Home
User`s Guide DeskTalk
Contents
1. 12 000 00 10 500 00 9 000 00 7 500 00 Zd 4 Fx ea PSI Ed m Ex BE Es E E E MA E E ES E ki B E ES KI FER Sd Bos ES Ez3 RI ES E Ki Ee E E e Es 6 000 00 Eo ka Bed Kal kal Del ki ka ka RID M Kal Es Bal ki Rs 000 00 E i Ee Pest 184 E Po Ed RST Pel Sd BS KO ES RSE E BR BSF ER RS koc SA ksal kad ksa boot Sed pool bo So poe basi kog RA BI 90d boe bed be IST BSA pos psa ER ES RA GH Ra KI el bep ES RSI bast Ted psa BST Deed PSA BAH on Eod kos 3 peat P9 RSH Ted BS BS PASTA Pe Ex Ea De ORS KA FS Nag BSS Kod BEY bes Med Des Pal Med post bee Rod KS D poe Kod pes bo 4 500 00 R39 KA ES hes RA Ed BA KA Led bed E Red P294 1903 BSS BOS Leod hes hasi Sd bes post led bes Eod pisi 1904 bed Doo z bey poe Dey ESS ksa BR BSR besi E 553 PS 194 BEY Poot Kod PRY BSC kisa Prot PSe sd pos beb Kael 199 PA PS ksa Pes RET Rod Pe eed RSET KS KA ES KSI Pe paga Red Pe oed Red Pe Deed hes BRS sd ksa Pew Pool oed Bes Deo Poo BSI psa Beet bosd 50 Boat bes psa peod ed Bet ood pool basi Sed poe kos Iso Boel best Iso Beet BS basi Sed Po BRS PSA BAST Se psa pasa e peat BA d peat kasi od SY SY 1829 Eod Mel ood Dod Pes od Dodi Pos Dod RSI pol RSE Ted D
2. mcollect trend label mw collect TRENDit trend export trendcopy trendtimer a E77 trend discover db delete data trend proc rmon collect KN db size util rcollect TRENDstep DataPipes Figure 1 1 TREND Applications and Processes 1 2 TREND Overview TREND 3 6 User s Guide TREND GUI Applications Background Processes Layer Information Report Schedule gt print N lpr_launch lt print N gt ps gif bmp trend_export lt cmd file gt data file Report Launcher TRENDweb age_files Presentation TREND Build gt qss qgr gos TRENDsheet lt gt qss files TRENDgraph qgr files Grade of Service lt gt gos files Processing TRENDit TREND sum lt sum file TRENDstep lt rot TRENDrank lt rnk Collection Collect Data DataPipes ee_collect ecollect mw collect mcollect rmon collect rcollect Setup Configure Users Groups Discover Import Nodes Autopilot Define Nodes Define Views Types MIBwalker TRENDdbview lt node list file lt package file mib file gt vwb files trend_discover trend Ibel TREND Overview L Si d O 2 o o 3 o 1 Introduction to TREND TREND N GUI Applications Background Processes Layer Database Data Manager trendcopy Mgmt db_delete_data db_size_util Utilities trendtimer lt trendtimer sched trend_
3. 11 16 Fixed Columns 0 000 c eee eae 11 16 Numeric Column Precision eee eee eee 11 16 Column Width aoaie e ee S e E E S 11 17 Exporting Data a A pee Lae a eee es 11 18 Format of a TRENDsheet Export File 11 19 Command Line Options 0 cee eee 11 21 Contents TREND apee lle AE LI e ERU MW ee 12 1 Main Display and Menu System 2 0 000 12 1 Menu Command Summary 0 e eee eee 12 2 Opening an Existing Graphical Report 00 2 eee eee 12 4 Modifying an Existing Report 00 e eee eee eee 12 5 Creating a New Report from an Existing Report 12 5 Creating a New Graph 0 0 cee tee 12 5 Selecting Statistics to be Graphed aa 12 6 selecting Data na soo PUNA bach heed eae edb BE bhala 12 8 Selecting Statistics eee 12 9 Displaying the Graph ee 12 11 Selecting Different Data lille 12 11 Displaying Data for a Different Node 12 11 Displaying Data from a Different Database 12 12 Setting the Time Period to be Displayed 12 13 Setting Report Display Parameters a 12 14 Changing Line Attributes 0 0002 eee 12 14 Changing Point Attributes 00 000 ee eee 12 15 Defining Fill Attributes ee eee 12 15 Adding Changing and Deleting Plot Lines 12
4. TREND 3 6 User s Guide BY M o z Q E trendtimer sched 9 5 DPIPE HOME bin mw collect n i 5 c 25 n E 10 DPIPE HOME bin mw collect n i 10 c 25 n EI 15 DPIPE HOME bin mw collect n i 15 c 25 n e 20 DPIPE HOME bin mw collect n i 20 c 25 n 60 DPIPE HOME bin mw collect n i 60 c 25 n 5 DPIPE HOME bin rmon collect n i 5 c 25 n 10 DPIPE HOME bin rmon collect n i 10 c 25 n 15 DPIPE HOME bin rmon collect n i 15 c 25 n 20 DPIPE HOME bin rmon collect n i 20 c 25 n 60 DPIPE HOME bin rmon collect n i 60 c 25 n 24 0041 00 DPIPE_HOME bin logbackup 24 00 42 00 DPIPE_HOME bin trendit s z 30 f 60 You can run parallel instances of TRENDit in a number of ways depending on your requirements For example you can add multiple TRENDit lines to the trendtimer sched file using the t option 4 trendtimer sched 5 DPIPE HOME bin mw collect n i 5 c 25 n 10 DPIPE HOME bin mw collect n i 10 c 25 n 15 DPIPE HOME bin mw collect n i 15 c 25 n 20 DPIPE HOME bin mw collect n i 20 c 25 n 60 DPIPE HOME bin mw collect n i 60 c 25 n 5 DPIPE HOME bin rmon collect n i 5 c 25 n 10 DPIPE HOME bin rmon collect n i 10 c 25 n 15 DPIPE HOME bin rmon collect n i 15 c 25 n 20 DPIPE HOME bin rmon collect n i 20 c 25 n 60
5. TrendCopy sh S source server S target server t table name This command copies only the delta table specified by table name from the source server to the target server and generates some logging informa tion Include a TrendCopy sh command line for each table name you want to copy TrendCopyLoop sh S source server S target server This command invokes concurrent instances of TrendCopy sh to copy all delta tables from the source server to the target server ie D o lt i o m 3 Z E 3 iv ko 3 apjajes e dn bumas a These scripts also generate some logging information Note The scripts are available for UNIX environments only For NT environ ments you must use a form of the trendcopy command described above Timing TRENDit and TRENDcopy Operations Consider the following items when you schedule TRENDit and TRENDcopy operations 1 First to generate delta data for a given day TRENDit requires at least two dif ferent samples whose timestamp is after 1 00 a m the next day Therefore if the satellite server is getting polled data every 15 minutes you should schedule TRENDit to run after polling completes say for 1 45 a m Polling operations do not always take the same amount of time so be sure to be sure to figure in a Timing TRENDit and TRENDcopy Operations D 5 D Setting Up a Satellite Server Environment buffer that gives polling time to complete before you schedule TREN
6. hostOut Errors hostOutPkts 100 0 Creating a New Tabular Report 11 7 11 TRENDsheet It is important that left and right parentheses are paired When all expressions have been created click Cancel to exit the Edit Expressions window Once you have finished defining the basic report in TRENDbuild you must save the query file you have created Use either the Save Save As and Exit options on the TRENDbuild File Menu to invoke the File Selection window You will be prompted to enter a file name under which the file will be stored The file will then be auto matically loaded into TRENDsheet After saving the report select Exit to quit TRENDbuild and return to TRENDsheet The new tabular report will be displayed If there are more columns than can be displayed at one time in the TRENDsheet win dow use the scroll bar at the bottom of the window Fixed columns see Formatting Columns in the Report on page 11 15 are always displayed in the window while the remaining columns will change as you scroll horizontally The default table length for a TRENDsheet report is 50 rows The maximum length is 2 000 rows The report title is displayed centered at the top of the table grid area It has three components file name the report file name without the qss extension node name the name of the node selected in the Select Node window key the value of the key chosen for the selected in the Select Node window Figure 11 4
7. 5 Select a Property Key table a To select an existing property table In the Existing Property Key Tables field select the property table you wish to be associated with your MIBlet Upon selection the index description appears in the Index description is field of the MIBlet s data table and the attributed node type is displayed in the Node type description is field b To create a new property table for the MIBlet or ius 2 x D Click on Create Property Key Table button The Create Property Table window appears Create Property Table x Description For a new property table you must specify its name and descriptions The name is the alias and not SQL name Enter Property Key Table Name Kib Il Foo Group Enter index description Enter node type description Cancel Back Next gt Figure 6 18 Create Property Table Window This dialog box Figure 6 18 defines the property table that identifies your MIBlet Creating a MIBlet 6 25 6 MIBwalker You may enter an alias name such as Router_Port_Properties in the Enter Property Key Table Name field This name will appear in the Existing Property Key Tables field on the Create MIBlet Assign Property Table window Enter a description of the instance identifier such as port into the Enter index description field Enter a node type description such as router into the Enter node type des
8. Click the Reset button to return to the default start and stop times Click the Cancel button to quit out of the window without changing the start and stop times Setting Report Display Parameters To complete the definition of the new report you must set its display parameters including defining the plot line types setting the X and Y Axis parameters and specifying report titles These modifications are handled through the Select Data Set window Figure 12 4 Changing Line Attributes 12 14 Line type color and width can be specified for each line plotted on a line graph The Type box is a list of available line patterns Select the pattern for the current line by clicking on the Type box s down arrow button then clicking on the desired type Use the scroll bar to view the list of available patterns The Color box is a list of the available line colors Select the color for the current line by clicking on the Color box s down arrow button then clicking on the desired color Use the scroll bar to view the list of available colors Setting Report Display Parameters TREND 3 6 User s Guide The Width box is a list of the available line widths in points Select the width for the current line by clicking on the Width box s down arrow button then clicking on the desired width Use the scroll bar to view the list of available line widths Note Keep in mind how the graph will look on paper while you set it up
9. DPIPE HOME bin rmon collect n i 60 c 25 n 24 00 1 00 DPIPE HOME bin logbackup 24 00 2 00 DPIPE HOME bin trendit t mib ii ifentry s z 30 f 60 24 00 2 00 DPIPE HOME bin trendit t mib ii system s z 30 f 60 24 00 2 00 DPIPE HOME bin trendit t ciscol00 lsystem perf s z 30 f 60 In considering a parallel implementation you need to assess the machine s CPU and RAM to ensure that running parallel processes do not have a detrimental performance impact TREND Processing Options C 21 C Performance Tuning Issues db delete data You may want to run several concurrent instances of the db delete data program to age records from multiple database tables thus enabling processor overlap If you do not specify a particular table with the t option db delete data processes tables having the appropriate aging option serially This may result in excessive run times By employing the t option which will only perform update statistics and record aging on the specific table you can run multiple instances of db delete data and thus improve throughput and reduce run times Intermediate Reporting Tables When CPU and memory constraints are not at issue the biggest factor that can negatively impact performance is table size The larger the table the longer it takes to process on any machine with any configuration Take care to retain only the data in the tables that is required for nightly TRENDit TRENDsum
10. Here are a few examples of launching trend_export from the command line trend export r trendsnmp reports Daily Lan Inventory qss o daily lan report txt f comma This would use the TRENDsheet report query file Daily Lan Interface Inventory qss as input and produce an output ASCII file with the name daily lan interface inventory txt Notice that the fully qualified path names were specified for both the report and the ASCII output file In addition note the use of the f option specifying that the ASCII file should be comma delimited trend export i trendsnmp reports ascii parm dat N j D m Z o o bad o s D Controls and Switches 20 5 20 TRENDexporter This invocation would use the input parameter file ascii_parm dat found in the di rectory named DPIPE HOMENlb to drive the ASCII file creation process See In put Parameter File on page 20 6 for the parameter file format trend_export i trendsnmp reports ascii_parm dat q hidekel This invocation would work like the previous one but would use the SQL server hidekel as the source of the data generated during processing Input Parameter File When using the i option to point the process to an input parameter file all other com mand line options are ignored except for q DSQUERY This is the preferred meth od for using the utility Only by using the input parameter file option can multiple reports and multiple instances of those report
11. lees 5 11 Running Discover from the Command Line 0 5 11 Using the Command Line to Run IP Discovery 5 11 Option Descriptions liliis 5 13 Notes About Some Discover Command Line Options 5 14 Specifying Community Strings for an SNMP GET Request 5 15 Community Strings Files 0 a 5 16 Running SNMP Type Discovery eee eee eee 5 17 Launching SNMP Type Discovery lille senes 5 18 SNMP Type Discovery Command Line Options 5 18 Option Descriptions leeren 5 19 Type Discovery Files 5 19 SNMP Tests 000 eee eee eee eee 5 21 Input File Examples 000 0000 eee eee eee 5 22 MIBwalket 22 2 1 DAL fae Sa aed oad Te ee a Pe RE GG 6 1 MIBwalker Process Flow 00 0c eee eee 6 2 Main Display and Menu System nassaan naana aana 6 2 Command Menu Option Summary saasaa aaan 00 e eee 6 6 Fils onines eR ut E O a ad E ASA alate Guat aT 6 6 VEW sedet deemed Are egeta dixe AGE a daten qs 6 7 TOONS east fere Me RETA Ri ewerdbenesies bte 6 7 Navigating the MIB Map 0 cee IIR 6 8 Using Object Path Radio Buttons 000 6 8 Using the Next Level Box 000 naana eee 6 9 Finding an Object in the MIB Map 0 2 0 eee eae 6 9 Loading a MIB aa eet AMANG Aa GAAN a aia AA 6 10 Preparing MIBs for MIBwalker 0 000 eee eee eee 6 11 Typographic Rules 2 000
12. 2 Use the ISQL utility to delete the invalid key table name from dsi_key_tables as follows isql Udsi dpipe Pdsi dpipe 15 delete from dsi key tables 25 where dsi key table invalid key tablename 3 gt go 3 Use the ISQL utility to delete the invalid key table 1 gt drop table invalid key tablename 25 go TRENDstep Column Definition You must use a column statement to define each column in the output table The column statement has the form column_name expression where expression can contain the following components Other column names Any rational number does not have to be an integer for example 5 12 5 Strings Strings consist of any sequence of characters enclosed in single quotes Examples this is a string is a blank string single quotes enclose a space character Mathematical operators See Mathematical Operators on page A 64 Logical operators See Logical Operators on page A 64 Using Column Names in Expressions gt Any string that is not a keyword constant number literal or comment is considered to be a column name There are four ways you can reference a column name 3 depending on the column type and how it is used in an expression D 7 Column Type How to Reference input data Names a column in the input data table Use the column SQL name input key Names a column in the input key table Use the KEY column_name syntax A 63 TR
13. 7 C i vesioz o7 7aezeszo4 4 a s mes 1 E emmmw s 5 9 eee 1 7oon5m emwem 6 6 am CA 2 RA mers E 2663469 40 2493934 67 atlanta 2663654 08 2494262 85 ce a Ka eae 2 BZ O E Ba 0 24 Figure 9 5 Sample LAN Connectivity Top Ten Report Ranking 9 25 v 3 2 D o D o 5 uonebaibby eleg 6 9 Data Aggregation and Manipulation Summary As introduced in this chapter the TRENDit TRENDstep TRENDsum and TRENDrank utilities produce and process cleaned data They provide the means for removing invalid or incomplete samples leveling performance spikes creating meaningful aggregation levels and computing statistics that provide useful measurements of enterprise health Man Pages on page A 1 describes the syntax and usage of these utilities in detail 9 26 Summary TREND 10 TRENDbuild Las i E m 2 g ley Q TRENDbuild is a report building utility that defines graphical and tabular TREND reports It allows you to specify data to be included in a report assign aliases to fields being reported on create expressions that derive calculated values from stored data for inclusion in a report and save the report definition as a query file TRENDbuild is used to create new reports for TRENDgraph TRENDsheet and Grade of Service applications Main Display and Menu System The
14. Figure 13 12 Grid Window The window shows the current grid display settings To turn grid lines on or off click the down arrow of the Show Grid box and select Yes to turn them on or No to turn them off To change the color of the grid lines click the down arrow of the Grid Color box and select the color you want Of course the color will only matter if you are display ing grid lines Click the OK button to activate the revised selections Click Cancel to retain the original values Drilling Down for More Detailed Data Grade of Service reports contain a drill down feature that allows the user to click a region of the graph to obtain a second detailed tabular report displaying the data of interest 13 20 Drilling Down for More Detailed Data TREND 3 6 User s Guide Command Line Options A Q mg o Q D h i D S 7 m The Grade of Service application is invocable from a command line the command agx gos h to see a description of the GOS command line options agx gos b database c printer copies gt e lt db end time in Sybase format gt f graph file h i print files k lt keyid1 keyid2 keyid3 5 p lt printer gt s lt db start time in Sybase format gt The Sybase date and time format is yyyy mm dd hh mm ss For example 1997 06 10 00 00 00 If you do not specify an extension for the lt print file gt file name the output is s
15. startserver f RUN_ lt HOSTNAME gt _SYBASE gt amp dev null amp where lt HOSTNAME gt _SYBASE is the name of your Sybase SQL Server or su root sh etc rc TREND DB up For NT make sure that the service for Sybase is started See the TREND Installation Guide for details E vv o 2 9 0 g TO 9 D O o m 3 Starting Sybase from the Command Line B 9 B Database Administration Recovering from Sybase Error 1105 Sybase error 1105 occurs when Sybase is unable to allocate space for a database The error message returned is 00 94 07 13 04 07 08 18 server Error 1105 Severity 17 State 3 00 94 07 13 04 07 08 47 server Can t allocate space for object syslogs in database dpipe db because th logsegment segment is full If you ran out of space in Syslogs dump the transaction log Otherwise use ALTER DATABASE or sp extendsegment to increase the size of th segment This message may be repeated If the State value is 1 your database is out of space and needs to be enlarged or cleaned out The error message may be accompanied by messages similar to the following 00 94 07 13 04 07 08 75 server Automatic checkpointing is disabled in database dpipe db because the log is out of Space It will continue when the DBO successfully checkpoints the database Please fr up some space or extend the database and then run CHECKPOINT Perform the following steps to
16. Avoid counting this data twice by subtracting this count from the count of all other IP counts else if KEY nt layer ip 1 amp amp KEY ap layer snmp If a row in a source data table con tains statistics about SNMP traffic over IP then snmp octets diststatsoctets010 Include the statistics in the resulting SNMP octet count column of the destination table other ip octets 0 0 diststatsoctets010 Avoid counting this data twice by subtracting this count from the count of all other IP counts A 71 gt D 5 v D e D 7 TRENDstep A 72 Line Item Comment endif This statement ends the IF block switch KEY dsi_table_key case ip 1 other ip octets diststatsoctets010 endswitch If a row in the source data table con tains statistics about IP only include the octet count in the network layer ip octet count of the output table This column represents all of the counts of IP traffic other than WWW and SNMP which is the rea son these counts were subtracted from the other_ip_octets field in the preceding IF blocks FINAL total ip octets ROTATE other_ip_octets ROTATE www_octets ROTATE snmp octets ENDFINAL This specifies the creation of an additional column in the output table which contains the total count of all IP traffic i e WWW counts SNMP counts other counts by_variable
17. Cancel Figure 12 9 Y Axis Window The window displays the current minimum and maximum Y Axis values from among all plot lines Note The minimum is not necessarily 0 To change auto scaling to the maximum value Click the down arrow of the Autose lect maximum value box and select Yes to enable auto scaling or No to use an ab solute value If you select No enter the desired maximum Y Axis value on the Maximum Value line or leave the current value shown To change auto scaling to the minimum value Click the down arrow of the Autose lect minimum value box and select Yes to enable auto scaling or No to use an ab solute value If you select No enter the desired minimum Y Axis value on the Maximum Value line or you can use the current value shown Setting Report Display Parameters TREND 3 6 User s Guide Enter a label for the Y Axis on the Label line Any printable characters are allowed The label will be displayed at the top of the Y Axis Click the OK button to activate the new scale Click on Cancel to retain the original scale If the range of values on the Y Axis differ by several orders of magnitude you may want to use a log scale when displaying a graph Highlight the Use Log Scale box to enact this functionality Specifying Report Titles Report titles are centered at the top of the graph area Of the three title lines two can be set by the user while the third is the date range for data pres
18. Change the value by typing a new one in the cell and hitting the return key Valid values are yes The data in this table will be rolled up no The data in this table will not be rolled up default Whether the data in this table will be aggregated or not is determined by the value set for Roll Up Data Automatically in the Default Storage Time window see Changing System wide Rollup and Aging Defaults on page 2 14 Only the user trendadm can modify a table s rollup option For all others clicking on the cell will simply highlight the entire row Changing Rollup on a Specific Table 2 15 2 TREND Configuration for trendadm Changing Aging on a Specific Table Change the value by typing a new one in the cell and hitting the return key Valid values O forever Never discard the data in this table number Actual number of days to keep a row of data in this table A row will be deleted when it is number 15 days old default The length of time data in this table will be stored is determined by the value set for Keep type Data in the Default Storage Time window where type matches the Type column value for this table see Changing System wide Rollup and Aging Defaults on page 2 14 Only the user trendadm can modify a table s aging option For all others clicking on the cell will simply highlight the entire row Truncating Tables 2 16 Truncating a table is when you delete of all rows
19. The source data table contains one row for each unique dsi key id for each collection period Thus if collections are made in 15 minute intervals there are 96 rows for each unique dsi key id four rows per hour 24 hours per day 96 TOWS If the o option is used the hour 0 23 specified in the option is compared to the ta period timestamp in rows of the source data table as follows 1 Get the timestamp for the current time minus 24 hours Thus if the current time is 11 22 35 on June 12 1998 the resulting timestamp is 11 22 35 on June 11 2 Make the hour portion of the timestamp equal to the value 0 23 specified in the o option Thus if o 4 is specified on the TRENDlabel command line the timestamp is 04 22 35 on June 11 mh Ce J m z o ley 82 3 Setthe minutes and seconds portions of the timestamp to zeroes In this example 04 22 35 on June 11 becomes 04 00 00 on June 11 Thus the target record in the source data table in this example is the one for the dsi key id where ta period is Jun 11 1998 04 00 00 4 If there is no row with the target time period the hour is decremented by 1 until a match is found For example if there is no row for the dsi key id where the ta period time is 04 00 00 TRENDlabel looks for a time of 03 00 00 decrementing the hour by 1 in this manner until a match is found Usage Notes 19 5 19 TRENDlabel If the o option is omitted the target row in the source tab
20. a JI m 2 o t o lt 18 TRENDcopy TRENDcopy is executed via a command line interface The following options can be used b Copies data for the length of the time period specified with the l option up to the baseline time specified with the b option length can be any integer and unit can have one of the following values h hour d day m month y year See Using the b and 1 Options To Copy Data on page 18 13 C Provides filtering by data table columns See Using the c and C Options To Filter Data by Column on page 18 10 C Provides filtering by key table columns See Using the c and C Options To Filter Data by Column on page 18 10 d Sets the debug output level Valid values are 1 2 and 3 The higher the number the more detailed the information Debug output is written to the standard output destination This option is for development purposes e Provides filtering by date A date is specified in the format yyyymmdd See Using the e f and g Options To Filter Data by Date on page 18 11 f Provides filtering by day of week Valid weekday values are su Sunday mo Monday tu Tuesday we Wednesday th Thursday fr Friday sa Saturday See Using the e f and g Options To Filter Data by Date on page 18 11 18 4 TRENDcopy Syntax TREND 3 6 User s Guide g Provides filtering by hour of the day Valid hour values are 1 24 where 24 means 12 midni
21. bles The control tables identify individual data tables for each group Data tables are logically divided into smaller tables according to control table rows Data tables identify the data sources The RMON MIB uses a control table with the history host and matrix tables to store the parameters that define what each instance of those tables represents For exam ple the control table stores the name of the user who created the instance the inter pon D fe 2 7 o Eei RMON Group Support 16 3 16 RMON Support face on the device that is the source of the data and how many buckets the instance contains To correctly collect the data in the history host and matrix tables the user actually defines polling on the control table however TREND collects both the control table and data table entries each time the control table is collected TREND polls control tables gets data and augments a key table with information associated with each data source This information includes the speed and type of the associated RMON probe network interface TREND polls for only one data table per source For example if history tables are setup as separate short and long interval ta bles TREND will only poll one instance of the data table based on a best fit concept If the data source is a Frontier Domain the associated domain description informa tion is also retrieved The Statistics group does not have a control
22. else if KEY nt layer ip 1 amp amp KEY ap layer snmp snmp octets diststatsoctets010 other ip octets 0 0 diststatsoctets010 endif switch KEY dsi table key case 96 ip 1 count network layer ip octets other ip octets diststatsoctets010 endswitch FINAL total_ip_octets ROTATE other_ip_octets ROTATE www_octets ROTATE snmp_octets ENDFINAL by_variable KEY dsi_target_name by_variable KEY dsi_data_source Figure A 7 Example of a TRENDstep Input Script File TRENDstep Figure A 8 shows the same Input Script File with comments on the significance of each line or block Line Item Comment source_table Rmon2_statsdata_ Identifies the source table destination_table rmon2_protdist Identifies the output table define IGNORE_NULL Specifies how TRENDstep will deal with null values See Handling Null Values in Expressions on page A 65 KEY if_speed KEY ifSpeed Include the column if_speed in the resulting key table and get its value from the ifSpeed column of the source key table if KEY nt layer ip 1 amp amp KEY ap_layer Www http If a row in a source data table con tains statistics about WWW traffic over IP then WWW octets diststatsoctets010 Include the statistics in the resulting WWW octet count column of the destination table other ip octets 0 0 diststatsoctets010
23. typename cisco_routers collection collection y n where y n is yes or no The default is no Specifies whether or not the mw collection table controls Type Discovery If the value is yes then trend_discover performs type discovery only if the mw collection table has an entry where username is the user running trend_discover the device_type is the name_of_type and the view_name is not set nodetype nodetype collect_type where collect_type is rmon or snmp The default is snmp Specifies whether this is an RMON or regular SNMP table The syntax rules for Type Discovery files follow 1 Every line in Type Discovery files must end with a semi colon except for comments Running Discover from the Command Line TREND 3 6 User s Guide 2 Comments begin with a slash and an asterisk and end with an asterisk and a Slash as shown in the following example this is a comment 3 Blanks are ignored in the input file unless they are inside a string between single quotes as shown in the following example this is a string SNMP Tests The actual definition of the type defines one or more SNMP Tests that a device must pass in order to be of the given type If multiple SNMP Tests are defined for a given type all of them must pass i e AND function is used The following tests are available Test Syntax Description Miblet miblet name This
24. 0 eee eee 2 9 The Specify Report Directory Window 2 11 The Description Editor Window llle 2 12 The Data Manager Main Window lle 2 13 The Default Storage Time Window 2 14 The Truncate Table Confirmation Message 2 17 The Delete Table Confirmation Window 2 18 The TREND Main Window 0 22 3 2 The TREND Discover Interface 5 7 MIBwalker Process Flow eres 6 2 MIBwalker Main Window naasna naana naaa aaa 6 3 Description Pop Up Window from MIBwalker 6 5 File Selection Window 0 6 6 View Selection Window 2 2 0 2 6 7 Tools Selection Window eee eee 6 7 Figures xxii 6 7 6 8 6 9 6 10 6 11 6 12 6 13 6 15 MIB Object Path Radio Buttons 000 6 8 Listing of Objects in Current Level aa 6 9 FING WINDOW oris te de EE PREISE TS 6 9 Select MIB Window lees 6 10 MIBlet Name Examples in Data Manager 6 16 Example of an error message uuauaan aan ua annaa 6 17 Viewing a MIB in Distinct Blocks aaa 6 18 MIBwalker with Variables of the ifEntry Table Selected 6 22 File Menu With Create MIBlet Selected 6 23 Create MIBlet Window 0 2 00 e eee eee 6 23 Create MIBlet Assign Property Table Window 6 24 Create Property Table Window 4 6
25. 123456 m your router2 sf JA r1234567 Z iv Pad o Output Format 20 15 z 20 TRENDexporter Tab Format Condensed The condensed mode of the tab separated format generates a tab delimited file but the columns are only as wide as the actual data rather than being padded to the de fined field width The columns are aligned with the next tab stop With headers y the output would be dsi_target_name dsi_table_key ifinoctets018 your router 1 123456 your router 2 123456 your router2 1 1234567 This same output with headers n would be your router 1 123456 your router 2 123456 your router2 1 1234567 Comma Format Condensed 20 16 The condensed comma format is a series of columns that are only as wide as the ac tual data they contain separated by commas Note that whenever a comma delimited file is requested all character based data is enclosed in quotes With headers the ASCII file generated would be dsi target name dsi table key ifinoctets018 your router 1 123456 your router 2 123456 your router2 1 1234567 If the headers no option was used then this would appear as your router 1 123456 your router 2 123456 your router2 1 1234567 Output Format N mrs E D m 2 is o 9 7 TREND 21 TRENDproc TRENDproc is a utility that groups together multiple interrelated commands This allows the TREND user to chain executable commands together allowing them t
26. For TRENDsheet the query file defines what columns will be displayed their order and name and expressions used to combine column values Similarly TRENDgraph files contain line and point col or pattern thickness as well as other attributes The TREND report query files are handled like any other files e g word process ing files or spreadsheets on your system For example a file could be stored in a standard default directory or subdirectory linked to a central library of query files A number of report types and formats are available A Brief Tour of TREND L TREND 3 6 User s Guide Graphical Reports By executing the TRENDgraph module from the main menu you can browse and select from a list of pre formatted reports called query files Query files contain the SQL commands which retrieve data from specific tables in the database Once a query file has been specified the user can select a variety of options to specify the node group or instance the date and day range X and Y axis scaling statistic se lection graphic display attributes and graph rendering i e line stacked bar or area graph Graphical reports can be printed directly to postscript printers scheduled for periodic execution or saved to files for customized post processing such as conver sion to Web page format Drill down reporting allows TRENDgraph users to point at an area in any TREND graph and click to produce a detailed tabular report Th
27. It has 22 chapters and 6 appendixes Chapter 1 Introduction to TREND Provides a general description of TREND functionality including the background information needed to understand how TREND works Preface xxvi Chapter 2 Chapter 3 Chapter 4 Chapter 5 Chapter 6 Chapter 7 Chapter 8 TREND Configuration for trendadm Shows the administrative user trendadm how to define users and groups and create polling views made up of different groups nodes and types for data collection and how to change defaults or rollup and aging parameters for data in the database trendadm can also truncate and delete tables in the database This chapter also discusses the issues related to maintaining the TREND license key The TREND Main Window Describes the main window for launching TREND applica tion modules and defining network configuration data Autopilot Describes TREND s automated approach to report genera tion which uses ReportPacks See the TREND ReportPack Guide for the current version of this chapter Discover Describes the Discovery module which automates the dis covery of network nodes MIBwalker Describes TREND s MIBwalker module which you use to load and view MIBs and define SNMP data collection Collect Data Describes the Collect Data module which you use to define polling policy for data collection Data Manager Describes how to use Data Manager s tabular display to administer an
28. Since we are only interested in the 1 40 p m time period Figure A 3 shows only that part of the report that includes 1 40 p m The line or bar for 1 40 p m is a representation of the IP traffic recorded by RMON agent on device 134 70 18 241 for that time period The stacked bar chart format is useful in that it provides a visual basis of comparison For precise numeric values the report shown in Figure A 2 with TRENDsheet is more suitable TRENDstep Input File A TRENDstep input file by convention has the file extension rot although you can use any extension You can use the following statements in this file Statement Description Name of the input table The table name you enter on this line may be the SQL real Table Name or the Object Alias Table Name Also TRENDstep cannot run against raw data You should only specify a data table against which TRENDit or TRENDsum or both has been run source_table table_name This statement appears as the first execut able line of the input file and is required gt D v o Q o o A 57 TRENDstep A 58 Statement Description destination_table table_name Name of the output table The table name you enter on this line must be the Object Alias Table Name for a raw data table TRENDstep automatically adds the appropriate prefix to reflect the table s rollup level rate hourly daily etc For example if you
29. TRENDdbview Opening an Existing View 22 19 Saving Changes to Existing View 22 20 promiscuous 16 2 A 34 Properties Key table 19 1 protocol 5 3 5 11 9 13 16 9 17 6 17 9 17 12 22 10 identifier string 17 16 invalid 17 13 parameter string 17 16 rename 17 14 Protocol Data Unit 5 13 5 19 Protocol Definition Interface 17 5 17 9 17 12 17 13 17 14 protocol definition interface RMON2 17 6 protocolDirTable 17 16 Q qualification clauses 22 15 text 22 15 qualifications 22 5 22 8 22 10 22 12 22 18 default 22 8 default setting 22 14 default settings 22 9 qualifier ecapVal 17 16 parmVal 17 16 query 13 4 file 1 11 13 9 query file 7 17 10 1 11 4 11 8 12 8 15 4 20 1 20 2 R ranges of scores 13 5 ranking statistics 9 23 ranking table 9 24 rate data 9 8 rate table 9 5 raw data 9 1 9 8 table 9 7 10 3 10 4 table 9 4 raw sample data 8 3 read community string 5 3 5 10 5 16 Read Community String option 2 6 read community 3 4 rebooted 10 3 reformat column parameters 11 14 Refresh Data Manager 8 5 Refresh option TRENDsheet 11 4 relational database A 33 relative weight 13 10 13 11 Remote Monitoring RMON 9 1 remote site 7 15 remove directory 14 10 rename protocol 17 14 reordering collection requests 7 11 replication 1 14 Report Builder 1 13 Launcher 1 5 1 12 report create a new 10 6 create with TRENDbuild 10 5 definition 12 5 14 7 directory changing the default 2 11 template 11 5 type 1
30. The key is shown in parenthesis If you have selected all nodes from the Node box in the Select Node window Figure 11 4 only the report name will appear 11 8 Creating a New Tabular Report TREND 3 6 User s Guide Modifying the Tabular Report To modify the definitions setup through TRENDbuild click the Edit Current View option under the Edit button This will put you into TRENDbuild in the context of the current report That is the TRENDbuild category group statistics aliases and expressions windows will contain the values used by the report being modified When you finish making changes in TRENDbuild you will need to reload the report into TRENDsheet using the Open option under the File button for those changes to be reflected in the report on your screen If you made any changes to the report definition before requesting Aliases Expres sions TRENDsheet will ask if you want to save those changes before proceeding Specifying Data to be Included in the Report Once a basic report has been defined and displayed specific nodes and keys can be selected the database can be changed and a time period can be specified Select the Node and Key Being Reported When you first run the new report TRENDsheet will display the data for all nodes and all keys for each node You can focus the report on a specific node and one or all of its keys for the table Click the Select Node option under the View button to bring up the S
31. The script follows and its name is BackupTrendDB csh Note Remember to replace the parameters identified by the angle brackets lt gt with the requested information Make sure that the bin isq line has all of its arguments on the same line bin csh f script to backup i e dump TREND and Sybase databases setenv DPIPE HOME TREND Directory setenv TAPE DEVICE lt Sybase Name of Tape Device gt Setup correct Trend environment source SDPIPE HOME lib Cshrc Log data to Sybase log file echo n Starting Sybase TREND backup on tee a SSYBASE install SDSQUERY log date tee a SYBASE install DSQUERY log Start isql session to perform the backups SSYBASE bin isql Usa P lt lt EOFEOF tee a SYBASE install DSQUERY log checkpoint Backing Up the SQL Server Installation TREND 3 6 User s Guide go dump database master to TAPE_DEVICE file master dmp with init go dump database sybsystemprocs to TAPE_DEVICE file sybsystem procs dmp go dump database dpipe db to S TAPE DEVICE file db dpipe dmp go quit EOFEOF echo n Completed Sybase TREND backup on tee a SYBASE install DSQUERY log date tee a SYBASE install DSQUERY log echo tee a SYBASE install DSQUERY log exit Remember to set the proper execute permissions such as 755 typically Place the script into the SDPIPE HOME scripts directory Make sure that the scheduler such a
32. Weekly and Monthly tables is comparatively insignificant TREND Processing Options C 23 C Performance Tuning Issues TREND C 24 TREND Setting Up a Satellite Server Environment w iv D lt D o m 3 Z E 3 iv ko 3 awpares e dp bumas a Satellite servers and central servers are installed in the same manner However there are numerous ways that they can be configured The most common and straightfor ward way is to 1 Set up polling to collect raw data at the satellite servers 2 Run TRENDit to convert the raw data to delta data 3 Run TRENDcopy to send the delta data to the central server A Sybase inter faces file defining the central server must exist on the satellite server the server that issues the TRENDcopy command See Creating the Sybase Inter faces File on UNIX Systems on page D 8 or Creating the Sybase Interfaces File on Windows Systems on page D 15 for details 4 Run TRENDstep TRENDsum and TRENDrank on the central server to create the base and summarization tables for reporting Reporting is done at the central server D 1 D Setting Up a Satellite Server Environment An Elan license manager should be installed at each server When you call DeskTalk Systems Customer Support to establish the keys for each license manager the sup port person will give you a license key based on your configuration The following sections point out considerations fo
33. by its RMON2 agent DataPipe Components 17 11 17 RMON Support 17 12 The protocols also appear in the Get Probe Protocols Data window Probe oinoo Retries B Community public Timeout fio Port Number hea ts 23 Valid Protocols IP Address wildcard ether2 ip 1 8 1 0 0 1 0 0 8 0 2 0 1 Unknown Address wildcard ether2 vip 8 1 0 0 1 0 0 11 173 2 0 0 DECNET Address wildcard ether2 drp 8 1 0 0 1 0 0 96 3 2 0 0 Unknown Address BB wildcard ether2 atalk 8 1 0 0 1 0 0 128 155 2 0 Unknown Address wildcard ether2 sna th 8 1 0 0 1 0 0 128 213 2 0 IPX Address wildcard ether2 ipx 8 1 0 0 1 0 0 129 55 2 0 0 Unknown Address wildcard ether2 ipv6 8 1 0 0 1 0 0 134 221 2 0 0 gt Add to Database View Invalid Change Invalid Name Quit Figure 17 6 Get Probe Protocols Data Window After a Get In Figure 17 6 we see that the Protocol Definition Interface was able to identify 23 valid protocols This means that the target device responded to the get request with 23 different OIDs each associated with the correct protocol name as defined by the RMONGQ protocol naming convention and that none of these 23 protocols already existed in dsi protocols The leftmost column describes each OID s address type If the address type is not know Unknown is displayed Since these 23 protocols are val id they should be inserted into dsi protocols To do this click Add to Database In Figure 1
34. host A are copied to host B 3 Edit the trendtimer sched file on host B to match the times and entries of the trendtimer sched file on host A especially the lines regarding trend proc f xxx pro for the packages and the lines regarding any custom reports scripts Trend license Permanent keys are tied to hostid and ip address Therefore a new license may be required Supply the new details to DeskTalk Systems so that we can generate a new key In the short term DeskTalk Systems can pro vide a temporary key F 3 n mz zo E O m D mp 3 gt ea o o gt m O F Relocating the TREND Database TREND F 4 Index Symbols DPIPE_HOME lib 2 22 DPIPE_HOME reports 2 11 3 7 14 7 SDPIPE HOME scripts 5 14 5 16 5 19 DPIPE_HOME tmp trend log 16 6 DSQUERY 18 7 18 8 SYBASE interfaces file 7 9 8 5 11 11 12 13 13 14 D 8 db used percent of database used 8 2 etc passwd file 2 4 A Abort button 6 33 About 1 19 absolute value 12 18 Access 6 4 Active Statistics box 10 10 10 11 actual range of values 12 17 TREND Add new view 2 8 type 2 10 Type button 2 10 views 2 8 add directory 14 10 new group 2 3 new type 2 10 node 2 7 operators 22 17 Add New Report option 15 4 Add Node button 2 7 Add to View button 10 14 age_files utility A 2 agent 16 6 aggregate raw data A 39 aggregated 8 3 aggregation 1 13 10 4 11 10 aggregation of data 9 1 incomplete 7 17 aggregation t
35. o TRENDstep Rules for the expressions used in a switch block include Expressions can contain constants column names and Expressions are evaluated from left to right no operator precedence is observed However expressions within parentheses are evaluated first The following logical operations are allowed lt lt gt gt The amp amp and ll boolean operators are allowed in a switch block but might make it difficult to understand the logic Use IF ELSE constructions instead if bool ean logic is needed If one of the case statements inside a switch block is true the rest are skipped A switch block s default case is executed only if every other case inside the switch block is false IF Blocks Figure A 6 shows the IF block construction if expression column expression else if expression column expression else if expression column expression else column expression endif Figure A 6 IF Block Construction A 68 TRENDstep String Comparison To perform wildcard character comparison use the percent 9o symbol to represent any sequence of one or more characters inside a string Only equal and not equal comparisons are allowed if wildcard characters are used TRENDstep Output Table Creation If the specified output table does not exist TRENDstep creates the new data table and its associated key table The new data table cont
36. report myreports lan interface load qss output myreports lan interface load max y instances mynodel 1 mynodel 2 mynode2 1 mynode2 2 instances mynode3 3 mynode4 1 end start report myreports lan_interface_load qss output myreports lan_interface_load max y instances mynodel 1 mynodel 2 mynode2 1 mynode2 2 instances mynode3 3 mynode4 1 style combine format comma condense y end Example 2 causes trend_export to run a set of reports 1 First it would run the same report as in Example 1 2 The second report run would be Serial interface Usage qss Its output would be a tab separated uncondensed data saved in a file called tmp Input Parameter File 20 11 N E D m Z o o bad o e s o 20 TRENDexporter 20 12 serial_interface_usage lt timestamp gt Notice that all the default values would be used for this second report The third report to be run lan_interface_load qss would produce six tab sepa rated uncondensed ASCII files because the keyword style individual is used The qss report source file is used as a template to run the query for each of the listed instances Note that the instances are on multiple lines They could have been all listed on the same line Also note that the max keyword was added to indicate that the ASCII files will only contain data for the most recent period available for each instance The output
37. that input to be 134 70 0 0 Running Discover from the Command Line TREND 3 6 User s Guide A CAUTION Running Discover from the Command Line When you use the D suppress domain name option translated host names will be truncated after including the first period in the host name Note When the node translation z option is on and a ping is successful Dis covery will perform a hostname lookup A previously discovered trans lated name can only be replaced by another translated name not an IP address In other words if the host name lookup fails to return translated name Discover will not replace the current name with an IP address If the host s name is not valid anymore the translated name has to be updated manually and node name entries in dsi_snmp_nodes dsi_nodes dl_type and mw_collection tables must also be manually updated If a node is defined via Define Nodes or any other application then MAC address in formation for this node is not available to Discovery This can cause duplicate polling Specifying Community Strings for an SNMP GET Request When Discover performs an SNMP GET request the community strings it uses can be defined in the following ways If you do not specify a community string Discover will use the default which is public If you want to specify a single community string for all the SNMP GETs in a discovery you can use the c lt community_string gt option W
38. the test is done on sysDescr oid and ipForwarding oid typename 3com_router collection no 1 3 6 1 2 1 1 12 3com 1 3 6 1 2 1 4 1 1 5 22 Running Discover from the Command Line TREND 3 6 User s Guide Example 3 a o a o o lt o 2 this file will define types that support mib II_ifEntry miblet and have ipForwarding turned on typename mibIl collection no mib II ifEntry 1 3 6 1 2 1 4 1 1 Example 4 this file will define rmonl ether stats type this type is limited to rmon1 devices that do not support rmon2 extioins typename rmon ethstats collection no nodetype rmon 1 3 6 1 2 1 16 1 4 1 1 doesn t support rmon2 extension defined for rmonl 1 3 6 1 2 1 16 1 1 1 1 must support etherStatsIndex oid Running Discover from the Command Line 5 23 iscover TREND TREND MiBwalker o Ww 2 2 The SNMP protocol supports management of network resources through a data structure called a Management Information Base MIB MIBs are composed of groups of managed objects which are tables of variables that can contain a node s statistical values and configuration parameters A node is used here to mean any network communications accessible device that can be polled using SNMP Each node supports the MIBs the node manufacturer feels are appropriate to the function of the node Definitions for standard MIBs appear in Request for Comment RFC do
39. utility A 11 lose information on the TREND main win dow 3 2 Ipr_launch 1 8 21 2 21 4 utility A 12 MAC addresses 16 2 MAC Layer 16 2 A 34 macro 21 1 Management Information Base MIB 9 1 manually initiated process 5 2 mask Subnet 5 9 Match Value window 14 13 Matrix 16 2 matrix group RMON 16 3 matrix table A 33 MAX 5 10 maximum length 11 8 maximum length of name MIB object 6 12 maximum value 9 8 maximum width 6 21 menu Configuration 2 5 2 7 2 9 3 5 Configure 2 6 File 3 4 Tools 3 7 View 3 5 menu bar Data Manager window 8 4 message GET 5 3 metrics 9 2 9 12 MIB 1 5 1 9 1 12 6 6 6 10 6 11 6 17 8 3 9 1 10 6 11 6 12 6 13 7 16 1 22 1 22 12 22 18 A 27 A 30 A 31 definition 6 1 file 6 1 6 10 truncated file name 6 16 map 6 1 6 3 6 8 6 9 6 31 16 3 object 6 7 6 20 6 21 6 31 7 5 10 3 10 7 maximum length of name 6 12 selecting in GOS report 13 7 TREND 3 6 User s Guide structures 1 9 syntax rules 6 11 table 6 2 6 19 6 33 7 7 7 11 A 22 tree 6 7 structure 1 5 6 8 values 6 8 variables 1 5 1 9 6 33 7 1 MIB II tables 5 16 MIBlet 6 21 6 29 7 5 9 10 10 3 creating MIBlets 6 7 MIBname 22 18 MIBname_View_viewname 22 18 MIBwalker 1 5 1 9 3 8 6 1 9 10 16 3 A 32 E 3 Community Strings option 6 8 6 30 Find option 6 7 Load Mib option 6 6 setting a variable 6 34 SNMP Tool option 6 7 6 31 Change Node option 6 29 Set Table 6 35 Set Variable 6 34 MIN 5 10 model 17 9 mo
40. 0 eee ee 12 19 Grid WING OW a za sepe na ad deed RE ge 12 20 Graphical Report with Supporting Drill Down Tabular Report 12 21 Grade of Service Report Window 0000 13 2 TRENDbuild Main Window lille 13 7 Edit Expressions Window else 13 8 Select Data Set Window 0 0 00 eee eee 13 9 Select Key Window 0000 cee ee 13 10 Hourly Interface Health Report Data Set 13 12 Change Database Window nananana 13 14 Set Time Period Window 0 2 22 13 15 xxiii Figures xxiv 13 9 13 10 13 11 13 12 14 1 14 2 14 3 14 4 14 5 14 6 14 7 14 8 14 9 14 10 14 11 14 12 14 13 15 1 15 2 15 3 17 1 17 2 17 3 17 4 17 5 17 6 17 7 17 8 17 9 22 1 22 2 22 3 22 4 22 5 X Axis Window 00 00 eee eee 13 17 YsAxiS WINdOW sss reheteS pobre eee GANA And 13 18 Graph Title Window leise 13 19 Grid WINdOW i5 odere sin Oe da ale bet anced bed ad NGA 13 20 TREND Main Window 0 222 eene 14 2 Report Launcher Window 00 eee eee eee 14 3 Report Launcher Window s File Menu 14 6 Change Database Window 00 eee ee eae 14 6 Report Launcher Window View Menu 14 9 Report Directories Window 00 00 e eee eee 14 10 Report Types List Box a a auauua ce eee eee 14 11 Search Lists Window a an eee eee 14 13 Devices Window 00000 0c eee eee 14 14 Match Value Window f
41. 14 3 14 Report Launcher Control Description Folders list box Lists the Top Level Report Directory and its subdirectories You can select any report directory to be the Top Level Report Directory We tell you how to do this in Selecting the Report File Directory on page 14 7 The Folders list box makes it easy for you to select a directory to be the current directory You can easily change the value of the current directory by double clicking on an entry in the Folders list box Reports list box Lists the reports defined in the current directory You select the report you want to create from this list box Report Types list box Enables you to restrict the contents of the Reports list box to a specific report type for example all types TRENDgraph TRENDsheet and so on Prev Day button Provides an easy way to produce a report for the pre vious reporting period The label on this button is Prev Hour Prev Day Prev Week or Prev Month depend ing on whether the selected report is an hourly daily weekly or monthly report respectively Set Time button Enables you to specify or change the report time period Select Devices button Enables you to select specific devices and elements to include in the report View Report button Launches the report Remaining sections in this chapter describe how to Select the database to be reported see Selecting the Database
42. 2 9 23 9 25 A 48 TRENDsheet 1 8 1 10 1 11 1 18 7 4 7 16 8 6 10 1 11 1 14 12 20 5 20 14 22 9 22 12 22 19 22 20 Change Database option 11 3 11 11 column format 11 15 11 16 Column Format option 11 4 11 16 11 17 create new report 11 5 Edit Current View option 11 4 11 9 Export option 11 3 11 18 Index 19 Index Index 20 query file 20 1 20 2 Refresh option 11 4 Select Node option 11 4 11 9 sort order 11 14 Sort Order option 11 3 Time Period option 11 3 11 12 TRENDstep 1 8 9 2 9 11 9 13 9 15 9 16 A 52 script file 9 14 TRENDsum 1 6 1 8 9 2 9 10 9 15 9 16 9 17 9 18 9 24 21 8 C 19 formulas 9 17 trendtimer A 12 A 26 F 1 utility A 90 trendtimer sched 19 1 21 2 21 4 21 5 file 1 8 21 10 trendtimer sched file A 11 D 3 F 3 TRENDview HTML 1 13 PRO 1 13 TRENDweb 1 13 troubleshooting 7 16 Truncate button 2 17 Truncate table s option 8 4 truncate tables 2 16 type 7 6 7 7 8 3 12 14 16 4 18 13 add 2 10 define 2 9 3 6 definition file 5 19 delete 2 10 discovery 5 17 E 2 files 5 19 edit 2 10 select 2 10 validation 5 5 type_name 3 4 U unauthorized usage 2 19 uncondensed 20 14 Unselect All 8 5 update statistics A 10 user 7 10 defining 2 2 user visible name 8 3 utility TRENDproc 21 1 V validity of collected data 9 8 value Days 8 3 95th percentile 9 18 default 8 3 forever 8 3 ta period 9 4 value range 12 4 version number of TREND 1 19 View Daily Data
43. 3 Unused at this time Database Size Verification i e db license Unused at this time Unused at this time Different TREND programs use the feature number to request a token from the li cense key Tokens are generally available to anyone who can reach the license serv er Clients reporting and polling are required to obtain a token from the correct location specified by user environment settings Once a license key is installed the system issues a token authorizing the client to use TREND Polling clients utilize either a local license key to obtain a token or must go to the server to obtain a token Reporting clients must go to the server to obtain a token If there is an error message in the trend log such as Server banyas 38 No licenses are currently available License Key Administration 2 21 uoneanByuo GNAYL Z 2 TREND Configuration for trendadm Contact DeskTalk Systems Technical Support to have a new License Key generated that covers the updated range of expected tokens This error is being generated by the db_license program which is expecting or using the 38 token for monitoring da tabase size and if still using a License Key generated for much earlier TREND re leases the updated range of tokens will not be available DSI ELMHOST 2 22 When an application starts it attempts to get a license from the license server on the network by sending out a broadcast The application will search
44. 4 11 5 11 6 11 7 11 8 11 9 11 10 12 1 12 2 12 3 12 4 12 5 12 6 12 7 12 8 12 9 12 10 12 11 12 12 13 1 13 2 13 3 13 4 13 5 13 6 13 7 13 8 Sample Frame Relay Executive Summary Report Baselines 9 23 Sample LAN Connectivity Top Ten Report Ranking 9 25 TRENDbuild Main Window lille 10 2 Select MIB Window 2 22 22 10 6 Select Group Window 2 2 2 10 7 Select Statistics Window 0 0 cee eee 10 9 Assign Statistic Aliases Window a 10 11 Edit Expressions Window cee eee ee eee 10 12 TRENDsheet Main Window llli 11 2 TRENDbuild Main Window lille 11 5 Edit Expressions Window 11 7 Select Node Window 00 eee eee eee 11 10 Change Database Window 000 005 11 11 Set Time Period Window 0000 02005 11 12 Constraints Window cece eee eee eee 11 13 Order Window 000 00 cee eee eee eee 11 14 Column Format Window 0 2 2 aaaeeeaa 11 17 Export Window 11 18 TRENDgraph Main Window eee eee 12 2 TRENDbuild Window 0 2 12 6 Assign Statistic Aliases Window aa 12 7 Select Data Set Window 2 12 9 Select Key Window 0 00 eee eee eee 12 10 Change Database Window 2 annaran 12 12 Set Time Period Window 2222 anaana naaa 12 13 X Axis Window 0 00 cee 12 17 Y Axis Window i a E a 00 ce eee 12 18 Graph Title Window 00
45. 5pts lt 6Pisc Stats Add Statistic Update Statistic Update All Statistics Delete Statistic Delete All Devi ia is Pts ua evice Key Descriptio Fill Patte Bar Colo E lt 35 lt 50 Percent Utilization pawnee avg 2 Solid Green 50 25 Percent Input Errors pawnee tot 2 Diag Cros Blue 10 Percent Output Errors pawnee tot 2 Vertical Cyan jo 21 f2 23 Received Pkts Discardi pawnee tot 2 Cross Red 1596 lt 1 lt 3 s5 Graph Cancel Figure 13 6 Hourly Interface Health Report Data Set Threshold labels are specified for the graph using the Y Axis command under the View menu see Setting Y Axis Parameters on page 13 18 Finally Save the fully defined report 13 12 Creating a New Grade of Service Report TREND 3 6 User s Guide Selecting Different Data o o o Q D e h i I 7 om Displaying Data for a Different Node When a GOS report is created statistics or expressions must be identified as well as which device key pair is to be the source of the data To change the data source node for a report select the Grade of Service Select Data option under the View menu This will bring up the Select Data Set window Figure 13 6 Follow these steps for each statistic 1 Click a row in the Statistics to graph box The window will highlight the def inition for that line 2 Set the desired value
46. 6 9 fixed column 11 16 floating licenses 2 19 forecasting 9 17 format comma 20 15 parameter file 20 6 tab 20 14 formulas computing statistics A 80 Frequency 7 11 frequency value 20 9 function roll up 9 10 summarization 9 10 G gauge 6 14 10 3 10 5 data 9 9 definition 1 10 generate report 14 20 GET 5 10 message 5 3 Get button 6 35 Probe Protocol Data window 17 9 GOS 1 5 1 11 1 12 9 13 9 15 GOS See grade of service grade device performance 9 2 Grade of Service 10 1 14 12 report 19 5 grade of service 1 5 1 10 1 18 adding changing graph titles 13 19 basic concepts 13 5 changing colors and fill patterns 13 16 changing databases 13 14 changing nodes 13 13 changing time scale 13 15 command line options 13 21 creating a new report 13 7 creating editing expressions 13 8 drill down 13 20 grid lines 13 19 Interface Health Report 13 11 principles 13 5 reporting 13 1 selecting keys 13 10 Grade of Service GOS 9 13 TREND 3 6 User s Guide graded 9 14 grading scale 1 5 13 1 grading ranges of input data 9 13 graph title 13 4 grid 13 4 13 19 13 20 grid lines 12 20 group 7 8 7 11 10 6 10 7 10 8 11 6 define 2 3 Group By Function 7 14 Information Available 7 14 Instance button 6 33 Location 7 14 Priority 7 14 Vendor 7 14 Group Name 6 25 group_name 3 4 field 7 8 groups definition 2 2 GUI 5 6 5 10 7 1 H History 16 2 history group RMON 16 3 history group polling RMON 16 5 history tab
47. 7 16 Hourly Data 7 16 Rate Data 7 16 Raw Data 7 16 Weekly Data 7 16 view 3 6 7 7 16 4 18 13 22 1 22 9 22 10 Define 3 6 define 2 7 new 2 8 View button 8 7 8 8 View menu 3 5 view name 3 4 viewing SNMP data 7 16 virtual data table 22 9 22 10 22 19 TREND 3 6 User s Guide virtual table 22 1 22 2 22 11 22 20 volatile element 9 24 volume of data collected 7 15 WwW wait 21 10 wait block 21 3 21 4 weighted 13 5 weighted stacked bar format 13 1 weighting 13 11 Whole Word 6 10 wildcard function 17 16 window Background Color 2 13 Define Nodes 2 5 Define Types 2 9 Define Users and Groups 2 3 2 5 Description Editor 2 12 Select File 1 16 Specify Report Directory 2 11 Write Community String option 2 6 write_community 3 4 X X axis 12 16 13 17 option 13 4 Y Y Axis 12 17 command 13 12 hash marks 12 20 option 13 4 13 18 parameters 12 18 Index 21 TREND
48. A 4 db delete data deletes data from a TREND table when that data has been stored longer than the retention period specified for the table in the TREND database You can find the data retention period for various tables from the TREND Data Manager display See Data Manager on page 8 1 for details db delete data is executed automatically according to the schedule specified in the trendtimer sched file You can also execute db delete data at any time from the command line Deleting obsolete data from the database and updating the database statistics page are separate activities You must invoke db delete data with the u and U options to update the statistics page db_delete_data Options db_delete_data has the following options Number of instances of db_delete_data to allow to run concurrently The default is 1 that is db_delete_data processes each table one at a time Sets a debug output level Values of 0 1 2 or 3 are valid The higher the number the more detailed the information The default is 0 which means no debug output Debug output is written to standard out Deletes data only in tables whose source is the value of this option Valid values are datalyze Legacy tables created populated by TRENDit the daily weekly and monthly rollup tables that result if the r option is omitted from the trendit command dsi ee Tables populated by ee collect MW Tables populated by mw collect rmon Tables
49. Discovery window appears the Subnet mask is set to eight bits by default as shown below If the Subnet you want to search requires a different Subnet mask you can change the mask using this text field Note The Subnet Mask you apply determines the limits of the IP Address range you can define for discovery Setting a Range for Discovery Discovery allows you to specify a range of IP addresses to discover You do this by specifying a minimum and maximum value using the fields shown below 134 70 65 0 134 70 65 255 Note The range you can discover depends upon your Subnet Mask You must ensure that the Subnet Mask field is set appropriately for your net work s Subnet Mask Using the Discover Graphical User Interface 5 9 5 Discover Ifthe discovery you want to perform is in your local Subnet simply set the Min and Max Ranges and click Start The Discover interface appears showing the IP Range of your machine s local Subnet If you are running Discovery from a machine that is not part of the tar get Subnet you will need to reset the MIN and MAX Range values to discover the target Subnet About Community Strings When Discover finds a device at an IP address it sends an SNMP GET to determine whether or not that device is SNMP manageable This SNMP GET includes a Read Community String The targeted device will only respond to Discover if its commu nity string matches the string in the GET For
50. Find the lines for maximum and minimum online engines in the Processors group in the configuration output listing Group Processors Parameter Name Default Memory Used Config Value Run Value max online engines 1 132 1 1 min online engines 1 0 1 1 In multiprocessor environments you need to configure Sybase to use the additional CPUs For systems with less than four processors set the max online engines to the total number of processors In systems with four or more processors you can gain additional throughput by allowing one or more processors to be available for client processes for example TRENDit mw_collect rmon_collect TRENDcopy For example on an eight processor system allocate six processors to Sybase 1 To change the available processors isql Usa P sp configure max online engines 6 go 2 Restart Sybase to enable the new processor allocation Transaction Log Cache In most cases you should create a separate named cache for the transaction log because the transaction log is primarily write intensive Excessive writing to a shared cache such as the default data cache causes unnecessary flushing of pages to disk and thrashing or rereading of pages thus decreasing overall system performance C 12 Sybase Configuration Parameters TREND 3 6 User s Guide To set up a named cache for the transaction log isql Usa P sp cacheconfig logcache 5M logonly go This example creates a 5MB named cac
51. Low Pairs Figure 6 21 Counter Pairing Selection Click on Add Counter Pair The pairing appears in the High Low Pairs and is removed from the selection blocks See Figure 6 22 High Low Pairs sitCtrSplitl High sitCtrSplitl Low Figure 6 22 View of Applied Pairing 10 Click Finish 11 MIBwalker displays the following message and creates the raw and rate MIBlet tables with all other required SQL objects to support raw to delta 6 28 processing i Raw miblet b_ii_foo_ii_group_ and rate miblet R_ii_foo_ii_group_ created Figure 6 23 Window Verifying Creation of Tables Creating a MIBlet TREND 3 6 User s Guide Once the MIBlet tables have been created you can use TREND s Collect Data tool to begin collecting data for it See Collect Data on page 7 1 for a detailed description of Collect Data Utilities Changing the Current Node Utilities MIBwalker uses the current node named in the header bar as the target in the SNMP Tool You can change the current node to a different device Select the Change Node option in the Tools menu to bring up the Connect To window Connect to 134 70 65 10 Figure 6 24 Connect To Dialog Box The window displays the name of the current node on the Other Host line The list of nodes already known to TREND is in the Current Hosts box To change the current node type in a new node name or the node s IP address on the Other Host line or selec
52. MB 1 GB Medium to large installations 4 000 10 000 interfaces 1 GB nGB Large installations 10 000 interfaces Hardware Considerations C 3 C Performance Tuning Issues However this table is just a guideline You cannot give a system too much RAM When in doubt get more RAM than you think you need Disk Drive Interfaces If a system is configured with enough RAM the next likely bottleneck is disk I O Choose an appropriate disk interface controller if the platform supplier provides one as an option to ensure that data can move quickly from the disk to RAM for processing In general SCSI interfaces outperform IDE or Enhanced IDE interfaces There are at least three flavors of SCSI SCSI IL 10MB s burst transfer rate Ultra SCSI 20MB s burst transfer rate Fast Wide Ultra SCSI 40MB s burst transfer rate As with RAM the faster the interface the faster data can be moved on the I O bus to RAM for processing Do not confuse transfer rate with access time which deals with disk actuator positioning on the drive itself Faster interfaces also provide greater throughput for systems that have multiple I O devices connected to the interface controller Some platform vendors have optional hardware caching controllers that can provide significant performance improvements over noncaching controllers For larger systems you can also use multiple interfaces Fast Wide Ultra SCSI is the interface of cho
53. MIBs for MIBwalker TREND 3 6 User s Guide The following value is not allowed and should be replaced with a legal one experimental Convert SNMP v2 Statements Comment out the MODULE IDENTITY section Comment out the MODULE COMPLIANCE section Remove the label TEXTUAL CONVENTION from statements where it occurs Replace MAX ACCESS with ACCESS Replace BITS with OCTET STRING Replace TimeStamp with TimeTicks RowStatus is actually a TEXTUAL CONVENTION Replace it with SYNTAX INTEGER active 1 notInService 2 notReady 3 createAndGo 4 createAndWait 5 destroy 6 j Other TEXTUAL CONVENTIONSs are defined in RFC1903 9 9 9 Replace multi part SIZE arguments with a range that includes the smallest and largest arguments For example SIZE 018120 is replaced by SIZE 0 20 Preparing MIBs for MIBwalker 6 13 o Ww 2 2 6 MIBwalker Validating Object Types 6 14 It has been observed that MIBs sometimes use inappropriate object types The most common examples of this are when fields that function as a Counter or a Gauge are defined as INTEGER Note that the field type labels are case sensitive These must be corrected if the data collected from those fields is to be aggregated correctly The DESCRIPTION field can give you a clue about the way the data in a field is used Fields that should be defined as Counters monitor how many of something occur over time T
54. MIBwalker TRENDgraph and Grade of Service applications The user must have a color monitor to use this function see Setting Application Background Color on page 2 13 Tools Tools The Tools menu contains two options for building and defining variables and reports Display and Menu System 3 7 o E o zT E 0T 20 D 3 The TREND Main Window Select MIBwalker to define new groups of MIB variables to poll for or to browse through the MIB tree structure using a point and click interface refer to MIBwalker on page 6 1 for more detailed information Select TRENDbuild to select variables assign custom aliases and build expressions that are written to a query file used by TRENDgraph or TRENDsheet refer to TRENDbuild on page 10 1 for more detailed infor mation TRENDbuild can also be directly accessed through TRENDgraph or TRENDsheet You can also invoke the following reporting facilities from the Tools menu TRENDgraph see TRENDgraph on page 12 1 TRENDsheet see TRENDsheet on page 11 1 Grade of Service see Grade of Service on page 13 1 3 8 Display and Menu System TREND 4 Autopilot Autopilot is used for installing and removing TREND ReportPacks Please see the ReportPack Guide Chapter I for more information A gt Lad e o e 4 1 4 Autopilot TREND c o a p e lt D mg TREND 5 Discover Thi
55. Median mid Delta time when the minimum value occurred min Minimum value mit Time when the minimum value occurred per90 90th percentile per95 95th percentile per98 98th percentile A 77 gt D v o Q o o TRENDsum A 78 e rct min max Count of samples greater than or equal to the min value and less than the max value i e count of samples gt min and max Note The definition of this statistic has changed See the g option to use the version 3 5 1 and earlier definition std Standard deviation the square root of the variance tct value Count of samples greater than or equal to value i e count of samples gt value Note The definition of this statistic has changed See the g option to use the version 3 5 1 and earlier definition tot Total vet value Count of samples equaling value wav Weighted average Name of the destination table for the TRENDsum output This option is required for a non baseline table For a baseline table the full name of the destination table will be SDnn where nn is the number of days used for the value of the y option plus the source table name t plus this name You may not use both the a and e options Name of an input file containing the aggregation instructions TRENDsum 8 h n r t V y Use version 3 5 1 compatibility for the ret and tct statistics in the c option rct min max Co
56. Miedo aca RC havc aye wha Ber bk 2 5 Defining VIeWS caletiq uice ie de id RII een td NLA REGNI UE 2 7 Defining Types rennene E RR RII 2 9 Changing the Default Report Directory 000 eee 2 11 Defining Keys vio doo etd PAR NENA LGA d xax beh dtu 2 12 Setting Application Background Color aaa 2 13 Changing Data Rollup Parameters llle 2 13 Changing System wide Rollup and Aging Defaults 2 14 Changing Rollup on a Specific Table 0 0a 2 15 Changing Aging on a Specific Table 0 0a 2 16 Truncating Tables cee 2 16 Deleting Tables cee cee eee 2 17 License Key Administration llis 2 19 TREND License Scheme 2 00 e eee eee ees 2 19 License Composition llle ee 2 21 DSILEEMHOST kot asma t maala ee xd ds edet ads 2 22 Message Handling 0 cece ete tte ees 2 24 Example amag ou nad ae GG PEARCE RERNIECREP ae eae a 2 24 General Message Format cece eee eeaes 2 25 Process Initiation Messages 00 eee eee eee 2 26 TREND 3 6 User s Guide Process Completion Message eee ee eee 2 27 Process Error Message cece eee eee eee 2 28 The TREND Main Window eeeeen a 3 1 Display and Menu System 00 0 anaana 3 2 Menu Command Summary 0c eaa eee 3 3 THe File Men esos Gitte AA nak Gre ph haa pte ad NAGA 3 3 IMPOR NOTES na nee pa nad
57. Next select whether the keys i e table columns should be sorted in ascend ing or descending order by clicking on the appropriate button in the Order box Se lect the first sort field from the Key box by clicking on the field name in the box The key and order you choose will appear on the Order by line You can choose multiple sort keys with a different order for each key The sort will occur in the order you select the fields Click the OK button to have the sort order you specified take effect Click the Reset button to return to the default sort order Click the Cancel button to quit out of the window without changing the current sort order Formatting Columns in the Report You can alter the order of the columns in the report as well as the precision of nu meric columns and the width of each column to make the report more readable Column Order You can reorder the columns in the report by dragging and dropping them into dif ferent positions ok J m z 7 7 i Grab a column by clicking on its column heading the cell that contains the column name and holding down the select key on the mouse The box around the cell will be displayed as a broken line Continue to hold the mouse button down while moving the mouse The broken lined box will move with the cursor on your screen Move the cursor on top of another column heading and release the mouse button The col umn you are moving will be place
58. Select Group 10 7 13 7 Set Table 6 35 Start 5 7 Stop 5 8 Truncate 2 17 View 8 7 8 8 by variable A 81 TREND 3 6 User s Guide C Cancel button 5 8 6 36 7 9 case sensitive 6 14 categories of service 13 5 category 7 11 8 7 10 6 10 7 11 6 12 6 13 7 Other 9 11 9 12 Show by Source 8 7 Show by Type 8 7 central library 1 10 Change Database command 8 5 default report directory 2 11 change data source node 12 11 change database 7 3 option 7 9 window 7 9 Change Database option 12 12 TRENDsheet 11 3 11 11 Change to box 7 9 Changing report directory See Specify Report Directory option 2 11 child collector 17 6 cleaned data 9 16 cleaning process 9 7 raw data 9 1 9 4 Collect Data 1 4 1 9 6 29 7 12 16 9 button 7 2 Periodically option 7 3 7 5 Periodically window 7 10 7 13 A 27 A 34 window 7 8 7 9 7 10 7 16 collect ed data 7 1 7 4 9 4 performance data 9 1 9 3 collection interval 9 3 involving multiple database TREND installations 7 8 request definition 7 16 reordering 7 11 collection of data incomplete 7 17 collection request saving 7 9 color 3 7 12 14 13 16 setting 2 13 column Rollup 9 10 type 9 10 column format TRENDsheet 11 15 Column Format option TRENDsheet 11 4 11 16 11 17 comma format 20 15 comma separated values CSV 20 2 command About 1 19 aga sheet h 11 21 agx gos 13 21 agx graph 12 22 buttons 5 7 Change Database 8 5 common 1 15 Constraints 11 4 Edit
59. Status 1 Up 2 Down Daily Interface Bandwidth Utilization Daily Interface Health Status Daily Interface Performance Statistics Avg 95th Percentile Peak Errors Discards ant Daily Interface Totals Number of Errors Discards Broadcast Packets and Total Packt Daily Interface Utilization Counts Number of Samples falling within a Utilization Rang All Types TRENDgraph Reports TRENDsheet Reports Figure 14 7 Report Types List Box Selecting a Report 14 11 14 Report Launcher 2 14 12 You can choose one of the following filters Report Type Retrieves No Filter The default report type which is all reports that are defined in the current Look in report directory TRENDsheet All TRENDsheet qss reports in the current Look in report directory TRENDgraph All TRENDgraph qgr reports in the current Look in report directory Grade of All Grade of Service gos reports in the current Look Service in report directory Hourly Reports All hourly TRENDsheet TRENDgraph and Grade of Service reports in the current Look in report directory Daily Reports All daily TRENDsheet TRENDgraph and Grade of Service reports in the current Look in report directory Weekly Reports All weekly TRENDsheet TRENDgraph and Grade of Service reports in the current Look in report directory Monthly Reports All monthly TRENDsheet TRENDgraph and Grade of Service re
60. TREND Estimating the Size of Your TREND Database While sizing a database is not an exact science you want to ensure that sufficient space is available to accommodate your raw and aggregated data tables While 500MB is sufficient for most evaluation purposes a minimum of 1GB is recom mended for data intensive applications of the TREND software To more closely estimate sizing for your TREND database follow these steps 1 Estimate the daily raw data table space requirement for each interface collected against nodes instances per node KB per row samples per hour 24 hours gt ga o 2 9 0 g TO 9 D 0 o m 3 B Database Administration 2 Estimate additional daily table requirements as follows Rate table Raw table Hourly table Rate table 3 Daily table Hourly table 24 Weekly table Daily table 7 Monthly table Daily table 30 3 Add the daily space requirements for each table to find the total space requirement 4 Multiply the total space requirement by the retention time specified for each data table in the TREND database raw rate hourly daily weekly and monthly to find the total space requirement for each data table 5 Add the total space requirement for all data tables 6 Multiply the total by 2 to find the recommended TREND database size Consider the following collection aggregation scenario Interface Number Instances Size
61. The TREND collection processes mw collect and rmon collect use port 161 by default You can modify the default on a per collection process invocation basis by using the P option uppercase on the invoking command Creating the Sybase Interfaces File on UNIX Systems You use different utilities to create the Sybase interfaces file for Sybase 11 0 x 11 5 and 11 5 1 on UNIX systems as described below Create the file on the server that runs the TRENDcopy command For Sybase 11 0 x Use the sybinit utility to install the Sybase interfaces file for Sybase 11 0 x on UNIX systems Perform these steps 1 As user sybase bring up a terminal window 2 Bring up the sybinit utility by typing the following command D 8 Creating the Sybase Interfaces File on UNIX Systems TREND 3 6 User s Guide SSYBASE install sybinit 3 Select Edit View Interfaces File from the main sybinit menu 4 Select Add a New Entry from the INTERFACES FILE TOP SCREEN 5 Enter 1 for Server Name press Return and then type in the name of the sec ondary Sybase server not to be confused with the server s host name after the prompt Press Ctrl a to accept your entry 6 Select Add a New Listener Service from the SERVER INTERFACES FILE ENTRY SCREEN 7 Enter 1 for Hostname Address from EDIT TCP SERVICE press Return and then type the secondary server s host name after the prompt 8 Enter 2 for Port from EDIT TCP SERVICE press Return and then type th
62. This option is equivalent to the bDelta_time parameter in TRENDpm A 21 gt s D v N Q o o mw_collect Z Sets the debug level for child instances of mw collect Values of 0 1 2 or 3 are valid The higher the number the more detailed the informa tion The default is 0 which means no debug output Debug output is written to standard out You should only use this option for testing in coordination with DeskTalk Systems Technical Support due to the additional overhead it places on mw collect The d option sets the debug level for the parent instance of mw collect The d and z options can be used together to control the debug level of parent and child mw collect processes independently Z This option specifies whether to turn on debugging or logging or both for the bcp gateway process When the logging option is turned on the information is written to the log file named SDPIPE HOME tmp bcp gateway dbg log Valid values for this option are 1 Turn on logging 2 Turn on debugging 3 Turn on both logging and debugging This option is in UPPERCASE Note If any logging option is set to on and you are running trendtimer sched add an entry to trendtimer sched to backup log files To do that add the following command to trendtimer sched DPIPE_HOME bin log_backup f lt input file name gt Description mw collect is TREND s SNMP polling application for generic MIB tables Its sim plest use is
63. To utilize this feature select Import Nodes from the File menu of the main TREND window A dialog box will request the name of the node view or type file Enter the name and click OK The format of the imported file must be node name read community write community poll type type name view name group name Required fields are node name read community write community and poll type snmp or rmon If type name or view name is specified group name must also be specified Missing fields in a record must be represented by the comma delimiter You can also issue the import nodes command from the command line The syntax is import nodes f import file h V where f Reads import file h Displays the command line options help V Displays the current version of the import_nodes utility Display and Menu System TREND 3 6 User s Guide The View Menu The View Menu contains one option Select Node Select Node Select Node is used to set the initial target node value for polling This value applies to the MIBwalker tool and the Collect Data application module Other targets can be set within those modules The Configuration Menu Configuration o o c E 0T 20 o Configuration menu options allow the user to define users groups views and types for set up and configuration of the system to be reported on Command options in the Configuration Menu are only available to your TREND admi
64. a max_processes wait_time debug_level thread wait time min disk pct s s i G debug level pm A c C d D E clock error value F g G h hostname i interval I check index k K suppress spikes E M min filter value n N retry interval o timeout p max entries per pdu P snmp port mw_collect log_info_level retries q r R min rows s round factor S snmp version t tablename w high water mark W high water mark log X Y delta time z child debug level Z debug log bcpgateway Options mw collect has the following options a Directs the child collectors of mw collect to output the collected data in ASCII format You may want to do this for debugging purposes gt D v o Q o o A Disables archiving of raw data The default is archiving When you use this option it turns off archiving This option is in UPPERCASE This option is equivalent to the bArchive parameter in TRENDpm C Specifies the number of collection processes to run concurrently When mw collect starts it starts child processes that actually do the collec tions The default is 5 mw_collect C d D E Specifies the number of minutes that each child process can run The system kills the child process if it runs longer than the specified n
65. an ANA Pa tee LARES 3 4 The View Menu 2 0c e eee eR III nh 3 5 Select Node dances macto x Ra Mai Poa eal Mee EO s 3 5 The Configuration Menu llle 3 5 Define Users and GroupS 0 00 eee eee ee 3 6 Define NOTES cs oor Rp ERR SEE ERE e s 3 6 Define VIOWS niece aa btn Sete ach d ete pama ERR 3 6 Define Types scc ute Npa cena at achame wd ae PUE Bie 3 6 Specify Report Directory 00 0 00 e eee 3 7 Edit Descriptions ser seeded aa Deu RI ale een UR a R 3 7 Set Application Color 00 ee 3 7 TOOIS 5 5 6x ad e NA Cas KANAN AND ANA EAE had 3 7 Autopilot PAA AA Bah ee BR eh ee Ee ee eS 4 1 DiSCOVEF AA he fates 5 1 What Discover Does ta cece ee eens 5 2 IP DISCOVOLy sse teeta eles ad ata mah pp hab ee uS 5 2 How IP Discovery Works lessen 5 3 IP Discovery Table Population l l 5 3 SNMP Type Discovery 00 AAEE ERREKEN RA Raie 5 4 How SNMP Type Discovery Works aaa 5 5 SNMP Type Discovery Table Population 5 5 Using the Discover Graphical User Interface 5 6 Launching the Discover Interface llli sls 5 6 The Start Stop and Cancel Buttons 5 7 Defining an IP Discovery eee 5 8 Subha Mask ocu s ue wd Mare pee Ab 5 9 Setting a Range for Discovery elles 5 9 Contents About Community Strings a 5 10 Host Name Translations
66. change the type in the sequence list too MIBlet Names Data table names in TREND have two names Object Name and SQL Name The Object Name is the name the user sees when selecting a table through the TREND GUI for example when defining data collection or creating a report The SQL Name is the actual name of the table in the database The Object name is derived by combining the name of the MIB file with the name of the MIBlet joined by an underscore _ The file name of the MIB file can be anything allowed by the operating system DeskTalk recommends that you do not Preparing MIBs for MIBwalker 6 15 6 MIBwalker 6 16 use dots or periods in the file name other than the mib extension since the mib file named used in the Object name is truncated at the first dot or period in the MIB file name The corresponding SQL name is created according to the following rules The string is converted to all lower case characters Hyphens are replaced with underscores If the string is more than 17 characters long it is truncated at the 17th character from the right If the resulting first character is not a letter it is changed to an x An underscore is appended to the resulting name If the resulting SQL name is unique it becomes the name used in the database 0 9 9 If the resulting SQL name is not unique the trailing underscore is changed to a zero 0 If it is still not unique the zero increases by one unti
67. database TREND is endowed with powerful features which trendadm can access for you This manual focuses on the installation and use of TREND by the normal user Some of the advanced features are listed here and described in detail in TREND Configura tion for trendadm on page 2 1 Configuration Distributed Polling Multiple Databases 9 9 Database and Partial Database Replication Advanced TREND features L TREND 3 6 User s Guide Interface Key Descriptions ET pia f Q 7 E 3 Commonly Used Commands Some commands appear in multiple TREND application modules for example The Open command appears in the File menu of the TRENDsheet TRENDgraph Grade of Service and TRENDbuild modules These common repeated commands will be described here rather than repeating their descriptions in each chapter The Open Command The Open command provides the user with access to the Select File window where report query files can be viewed To run a report by name select the Open option under the File menu This brings up the Select File window Select File 2 x1 J BayNetworks Drilldown Serial Interfaces Rmon l Frame Relay CO RMON Basic CJ Getting Started Z3 Scripts C3 Mib ll Drilldown Ethemet RMON CJ Misc C9 Drilldown LAN Interfaces Qj Performance Figure 1 2 The Select File Window Commonly Used Commands 1 15 1 Introduction to TREND Initially the Select File windo
68. define device groups 7 13 groups 2 3 nodes 2 5 types 2 9 users 2 2 views 2 7 define graph attributes 12 11 Define Nodes option 2 5 read community string 2 6 write community string 2 6 define plot lines 12 8 Define Types option 2 9 Add Type 2 10 Current Types 2 10 Delete Type 2 10 edit a type 2 10 New Type 2 10 Define Views 3 6 Add View 2 8 Define Views option 2 7 Current Views 2 8 Delete View 2 9 New View 2 8 Defining nodes node type 2 6 Index 5 Index read community string 2 6 write community string 2 6 types 2 9 2 10 add a new type 2 10 add type 2 10 delete a type 2 10 edit a type 2 10 list current types 2 10 views 2 7 add a new view 2 8 add view 2 8 delete a view 2 9 list current views 2 8 delete line 12 16 node 2 7 tables 2 18 type 2 10 views 2 9 Delete button 2 18 7 13 delete data out of database A 4 delete files automatically A 2 Delete Node button 2 7 Delete table s option 8 4 deleting a table 2 17 delta data 9 1 9 5 9 7 9 9 9 16 rank 9 24 value 1 10 9 5 16 5 delta time 10 3 delta_time 9 5 9 6 9 9 9 12 9 14 A 80 Description 6 5 DESCRIPTION field 6 14 destination table name 18 7 device 5 1 5 5 5 11 5 16 7 1 7 6 14 14 17 6 17 9 17 10 20 2 description 14 15 dynamic 7 1 level of performance 13 10 malfunctioning 9 8 Index 6 name 14 15 offline 9 8 SNMP manageable 5 2 tables 3 7 Device Filter list box 14 15 device key pair 12 12 13 13 directed instance poll
69. different than the type used for Vendor B s router Vendor functional groups overcome that distinction This can be viewed as a way to subset larger groups For example in a large network there may be dozens or hun dreds of routers You may want to focus on the core routers for polling purposes Create a Core Routers group which is a subset of the Routers group discussed above Some Considerations When Organizing Polling TREND 3 6 User s Guide Remember that the polling processes ee_collect mw_collect and rmon_collect read the membership list for the group being polled each time it starts a polling cycle Changes made to the group are automatically reflected in the list of nodes actually polled for the data Note Views and types are defined separately for each user group Polling must be defined by a user in that group for polling to occur Any user can only belong to one group Polling Frequency The type of data collected in a polling request and the reason it is collected directly affect your decision regarding how often to poll for the data For example system configuration data that rarely changes or disk utilization data that changes gradual ly may only need to be collected once a day Dynamic performance statistics with characteristics that vary with the time of day will need to be collected many times each day to give an accurate picture The number of minutes between polls also de pends on how much th
70. e eee ees 6 11 Syntax Rules nana caw ERU KING rad eee De pitna Deed 6 11 TORO MIB Tr60 52 i e he Rm NA AN edhe Sess 6 11 Description Lengi reris eeri tida te opi ae AEE e ERAR 6 12 Object Name Length aana naaa 6 12 STATUS Field Values 0 00 LASERE ERENER RES 6 12 Convert SNMP v2 Statements 00 0c eee eee 6 13 Validating Object Types cee ee 6 14 vi TREND 3 6 User s Guide MIBlet Names anan wan xad CEERI REER RESY 6 15 Troubleshooting a MIB eresse tanneri cee eee 6 17 Recommended Procedure a a auauua lees een 6 18 MIBlets 52 55 codecs CR Ba ADAN oun Na ede beads 6 19 Utilizing 64 Bit Values BB 6 19 Creating a MIBlet R cee ees 6 20 Utilities xo Abe Eee NG EN p iad BA t oi 6 29 Changing the Current Node eee eee 6 29 Changing Community Strings lees 6 30 Polling Data to Screen 1 2 2 ee 6 31 SNMP SE 2 e KA bea aie ce oo ee ot en ag 6 33 Setting A Single Variable llli 6 34 Setting Multiple Variables in a Table 6 35 Adding a New Key 00 eee ete 6 36 Collect Data a GA ne Glo ee ease e a Goda a REN IRE 7 1 Colect Data tar souks fan Peete eh oe CREAN NG MAAM ANAN 7 2 Menu Command Option Summary llle 7 2 ThesFile Men ceased due cce be fd ach eed ARR RETE 7 3 Collect Data Periodically 0 00000 e ee eee 7 3 Change Database 0 0 c eee ee eee 7 3 The View Men
71. entries in the Select Device list box that match the specified search value Click Clear to clear the text in the Search text box 6 Click on the device name to select it The name of the selected device is highlighted After you select the device the elements for that device are retrieved from the database and are listed in the Select Element list box Note If you selected All Devices the Select Elements and Element Filter lists are inactive 7 Ifyou selected a specific device you can now select the element on which you want to report from the entries in the Select Elements list box All Elements is the default You can restrict the entries on the list by selecting a filter in the Element Filter list box You can choose one of the following filters Filter Retrieves No Filter No filtering criteria is applied to the list the default 14 16 Selecting Devices and Elements TREND 3 6 User s Guide Filter Retrieves Element Name The Match Value window for the element name appears if you select this filter The window is similar to the Match Value window for the device name see Figure 14 10 You can specify an element name such as Async1 1 or a partial word string such as Async in the text box Select the Partial Word check box if you specify a partial word string The filtering criteria is applied to the Element Name column in the database the elements that are retrieved f
72. files will be named lan interface load mynodel 1 timestamp lan interface load mynodel 2 timestamp lan interface load mynode2 1 timestamp lan interface load mynode2 2 timestamp lan interface load mynode3 3 timestamp lan interface load mynode4 1 timestamp The fourth report would produce the same data as the third report but it would only produce a single output file called lan interface load timestamp In addition the output file would be comma delimited and condensed Example 3 start report myreports lan interface load qss output myreports lan interface load max n begindate 03 01 97 finishdate 03 31 97 instances mynodel 1 mynodel 2 mynode2 1 mynode2 2 instances mynode3 3 mynode4 1 style combine format comma condense y end Input Parameter File TREND 3 6 User s Guide Example 3 causes trend export to run a set of reports in a combined format i e all the data from the listed instances is combined into one output file for only data whose ta_period lies between March Ist and March 31st Example 4 start report myreports critical_interfaces qss output tmp critical_interfaces_list frequency daily exception y max n begindate 03 01 97 finishdate 03 31 97 format comma end start report myreports lan_interface_performance_report qss output myreports heavy_hitters frequency daily max n instan
73. finishes with a process id number Pid attached to each entry Error messages appear in the same format as general messages The following log entry shows an example of a process initiation message followed by a process error message and then followed by a process completion message trendtimer Fri Apr 10 01 20 01 1998 Pid 6675 usr TRENDsnmp bin trendit s trendit Fri Apr 10 01 20 01 1998 Server net xyz com 30 No licenses currently available trendtimer Fri Apr 10 01 20 01 1998 Process id 6675 terminated 1 sec Trendtimer initiates the process called trendit Trendit displays an error message and then trendtimer displays the termination message for trendit Process Initiation Messages The format of the message portion that appears at process initiation is the following Pid nnnn lt schedule file name gt lt options gt where nnnn is the process id number schedule file name gt is the path name and file name associated with the specific process lt options gt designate the parameters that modify how the process will run Message Handling TREND 3 6 User s Guide For example the format of the message portion that launches a collection request such as mw_collect or rmon_collect is the following Pid nnnn schedule file name gt n i xx pyy where nnnn is the process id number lt schedule file names is the path name and file name associated with mw collect or rmon_collect N E n indic
74. free up space in the log 1 Determine approximately how many pages the transaction log occupies using the commands 1 use dpipe db 2 go 1 select data pgs 8 doampg 2 from sysindexes where id 8 3 gt go Recovering from Sybase Error 1105 TREND 3 6 User s Guide where 8 is the ID for syslogs The result is the number of data pages 2K pages on most operating systems that the transaction log occupies Note that the query results may be inaccurate by as many as 16 pages but using the query is much faster than counting the rows in the transaction log 2 Dump the inactive portion of the transaction log using the dump transaction command with truncate_only option If this command also fails with the 1105 error run dump transaction with no_log 3 Repeat step 1 If however there are still a large number of pages in the syslogs table an outstanding transaction is probably preventing the log from being cleared If this is the case check syslogshold for old transactions for the database for which the error occurred in this case it is dpipe db 1 use master 2 go 1 select from syslogshold 2 where dbid db id dpipe db 3 gt go Determine whether the oldest active transaction can be terminated it may have been left inactive intentionally Continue this procedure until there are no other old transactions that can be terminated You can clear a long running transaction in one of two ways Using the
75. hard copy of a report select the Print option under the File menu in the TRENDgraph TRENDsheet or Grade of Service windows This action will bring up the Print window For Windows NT and Windows 95 systems the following Print window will be seen T Print to File tmp graph bmp Print Background Cancel Figure 1 3 The Print Dialog Box In addition to postscript the default TREND also supports bmp and gif files To save reports in these formats click on the Print to File box and add the appropriate extension either bmp or gif to the file name You can also modify the default file name to save the report file under any desired name Enter the name of the printer where you want the report run on the Printer line for Windows use the Printer Setup window Be sure to send the report to a postscript printer for a postscript bmp or gif file There is no default value for the printer Enter the number of copies on the Number of Copies line The default value is 1 Report query files can be printed to a printer or a file with or without a background by clicking on the Print Background box Click on the Print button to send the report to the printer Click on the Reset button to change the values back to their original values Click on Cancel to quit out of the window without printing Commonly Used Commands L TREND 3 6 User s Guide Exit 3 Pe f ex 7 3 s T
76. in a file that can be stored anywhere the user chooses allowing cre ation of a report library TRENDstep A command line application that conditionally processes each row of a rate input table and stores the results in a new output table This processing enables you to combine data from multiple rows of an input table into a single row in the output table and compute new columns of data in the output table TRENDsum An application that takes the cleaned rate data provided by TRENDit and summarizes it into many useful statis tics trendtimer A process that controls the collection trending and delet ing of data The trendtimer sched file contains all the entries for mw_collect rmon_collect ee_collect TREN Dit db delete data and Ipr_launch 5 of 5 1 8 TREND Overview L TREND 3 6 User s Guide A Brief Tour of TREND EI mg fo Q 7 E 3 ap Collecting Data The Collect Data application is used to define enterprise data polling policy It was designed specifically to handle SNMP data structures MIBs A polling policy is a set of requests to collect groups of MIB variables from one or more devices at reg ular intervals The actual polling of data is carried out by TREND s polling agents ee_collect mw_collect and rmon_collect Launched periodically these agents de termine their polling activities based on the defined polling policy TREND users can collect spec
77. is added in all roll up tables It About Data and Data Tables in TREND 10 3 10 TRENDbuild 10 4 reflects the time since the data source was rebooted in hundredths of a second at the end of the measured time Each raw data table and its associated roll up tables have a corresponding key table Key tables are used to maintain a unique index like value for each unique data source e g a router name and interface number ifIndex for MIB II ifEntry data Key tables also contain a description column where information about the data source can be placed In addition some key tables contain additional configuration and descriptive information about the data source e g RMON key tables contain interface type and speed information These values can be selected by TRENDbuild for use in reporting Rate table rollup primarily deals with counter variables The structure for the raw and rate tables is virtually identical Rate tables are generated by selecting pairs of rows from the raw data tables successive data points The delta between the points is calculated and the result inserted into the rate table The values for most columns in the new rate table row are the same as the more recent of the two samples Note The term rate table is something of a misnomer as used here Rate tables can be used to calculate rates by dividing counter variable deltas by delta time Unless captured as gauges no rates are directly stored in the rat
78. is displayed in alphabetical order in the Current Views scrolling list To select an existing view click on the downarrow of the Current Views box to bring up the list then select the name of the view you want 2 8 Defining Views TREND 3 6 User s Guide To delete a view select its name in the Current Views list and click on the Delete View button Note To edit a view you must delete the old view and create a new one using the methods described above Click on the Cancel button to quit out of the window N E m 032 ag no 3 o as o Q Q EI P at e D Defining Types You can add types to the database through various methods One method allows you to automatically identify the device types on your subnets by using Type Discovery from the command line See Running SNMP Type Discovery on page 5 17 You can also create a type by selecting the Define Types option under the Configuration menu which will bring up the Define Types window pa 134 70 18 255 wnee pawnee desktalk con 134 70 65 10 slurpee 134 70 65 220 slurpee desktalk corr 134 70 65 255 cedar desktalk com cowpasture desktalk dirty devil dirty devil desktalk c hardware desktalk cc hidekel desktalk com Figure 2 5 The Define Types Window Defining Types 2 9 2 TREND Configuration for trendadm You can add a new type by typing its name in the New Type box You can then click on the nodes you would like to incl
79. is insufficient application problems such as the following can ensue Cannot obtain lock Failure to connect Be sure the number of locks is adequate Over allocating locks is rarely detrimental By default Sybase defines 5 000 locks Desktalk Systems recommends a minimum of 20 000 locks Implementations that want to run a large number of baby collectors or parallel instances of TRENDcopy and TRENDit should use 40 000 to 50 000 locks Sybase Configuration Parameters C 15 C Performance Tuning Issues 1 To change the number of locks isql Usa P sp configure number of locks 40000 go This example sets the number of locks to 40 000 2 This change is static Restart Sybase to implement the change Binding the Key Tables C 16 Placing all key tables into their own cache can help performance by requiring no page I O when the key tables are referenced thereby reducing disk access There are two necessary phases to implement this functionality they are creating the named cache and the binding of individual key tables to the named cache Creating the Named Cache The first step is to create a named cache to bind the key tables into such as keycache It is advisable to implement this functionality after the TREND database has reached its normal size to avoid under or over allocating the amount of cache necessary After you create the cache named keycache you must shut the Sybase SQL Server down and restart it for th
80. list of all users presented Reordering the List of Collection Requests You can change the sort order of the displayed requests Click the Sort Order option in the View menu to bring up the Order Requests window N e o D oO D r 5 Managing Polling Requests 7 11 7 Collect Data Order Requests x Order by node 2 a 0K Cancel Figure 7 5 Order Requests Window The current sort order is shown in the Order by box By default the requests are sorted by node name To change this first click on the Clear button to remove the current sort order Next click on a column name in the Select box that you want used as the new sort criteria Activate the selected value by clicking on the Add button you can add more than one field to the Order by list The order in which you add the fields determines the sort order of your results Collect Data does a nested sort beginning with the first field listed in the Order by box then the next field within the first and so on When the Order by list is as desired put it into effect by clicking on the OK button Click on the Cancel button to exit the window If you cancel without applying a new Sort order the order will not change Modifying Request Definitions Existing requests can be modified by the user who created them as shown in the User column or by the user trendadm To modify a request click on the request s line in 7 12 Managing Polling R
81. main TRENDbuild window Figure 10 1 can be invoked from the Tools menu of the main TREND window or from the TRENDgraph or TRENDsheet applications 10 1 10 TRENDbuild untitled Figure 10 1 TRENDbuild Main Window The File menu shown below is used to create access and exit reports About Data and Data Tables in TREND TRENDbuild creates and or edits report templates which are comprised of files con taining report set up definitions and a specially formatted SQL query TRENDbuild requires neither SQL nor network expertise to use it In order to make full use of TRENDbuild s functionality however some basic understanding of the data and data formats available in the TREND database is required 10 2 About Data and Data Tables in TREND TREND 3 6 User s Guide Each group of MIB objects or miblet being collected has a corresponding raw data table in the TREND database Raw data tables in TREND are populated with timestamp constants statistics and other values Statistics may be counters or gauges A counter is a raw count of a statistic since a device was rebooted A gauge is an instantaneous measurement Examples include percent CPU utilization percent buffer utilization or the number of buffers used at a point in time Gauge values apply to a particular instant in time while counters are cumulative over time TREND includes a module called TRENDit that aggregates or rolls up raw data TRENDit runs on a batch mo
82. many others RMON2 agents are those agents that can store some or all of the information that is unique to the RMON2 MIB The RMON2 DataPipe poller vastly increases TREND s RMON reporting capa bility both for RMON and for RMON2 Note In the sections below each category of table shows two distinct tables Each of these pairings represents a data table in which collected data is actually stored with its associated control table which defines the for mat of the data table Data tables are identified by the suffix data in the Database Name column Control tables have no suffix 17 1 17 RMON2 Support Source Data Available from RMON2 RMON Data The RMON DataPipe supports extensions to standard TREND RMON tables The only difference between the regular RMON tables identified by the prefix rmon_ and the extended RMON tables identified with the RMON2 prefix rmone is that the extended tables contain Dropped Frames and Table Creation Time information To avoid duplicate collection be sure to collect rmone_ tables only from probes that support the RMONZ2 extensions Note You should use only the RMON2 DataPipe to collect data from agents that support the RMON2 MIB The table extensions described below cannot be collected from regular RMON agents Use TREND s standard RMON DataPipe to poll agents that support the standard RMON MIB Supported RMON Table Extensions There are twenty RMON table extensions suppo
83. more than one column at a time from the Columns in Selected Table Box There are three ways to do this You can click and drag the mouse pointer over a series of columns Each column is highlighted Working with TRENDdbview 22 15 22 TRENDabview You can highlight one column hold down the SHIFT key and select another column Those two columns and all of the columns between them will be highlighted Ifthe columns you wish to select do not appear in sequence hold down the CONTROL key and select the columns you want Each of the columns you click will be highlighted If you insert columns by accident you can select and delete them either one at a time or as a group Highlight the column s you wish to delete and click Delete Any highlighted column is removed from the Columns in View Box If you wish to restore the Columns in View Box to its default setting click Default The Columns in View Box is returned to its default condition 4 When the information that appears in the Columns window is correct click OK The Columns window disappears and you return to the TRENDdbview main window Define Expressions for a View 1 Click the Define Expressions Button The Define Expressions window appears with your selected data and key tables showing in the Selected Data Tables Key Tables Box Since it is not necessary to define expressions for any view TRENDdb view does not apply predefined default expressions Only those ex
84. name The ASCII file will not include column headers C Condense the resulting y Condense the ASCII file ASCII output file In an uncondensed file each column is padded with trailing spaces up to its default width from its database definition A condensed file has these trailing spaces removed n Do not condense the ASCII file DEFAULT d Show debug output If 0 No debug output this option is included the 1 2 or The higher the number the entries made in the log 3 more detailed the debug output file are also displayed on your screen This option should only be used for testing in coordination with DeskTalk Systems Technical Support due to the additional overhead it places on rmon_collect 1 of 3 Controls and Switches 20 3 N j D m Z o o bad o s D 20 TRENDexporter line options Flag Description Argument f Specifies the format of tab Generate a tab separated ASCII the ASCII output file output file DEFAULT comma Generate a comma separated ASCII output file Character data is enclosed in quotes h Help Display command N A i The full path and file name of the input param eter file The name and directory of any valid parameter file m Only include data with the most recent times tamp i e the maximum ta_period value in the source data table y Only include data with the most recent timestamp n Include all
85. name with the Transact SQL shutdown command 1 Use isgl to log into a SQL Server account with System Administrator privileges isql Usa Ppassword Sserver name Enter the following command to shut down the specified backup server 1 shutdown SYB BACKUP 2 go Backing Up the SQL Server Installation TREND 3 6 User s Guide When you use the shutdown command to stop a backup server specify the svrname listed in sysservers for Backup Server SQL Server s default svrname is SYB_BACKUP Creating a Dump Device The syntax used to create a dump device is sp addumpdevice tape disk devicename physicalname size Examples Create a dump device to a tape For UNIX isql Usa P 1 sp addumpdevice tape tapedump dev rmt 0mn 200 2 go For NT isql Usa P 1 sp addumpdevice tape tapedump tape0 200 2 go 2 Creating a dump device to a disk For UNIX 1 sp addumpdevice disk diskdump export home diskdump dat 2 2 go For NT gt 2 w 3o 3D a uo Y o a0 om El Backing Up the SQL Server Installation B 19 B Database Administration B 20 15 sp addumpdevice disk diskdump C HOME diskdump dat 2 25 go In the first example the tape device must be a non rewinding device and the size parameter is mandatory This example specifies a size of 200MB the actual value should be just larger than your actual database size The
86. oe eee En 10 1 Main Display and Menu System 000 eee ee ee 10 1 About Data and Data Tables in TREND 00 0000 10 2 Creating a New Report ea e eee 10 5 Selecting MIBs and Groups 0000 e eee ee eee 10 6 Selecting Statistics 0 0 aaa auauua 10 8 Assigning Aliases and Creating Expressions 10 10 Build an Expression n EEEE EEEE RERI 10 13 Modifying an Existing Report 00 0 0 cece eee 10 14 TRENDSheet si eye ee sea eee ANDENG 11 1 Main Display and Menu System 000 eee eee 11 1 Menu Command Summary cee eee eee 11 2 Opening an Existing Tabular Report 0 00000 e eee eee 11 4 Using an Existing Report as a Template 11 5 Creating a New Tabular Report 11 5 Selecting Statistics 0 0 ee ee 11 6 Modifying the Tabular Report 000 0 ce eee ee eee 11 9 Specifying Data to be Included in the Report 11 9 Select the Node and Key Being Reported 11 9 Displaying Data from a Different Database 11 11 Setting the Time Period liliis 11 12 Establishing Constraints llle 11 12 Formatting the Report cece eee ee 11 14 Sorting the Data in the Report 0 ee eae eee 11 14 Formatting Columns in the Report 000 eae 11 15 Column Order co eair 00 ee tee 11 15 Column Order and Report Sort Order
87. or Creating the Sybase Interfaces File on Windows Systems on page D 15 for details Be aware that TRENDcopy has the following conditions TRENDcopy only copies the definitions of key and data tables to another data base when the key table of the destination table is missing an identity column such as a Report Package demo In this case use either the BCP or BCPfile utility to complete the transfer TRENDcopy only copies a source data table with a corresponding key table that is a view when the destination table has a key table present and is also con structed as a view TRENDcopy copies a source data table constructed as a view to an existing destination data table constructed as a view otherwise it creates the destina tion data table as a table TREND 3 6 User s Guide TRENDcopy Syntax trendcopy TRENDcopy Syntax b length unit c column name exp value C key_column_name exp value d debug level e yyyymmdd yyyymmdd f day of week day of week g hour hour 4 length unit m minimum threshold for bcp transaction 100 M maximum threshold for bcp transaction 100 n node type p polling list id r table type R table type s source server S target server t source table destination key table destination table T source topology database u U user name v view name V w high_water_mark 90 18 3
88. or poor is applied to the weighted met rics This information allows network service providers to concentrate on areas where performance is substandard Since GOS can be set up with metrics from any SNMP MIB users can develop customized GOS reporting for routers servers LAN segments and WAN links Custom Reports The Report Scheduler module selectable from the main menu allows you to create customized charts and graphs using a point and click interface You select the data source and the set of statistics you want in your report TRENDbuild allows you to assign aliases to MIB statistic names and build computed variables using selected statistics constants and arithmetic operators Expressions are very useful for com puting relative utilization for example Once you have created the report specifica tions you can specify the name of the report query file Report Launcher The Report Launcher is an easy way to run an existing query file against a network element This interface allows you to select a report group e g routers report type e g GOS and a network element Report Chaining Related reports can be linked together for automatic printing via the Report Sched uler option A series of related drill down reports provide a detailed picture that is unavailable from any single report A Brief Tour of TREND L TREND 3 6 User s Guide TRENDweb Report Building and Viewing Facilities TRENDweb 3 2 provi
89. perf raw default day framerelay dte rfc1315 D5 frcircuitentry daily hour framerelay dte rfc131 H5 frcircuitentry hourly rate framerelay dte rfc1315 R5 frcircuitentry rate week framerelay dte rfc131 W5 frcircuitentry weekly month framerelay dte rfc1 5 M5 frcircuitentry monthly 0 32 default framerelay dte rfc1315 frCi x5 frcircuitentry raw default 756 256 default wfline 780 CSMACD Error x80 csmacd errors raw default 32 default day wf line 780 CSMACD D80 csmacd errors daily 0 32 default hour wf line 780 CSMACD H80 csmacd errors hourly 0 32 default 4 Figure 8 1 Data Manager Main Window Database Information The Database Summary Information area located between the menu bar and the tab ular display provides the following information SQL Server Name TREND can read and store data on any SQL server in the user s network so it is important to be able to monitor all servers from a single location This box indicates which server s tables are currently being displayed db used This tells the user how much of the space available in the TREND database on the SQL server currently being monitored has been used The database is full at 96 Data collection will cease at 90 until the size of the database has been reduced 8 2 Main Display and Menu System TREND 3 6 User s Guide Tabular Display The tabular portion of the main Data Manager window includes the following Object Name
90. plify administration Not all of these suggestions apply to every site Use the ones that suit your needs Group By Function Group By Information Available Group By Location Group By Vendor Group By Priority In general a particular type of device plays a particular role in the network This may be obvious but it needs to be taken advantage of by actually creating a polling group for nodes that provide that function For example create a group that contains all routers all file servers or all hubs While TREND supports polling by device type different types may be assigned to the same kind of device from dif ferent vendors or even to different models from the same vendor Functional groups overcome that distinction You may choose to create node groups according to a par ticular type of information available from all the nodes in the group In a large network it may make sense to subdivide groups based on the devices physical locations so polling can eas ily be distributed to the polling system closest to the target nodes Grouping by location makes it simple to setup poll ing of group Location A by Poller A and of Location B by Poller B While all devices of a particular type can be polled as group by specifying Poll by Type in the request definition it may be the case that different types are assigned to the same kind of device from different vendors For example the type used for Vendor A s router may be
91. polling interval is shorter than the amount of time spanned by the history table instance or if there will be a gap in the data because the interval is longer than the time spanned by the history table instance e g If a history table instance has 50 buckets each covering a 30 second interval polling that instance of the table once an hour with a i value of 60 will leave a gap of 35 minutes in the data at each poll Information about these gaps and intervals is written to DPIPE_HOME tmp trend log By understanding how rmon_collect works and defining history table instances according to the poll ing interval you use you can maximize the efficiency of your history table and your polling scheme The Matrix Table rmon_collect only collects the matrixSDEntry version of the matrix table Using the capabilities of the database the data in this table can read for any pair of nodes re gardless of which is the source and which is the destination node A 33 gt D v o Q o o rmon_collect Token Ring MAC Layer and Promiscuous Tables When polling a Token Ring RMON probe rmon_collect will automatically collect both the MAC Layer and Promiscuous history tables However because the user specifies which version of the Token Ring Statistics table they want when they de fine polling for those tables only the one selected in the Collect Data Periodically window will be collected by a given polling request Log File rmon
92. populated by rmon collect sumit Tables populated by TRENDsum tcopy Tables populated by TRENDcopy This option is mutually exclusive with the s and t options If you omit the f s and t options tables from all sources are aged which is the default Displays command format help A 5 gt s D v o Q o o db_delete_data A 6 i m S t The number of days that data can be retained in a table Data that has been in a table for one day more than the number specified is deleted The default is to use the aging value set for the table in the database Use Data Manager to view the table s default aging value See Data Manager on page 8 1 for details Performs key ID based deletions depending on the value of this option Valid values are data Deletes data using the key IDs selected from the data table key Deletes data using the key IDs selected from the property table ta period The data is removed based on time period only The default is data Ages data only in the table specified Enter the name of the table as shown in the SQL Name column of the Data Manager display See Data Manager on page 8 1 for details This option is mutually exclu sive with the t and f options If you omit the f s and t options all data tables are aged which is the default Ages data only in the table specified Enter the name of the table as shown in the Object Name column of the Data M
93. processing To this end an optimal configuration retains data as follows Type of Data Retention Period Raw 1 day Rate 1 day Hourly 7 days Daily 31 days Weekly 31 days Monthly 6 months C 22 TREND Processing Options TREND 3 6 User s Guide If you need to retain data longer you should use TRENDcopy to create an intermediate archive or reporting table For example assume you want to keep hourly mib_ii_ifentry Hib_ii_ifentry_ data for one month Use the following TRENDcopy command to create and maintain an intermediate reporting table 9 O D 2h 3 o O D el 3 Ce trendcopy t hour mib II ifEntry hour mib II ifEntry rpt s source server S target server The first invocation of TRENDcopy creates all the necessary data dictionary and table entries and copies the entire source table in this case hour mib II ifEntry to the target table hour mib II ifEntry rpt Subsequent TRENDcopy invocations copy only data from the source table that is newer than the data in the target table After you create the reporting table use the TREND Data Manager to change the retention period of the source table to 7 days and of the target reporting table to 31 days This ensures that TRENDit processes only the minimum amount of data on a nightly basis Note that the Raw and Rate tables typically have the most data then the Hourly tables The number of rows in the Daily
94. provide information on various aspects of TRENDlabel command usage Note that omitting alias from the command may produce an ambiguous error message such as prog name date time Ambiguous column name XXXX which indicates a situation where the key table and data table both have a column called XXXX The query uses both key and data tables in the from clause and uses only XXXX in the select clause it does not specify the owner table name To resolve this situation do the following l 2 19 4 Start an ISQL session from a command line with user dsi_dpipe Perform sp help on the key table to get column name listing Drop the improper column from the key table with the following syntax alter table lt tablename gt drop lt column name gt Exit the ISQL session Correct the TRENDlabel syntax to use alias column Usage Notes TREND 3 6 User s Guide Populating the dsi_descr Column When you generate a TRENDgraph or Grade of Service report the report title appears above the graph and the legend appears below the graph If the value of the dsi descr column in the key table the report uses is null or blank the legend simply displays the name of the node on which the report is based However if the dsi descr column is populated the legend displays the description that appears in the column This can make graphical reports easier to identify and more comprehensible Locating the Target Row in the Source Data Table
95. provides the data to populate the dsi descr column in the key table Displays the syntax of the trend label command Identifies the destination key table Specify the SOL name of the table This option lowercase letter o specifies the hour of the previous day Valid values are 0 midnight through 23 This value is used to locate the row in the source data table associated with the dsi key id that is used to populate the key table columns If this option is omitted the default target row in the source data table is the one where the maximum ta period that is less than or equal to the current time the minutes and seconds portions of the ta period value are set to zeroes minus 1 hour on today s date See Locating the Target Row in the Source Data Table on page 19 5 for a detailed explanation of how the target row is located Has the same syntax and application as the c option with the following exception The target value from the source data table populates the associated column in the key table even if that column already exists and has a nonnull or nonblank value 19 3 e J m z o ley a 19 TRENDlabel t Identifies the SQL name of the source data table The table is typically a rate table but can be a raw table Summary tables cannot be used as a source Displays the current TRENDlabel command version For example a return value of 3 5 means TREND version 3 5 Usage Notes These notes
96. r 5 Figure 7 2 Collect Data Periodically Window The Poll For box in the window displays the names of the variables or tables that can be defined Click on the down arrow next to the box to see a list of variables and tables available for selection Setting Up New Polling Requests 7 5 7 Collect Data Specifying Which Nodes to Poll The Polling Mode box lets you define the polling target grouping for this collection request Membership in so if nodes are added or the list of target nodes is determined at each polling cycle deleted the change will automatically affect which nodes are polled for this data Click on the down arrow next to the box to see a list of the nodes and select the one you want Alternatively type in a node name if Commu nity String values are public and private The possible values are individual node individual node by key all nodes of same type 7 6 Polls only the node named on the Select Node line This is the node named in the header bar of the window You can change the default node name using the Node List box Click on the down arrow next to the box to see a list of the nodes available and select the one you want TREND allows you to conduct directed instance polling using previously assigned keys Use the Select Keys but ton to view and select keys which define the data to be collected for a given node Each node in your network is some type of device It can be useful
97. same model Since the data set collected by each of these probes is identical to the others getting the protocols from one probe is the same as getting the proto cols from all probes When you are ready to get the protocols from an RMON2 manageable device launch the Protocol Definition Interface and click the Get Probe Protocols button The Get Probe Protocol Data window appears DataPipe Components 17 9 17 RMON Support Get Probe Protocol Data Figure 17 4 Get Probe Protocol Data Window In order to get probe protocols you need to supply the IP name or address and com munity string of an RMON capable device Enter the name of the device in the Probe field and the community string in the Community field Optionally you can also Specify a port number in the Port Number field Define the number of retries in the Retries field Define how much time must pass before the get request is timed out in the Timeout field 17 10 DataPipe Components TREND 3 6 User s Guide Figure 17 5 shows the window with a node specified Get Probe Protocol Data Lj N E fe 2 N o Ko 3 Add to Database View Invalid Change Invalid Name Figure 17 5 Get Probe Protocol Data Window with node identified When you have specified the node click Get Data The Protocol Definition Interface discovers all of the RMON2 protocols being collected by the target device actually
98. second example shows a disk dump device For disk dumps the size parameter is ignored but must be included 2 MB is entered here Note You can use Sybase Central in a Windows environment as an alternative to entering the above commands Dumping the Database The following example illustrates the procedure used to dump a database to a pre defined dump device isql Usa P 15 dump database dpipe db to tapedump with init 25 go OT 1 gt dump database dpipe db to diskdump with init 25 go The init option is recommended as it overwrites any existing data on the destination dump device Example of a Backup Script The following example shows a sample script with the corresponding procedure for backing up the database The script contains the commands to perform the backup Backing Up the SQL Server Installation TREND 3 6 User s Guide it can be regularly scheduled using a scheduler program such as cron on UNIX or At on NT systems First decide if you want to back up smaller databases such as master or sybsystemprocs to a filesystem rather than a tape device Perform either of the following sets of steps step A or step B for each database A Use fixed diskfiles defined via sp_addumpdevice 1 Adda new setenv line for defining the variable name to use an existing Sybase dump device that utilizes a disk file such as setenv MASTER DEVICE masterdump 2 Change TAPE DEVICE on the respective dump command to b
99. set for z TRENDit ignores the data for this sample interval on the assumption that the difference is due to the polled device being down for part of the interval It has been ob served that not all system clocks tick at the same speed sometimes causing the percentage difference to appear artificially high Set the option high enough to compensate A value of 30 will usually be enough The default value is 10 Examples A 42 To have TRENDit aggregate the data in all raw data tables in the database allowing data received up to 3 minutes after the hour to be included as part of the previous hour trendit f 180 To have TRENDit aggregate the data in the RMON host table omitting any rows where a host transmitted less than 100 octets packets and errors between polling samples and emptying the raw host table after rolling it up trendit t rmon hostdata p v 100 To have TRENDit aggregate the data in the mib II ifEntry raw data table trendit t mib_ii_ifentry_ TRENDpm TRENDpm Manages stored procedures identified by TREND trendpm c code gen file a db trace group trace level ela g n o c output file ot target table name pe parameter Zvalue parameter2 value pg parameter Zvalue parameter2 value x s db server name t table name proc app type proc class proc type v gt D EI UU i amp Options o c Specifies the name of code generated
100. set of nodes that are grouped together so they can be polled for the same data by a single collection request They may be grouped together because they share acommon function location or other characteristic which leads the user to want to monitor their performance the same way Views can belong to different groups To create a view select the Define Views option under the Configuration menu This will bring up the Define Views window Defining Views 2 7 2 TREND Configuration for trendadm Current Groups netadmin 3D Current Views fouer H New View fo Nodes in View Current Nodes 134 70 18 255 134 70 65 10 134 70 65 220 134 70 65 255 cedar desktalk com cowpasture desktalk dirty devil dirty devil desktalk c hardware desktalk cc hidekel desktalk com Add Node Delete Node Delete All n Add View Update View Delete View Cancel Figure 2 4 The Define Views Window You can add a new view by typing its name in the New View box You can then click on the nodes you would like to include in the new view As you select a node it will become highlighted hold down the shift key to select addi tional nodes When you have selected all the desired nodes click on the Add View button The name you entered will be included in the Current Views list Note To add another view you must first delete the nodes from the Nodes in View box then add new ones The list of existing views
101. solve these problems there is an extended form of the dotted name format which can be used to identify entries in the protocol directory table The format of extended dotted name notation is shown in Figure 17 9 lt extendedDottedName gt wildcard lt qualifiedProtoName gt lt qualifiedProtoName gt lt qualifiedProtoName gt name lt ecapVal gt lt parmVal gt lt ecapVal gt uint lt parmVal gt uint Where name protocol name from PI definition uint unsigned integer between O and 4G 1 Figure 17 9 Extended Dotted Name Notation The wildcard prefix is specified when the wildcard function has been specified with a base protocol layer identifier An ecap Val qualifier is only specified if the protocol has more than one encapsulation under a parent A parmVal qualifier is only speci fied if the value of the protocol parameter is nonzero For example the IP protocol has parameter countFragments This parameter is bit zero and thus has a value of one when turned on Putting it all together to specify that the wildcard function has been selected and the countFragments parameter has been selected for the UDP over IP over Ethernet II encapsulation the following extended dotted name is used wildcard ether2 ip 1 udp This specifies a value of 01000001 00000800 0000001 17h 1 0 0 1 0 0 8 0 0 0 0 17 for the protocol identifier string and a value of 000100 h 0 1 0 for the protocol parameter
102. startup file the entire trendtimer command is quoted to insure that the su command runs the en tire trendtimer command line rather than interpreting the trendtimer options as op tions to su The simplest way to start trendtimer manually is to become root and execute the command sh etc rc TREND timer up trendtimer The Schedule File trendtimer determines what it is going to do by reading a schedule file when it is first invoked By default it looks for the file in DPIPE_HOME lib trendtimer sched This can be overridden by the s command line option The schedule file is a series of statements in this format interval command line or lt interval gt lt offset gt command line lt interval gt is how often the command on this line will be run It can be either a num ber of minutes specified as a simple number MM or a number of hours and minutes specified in the format HH MM For example lt interval gt values of 120 and 2 00 are the same thing All intervals are counted in lt interval gt minutes or hours and minutes from unix 0 time which is midnight Thursday January 1 1970 Intervals chosen should therefore divide evenly into 60 minutes to insure that the process runs at a predictable time For example an interval of 5 minutes means the program invoked will run on the hour and at 5 10 15 20 etc minutes after every hour because these are 5 minute intervals since midnight Thursday January 1 19
103. string DataPipe Components TREND 18 TRENDcopy a JI m 2 o t So lt TRENDcopy allows the trendadm to copy data from one database to another Note Data transported by TRENDcopy goes from data table to data table Data copied in this way is only accessible by TREND The TRENDex porter see TRENDexporter on page 20 1 allows you to export data in ASCII text format TRENDcopy can Copy all the data in a database table and add that data to a similarly named table in another TREND database Copy only specified columns in a data table and insert them in another database Create a new data table in a database using information copied from a table in another database Filter the data it copies by type view date or time of day Filter the data it copies by columns in a data table or key table Copy only aggregated or summarized data 18 1 18 TRENDcopy 18 2 TRENDcopy has two common applications In a TREND system that uses Satellite Servers to clean raw data use TRENDcopy to copy the cleaned data from a TREND Satellite Server to the TREND Primary Server In a TREND system that includes an Archive Server use TRENDcopy to copy data from the Primary Server to the Archive Server Note A Sybase interfaces file defining the target server must exist on the server that issues the TRENDcopy command See Creating the Sybase Interfaces File on UNIX Systems on page D 8
104. take a rate table a TRENDsum output table or another TRENDstep output table as input TRENDstep processing is designed to Provide a method for intercepting a potential divide by zero condition in the input data and assigning an alternative value to the result Provide a way to group data into an other category For example RMON2 host or matrix data contains rows of data for many protocols TRENDstep can combine these rows into a few columns in a single summary row Assign grades to a range of input data values The Base Table 9 11 9 Data Aggregation and Manipulation Compute metrics such as utilization and store them as new columns in the output table The Divide By Zero Issue When you compute statistics potential divide by zero situations can frequently arise For example input utilization of an interface is typically computed as inoctets 8 delta time ifspeed 100 That is inoctets bytes is multiplied by 8 to give a bit count which is divided by delta time to give bits per second That value is divided by the interface speed ifspeed which is estimated bandwidth for an interface in bits per second The result is multiplied by 100 to give a percentage If the interface is down for some reason when the data sample is collected ifspeed is reported as zero which leads to a divide by zero condition if utilization is computed for the interface TRENDstep provides a set of statements if
105. the box by clicking on the down arrow in the box then using the scroll bar in the box to move through the list If you click on the name of a statistic in the box its data type will be displayed in the Data Type box Assigning Aliases and Creating Expressions The names of statistics shown in the Select Statistics window are used to label their associated data in reports You can override these names with more readable and de scriptive labels by assigning aliases to the statistics This is done in the Assign Sta tistic Aliases window To bring up the window click on the Assign Aliases button in the main TRENDbuild window 10 10 Creating a New Report TREND 3 6 User s Guide Assign Statistic Aliases request _ts gt Request TS Figure 10 5 Assign Statistic Aliases Window La i E m z o ley Q The Active Statistics box in the window displays the currently active statistics as set up in the Select Statistics window You can assign an alias to any of these names Create an alias by typing it in the New Alias box The alias will appear in reports exactly the way you type it here When it is the way you want it make it available for assignment to a statistic by clicking the Add button The alias string will appear in the Select Alias box You can modify an alias string by clicking on it in the Select Alias box which will cause the string to appear in the New Alias box Make your changes then click on the U
106. the collection request only if the hostname field in the mw collection table record for this collection request matches the hostname of the machine on which mw collect is running If you omit this option mw collect executes all polling requests whose interval matches the value of the i option regardless of the hostname specified to do the polling in the polling instructions Sets the number of seconds the procedure needs to wait in order to acquire a lock on an upload table The default value is 10 which is 10 seconds This option is in UPPERCASE This option is equivalent to the retry_interval parameter in TRENDpm Number of seconds mw collect is to wait for a response after sending an SNMP request Default is 1 second SNMP timeout gt D v o Q o o mw_collect A 20 P P q S The number of SNMP variables to include in the varbind list in the GET PDU Protocol Data Unit request It is possible to generate a GET request that yields a response that is too long to transmit The number of variables needed to exceed the maximum response packet size depends mainly on whether the request is being sent over a WAN link which shortens the allowable packet size compared with a LAN link The default p value is 20 For wide area serial lines you may need to use a smaller value If so begin with a value of 15 If that is successful increase it by 1 until you begin to receive error messages that PDU
107. the debug output level Valid values are 1 2 and 3 The higher the number the more detailed the information Debug output is writ ten to the standard output destination This option is for development purposes A 52 TRENDstep f Identifies the input script file This file contains the commands that generate the TRENDstep output file Unless the path is specified as part of the file name this file is assumed to be in SDPIPE HOME scripts directory l TRENDstep checks the dbstats tables to determine the current size of the database log and does not run if the log used size exceeds the per centage specified in this parameter The default is 90 for 90 m TRENDstep checks the dbstats tables to determine the current size of the database and does not run if the database used size exceeds the percentage specified in this parameter The default is 90 for 90 u Displays the syntax of the TRENDstep command V Displays the version of the command Description In a typical TREND data table each row provides many different statistics for a single table key A report based on this row can show any number of statistics for the table key but only for that particular table key Without TRENDstep if you want to compare the same statistic for many different table keys you have to generate multiple reports gt D v o Q o o TRENDstep however allows you to create a new output data table with the statistics you want
108. this reason the Community String field shown below is an important one Community String public In the Discover GUI you can only specify one community string at a time The dis covery process can only identify SNMP manageable nodes that have this community string Note When IP Discovery is run from the command line it has an option that allows you to discover for multiple community strings See Running Dis cover from the Command Line on page 5 11 To change a Community String select the character string that appears in the Com munity String text field by default Public and overwrite it Using the Discover Graphical User Interface TREND 3 6 User s Guide Host Name Translations c J o 7 fo lt D mg Host names are typically English language names used to identify devices If you select the Do Hostname Translations option shown below Discover will attempt to find the host name for the IP address of each node it finds 7 Do Hostname Translations Different systems use different protocols to perform this translation Discover will try to use whichever protocol is active on your system If the host name is discov ered Discover uses it to identify the device If no host name can be found for the target device Discover uses the IP address Running Discover from the Command Line Although there is a GUI for running IP Discovery this process can also be run from the command line SN
109. to poll all similar nodes for common data Select this option to have TREND poll all nodes of the selected type for this collection request If you select this mode the Select View box is grayed out and the Node lines in the window are ignored Select the desired type by clicking on the Type box s down arrow then clicking on the desired type from the displayed list The value you select will appear on the Select Type line Setting Up New Polling Requests TREND 3 6 User s Guide all nodes in view The nodes in your network can be grouped into views A view may be based on a common location function or other criteria that indicate a set of devices that you want to share a common polling policy Select this option to have TREND poll all nodes in the view you select for this collection request The user must belong to a group to use the following modes If you select this mode the Select Type box is grayed out and the Select Node and Select View lines in the window are ignored Select the desired view by clicking on the View box s down arrow then clicking on the desired view from the displayed list The value you select will appear on the Select View line all nodes of same This mode lets you focus a target list even further by type in view causing only nodes of the specified type which are also members of the specified view to be polled for the selected MIB table If you select this mode you must next select one valu
110. to populate the column in the key table alias is the name assigned to the column in the key table and column is the name of the column in the source data table TRENDlabel searches the key table for the column named by alias If the column named alias exists and if its value is null or blanks TRENDla bel populates the column with data from the column named column in the source data table If the column exists and its value is nonnull or nonblank the existing value in the target key table row is not changed If the column named alias does not exist in the key table TRENDlabel creates it and then populates it in the manner described above Synopsis TREND 3 6 User s Guide Options e h 0 You can repeat this option multiple times in the TRENDlabel statement to identify different key and data table columns or to concatenate column substrings or columns in the source data table to populate the key table column If you omit alias from the command the key table column is assigned the same name as the column in the source data table See Usage Notes on page 19 4 and Examples on page 19 8 for a complete description of this option with examples Sets the debug output level Valid values are 1 2 and 3 The higher the number the more detailed the information Debug output is written to the standard output destination This option is for development purposes Identifies the column in the source data table that
111. variables database vendor A Typographic rules for raw table device B Syntax rules table C Validate Object Types repeat steps A B until MIB loads Figure 6 1 MIBwalker Process Flow Main Display and Menu System MIBwalker is accessed from the main TREND window To launch MIB walker do the following 1 Select Tools from the Menu Bar 2 Select MIBwalker from the drop down menu 6 2 MIBwalker Process Flow TREND 3 6 User s Guide Figure 6 2 shows MIBwalker s MIB map in a tree diagram with the following items Node powder MIBwalker C 1TREND 1mibsAmib Il mib Figure 6 2 MIBwalker Main Window 1 The header bar identifies the active node powder and the complete pathname of the loaded MIB file C TREND mibs mib II mib The MIB path shows the path through the MIB tree to the item selected in the MIB map MIB Group names appear in bold The currently selected managed object is highlighted which is ifType in this example The definition for the currently selected object ifType contains many fields See the descriptions of the eligible fields below The enumerated list pull down menu contains the set of valid values for the selected object ifType This list only appears if it exists for the selected object Main Display and Menu System 6 3 o gt w 2 x D 6 MIBwalker The left side of the MIBwalker window displays information item 5 that defines th
112. vr floor 50 cnt 1 where vr is an array of vals sorted into ascending order vr floor 90 cnt 1 where vr is an array of vals sorted into ascending order vr floor 95 cnt 1 where vr is an array of vals sorted into ascending order vr floor 98 cnt 1 where vr is an array of vals sorted into ascending order cnt E At At EAt E At where At ta period column value of this row minus the ta period column value of the first row in the group cnt X val At tot At d X At cnt TRENDsum ybar tot cnt a ybar b xbar 30 a b Atg 86400 30 where Atg is the ta_period column value of the last row in the group minus the ta_period column value of the first row in the group f60 a b Atg 86400 60 where Atg is the ta_period column value of the last row in the group minus the ta_period column value of the first row in the group 90 a b Atg 86400 90 where Atg is the ta_period column value of the last row in the group minus the ta_period column value of the first row in the group dtt ceil threshold_value a b Atg 86400 where threshold value is the user specified threshold value and Atg is the ta period column value of the last row in the group minus the ta period column value of the first row in the group TRENDsum limits dtt to or 1000 Note In all formulas where it appears val is the resolved value of the expres sion to whic
113. when you compute a total For example expressions such as the following in 3 4 x TRENDsum files column ifinoctets018 avg max tot need to be changed as follows in 3 5 1 TRENDsum files column ifinoctets ifinoctets018 delta time004 avg max column ifinoctets018 tot A 89 gt D v o Q o o trendtimer trendtimer A 90 Daemon that invokes other TREND processes trendtimer s schedule file Description trendtimer invokes other TREND applications which need to be run at specific times or regular intervals It must be run by the user trendadm trendtimer passes trendadm s environment to each process it starts including the executable s search path and environment variable values to insure that the process runs correctly trendtimer runs on a 5 minute cycle Every hour on the hour and at 5 minute inter vals in between it examines the contents of its schedule file and determines which processes to run at the current time These processes are run one at a time in the or der in which they appear in the schedule file Once they are finished trendtimer waits for the start of the next 5 minute cycle trendtimer can be run at any time but it should be invoked at system startup The TREND installation adds a command to the system startup file to run etc rc TREND timer up which sets up the TREND environment for trendadm su s to SDPIPE HOME tmp and invokes trendtimer In the su command in the
114. window open and launch the reports from it Refreshing the Report List 14 20 If you think the contents of the current report directory has changed for example another user has added or deleted report definitions from the report directories you are working with you can easily refresh the display in the Reports list box To do so select Refresh Reports from the File menu see Figure 14 3 Generating the Report TREND 15 Report Scheduler Report Scheduler allows the user to print a report or a series of related reports on a periodic basis Report Scheduler s automatic printing function processes selected re ports to be printed at specified printers or to specified files at desired times on daily weekly or monthly intervals In most networks there is a set of devices e g core routers that need to be moni tored on a regular basis Using Report Scheduler the network staff can receive a se ries of related reports providing a complete picture of a given network anomaly e g router port utilization threshold exceeded E oa D o 2 n 1 7 a o mg Printing a Report Selecting Report Scheduler from the Main window invokes the main Report Sched uler window 15 1 15 Report Scheduler EF Report Scheduler 0 x File Scheduled Reports Daily s Delete default Print File 4 Report Directory Drilldown Report Directory Printer Print F a T Repor
115. window without changing servers 13 14 Selecting Different Data TREND 3 6 User s Guide Setting the Time Period to be Displayed Grade of Service allows you to set a Start Time and an End Time for data to be dis played in reports If you select Time Period from the View Menu the Set Time Pe riod window is displayed o o Q o e n 7 om Set Time Period Figure 13 8 Set Time Period Window Change the Start and End times by selecting the button that corresponds to the por tion of the time you wish to set i e Hour Day or Week Next click either the or buttons to increase or decrease this value respectively Click the OK button to have the sort order you specified take effect Click the Reset button to return to the default start and stop times Click the Cancel button to quit out of the window without changing the start and stop times Selecting Different Data 13 15 13 Grade of Service Setting Report Display Parameters To complete the definition of the new report you must set its display parameters This includes defining the plot line types setting the X and Y Axis parameters and specifying report titles Changing GOS Bar Attributes Color and fill pattern can be specified for each bar segment plotted on a GOS graph Select the color for the current statistic by clicking on the Color box s down arrow button then clicking on the desired color Use t
116. wish to update from the Report Instances box by clicking on the down arrow and then selecting the desired instance Next click the Parameters button to display the Report Parameters window see Setting Report Parameters on page 15 6 When you have the Report Parameters set the way you desire you can click the Update button in the Print window This will change the fields in the table display Defining Scheduled Reports 15 5 15 Report Scheduler for the instance selected in the Report Instances box and update them with the new values you have chosen To delete an instance simply highlight it in the table display select the instance you wish to delete from the Report Instances box and click the Delete button Note You must have at least one Report Instance for a report listed in the table display area Setting Report Parameters If you clicked on the Parameters button in the Print window the Report Parameters window is displayed CT a Node dirty devil z Sort Order Cancel Figure 15 2 Report Parameters Window This window allows you to select different nodes and different keys associated with a particular node Furthermore you can change the Sort Order to see reports sorted 15 6 Defining Report Instances TREND 3 6 User s Guide on a different field or in ascending rather than descending order see TRENDsheet on page 11 1 for more details on nodes keys and sorting Displa
117. you to the Server Entry Editor dialog 7 Click OK This returns you to the Directory Service Session dialog 8 Click the Close Session button This returns you to the Select a Directory Ser vice dialog 9 Click Exit 10 Click the Yes button when your are prompted about exiting dsedit This returns you to the UNIX command line You can validate the result by viewing the SSYBASE interfaces file D 14 Creating the Sybase Interfaces File on UNIX Systems TREND 3 6 User s Guide Creating the Sybase Interfaces File on Windows Systems Use the Dsedit utility to install a secondary server interfaces file entry for Sybase 11 5 1 on Windows systems Perform these steps 1 As user trendadm select Programs gt Sybase gt Dsedit from the Start menu The Dsedit window appears with the Select Directory Service window displayed within it Select Directory Service InterfacesDriver ie i o lt Dm o m 3 Z E 3 iy ko 3 apjayes e dn bumas a Creating the Sybase Interfaces File on Windows Systems D 15 D Setting Up a Satellite Server Environment 2 Select InterfacesDriver and click OK The DSEDITI InterfacesDriver win dow appears EB dsedit DSEDITI InterfacesDriver Server Entry Version SQL Server 4 Unknown Server Address NLMSNMP tpipeisybaselquery NLWNSCK suwanee 5000 3 From the Server Object pull down menu select Add The Input Server Name window appea
118. you use the t and r options TRENDcopy copies the rolled up version s of the specified table in addition to the table itself Here the r option increases the number of tables copied For example assume you want to copy a raw table named rmon_host and its hourly and daily versions from PRIMARY_SYBASE to BACKUP_SYBASE The command is trendcopy t rmon host s PRIMARY SYBASE S BACKUP_SYBASE r HD Note When you use the r option with the t option do not use the r value as it is seldom possible to determine the raw table name from the rolled up table name Using the c and C Options To Filter Data by Column To specify filtering on column values use the c option for data table columns and the C option for key table options Only one column filter can be specified per com mand line for each data and key table The format for column options is trendcopy c column name exp value where column name Is the name given in the Full Column Name column in the Data Manager Column Info window for the table being copied See Data Manager on page 8 1 for details 18 10 Usage Notes TREND 3 6 User s Guide exp Is one of the following operators equal not equal lt less than or equal gt greater than or equal lt less than gt greater than For example assume you want to copy table routerl You want only keys corre sponding to table key 1 ALL and only rows where column InPackets is greater than 3
119. zi Pata En Figure 8 4 Show by Type Window 3 Click the desired value in the popup window 4 Click OK to have your choice take effect Click Cancel to leave the contents of the Data Manager window unchanged Changing the Sort Order of the Data Manager Window By default Data Manager sorts the rows in the Data Manager window area by group ing a raw data table and its associated rollup tables together You can resort the rows by following these steps 1 Click the View button in the Menu Bar 2 Select Sort from the View menu This popup window will show you the fields which can be used to sort the rows 8 8 Showing Particular Tables TREND 3 6 User s Guide Sort Group Rollup With Data By Size in KB By Number of Rows By Object Name By SQL Name Figure 8 5 Sort Window 3 Select the field on which you want to sort the rows 4 Click OK to have the new sort order take effect Click Cancel to leave the sort order unchanged Monitoring the Rate of Database Growth The database occupies a fixed amount of pre allocated space on the disk Although TRENDsystem provides a number of features for controlling the growth of the data occupying this space it is important to monitor how quickly it is being used up and what is using it up so the growth can be controlled and accommodated The first place to watch for database space use is the db used in the Database Summary area Besides keeping track of what percent
120. 0 of the largest table in a mature database See the TREND Installation Guide for the procedure to follow to enlarge tempdb Once you have installed TREND you need to enlarge tempdb if you see any of the following messages either in your error log or on your screen 97 02 05 10 18 11 97 server can t allocate space for object syslogs in database tempdb because the logsegment segment is full If you ran out of space in syslogs dump the transaction log Otherwise use ALTER DATABASE or sp extendsegment to increase the size of the segment The example below demonstrates the process of increasing the size of the dpipe db database by 10046 Because a transaction log is usually 2046 the size of its associated database the example also shows dpipe db s associated transaction log being increased by 20 MB You can apply the following steps to enlarge any Sybase database If you are enlarging tempdb substitute tempdb for dpipe db Enlarging the Database TREND 3 6 User s Guide 1 Log onto Sybase and proceed to the master database user su sybase sybase isql Usa P 1 gt use master 2 gt go Note If you are enlarging tempdb skip step 2 and step 3 and continue with step 4 2 Determine the options that are configured on the database 15 sp helpdb dpipe_db 25 go 3 Inthe output from this command look for the Name and Status columns In the Name column you expect to see the value dpipe_db
121. 00 Use the following command trendcopy t routerl s PRIMARY_SYBASE S BACKUP_SYBASE c InPackets gt 300 C table key 1 ALL a JI m 2 o t ke lt Using the e f and g Options To Filter Data by Date You can use TRENDcopy to copy data filtered by date The following examples ex plain the use of the e f and g options for filtering by date day of week and hour of day You can specify more than one filter each filter can be a single value or a range of values Example 1 The following command copies any data for which the ta_period is 1 00 p m on ei ther Monday or Friday trendcopy f mo f fr g 13 s xyz S XYZ Usage Notes 18 11 18 TRENDcopy Example 2 The following command copies data with ta_period greater than or equal to Monday and less than or equal to Wednesday and ta_period greater than or equal to October 1 1998 and less than or equal to November 1 1998 trendcopy f mo we e 19981001 19981101 s xyz S XYZ Note The weekday range used with the f option is Sunday through Saturday and does not wrap around to the following week To specify a day range that wraps around to the following week two command line options are required e g trendcopy f we sa f su mo s xyz S XYZ will copy data from a Wednesday through the following Monday Example 3 The following command copies data with ta period greater than or equal to 1 a m and less than or e
122. 1 4 11 9 Select Statistic box 11 6 12 9 Select Statistics box 10 13 server source 18 7 target 18 7 Service Level Agreement 1 11 Set Default Storage Time option 8 5 Set Table option 6 35 Set Time button 14 19 shorten aging periods 8 10 Show by Source option 8 5 Show by Type option 8 5 Simple Network Management Protocol SNMP 9 1 Size KB 8 3 Rows 8 3 size using Data Manager to monitor database size 8 1 SLA 1 11 Small Fonts 3 2 SMI 6 11 SNMP 5 5 5 10 9 1 16 1 16 3 17 5 A 14 C 19 E 3 TREND 3 6 User s Guide collection policy 7 4 counter data 9 5 9 16 device RMON manageable 5 4 GET 5 10 5 11 5 12 5 13 5 14 5 15 5 19 GET message 5 3 manageable 5 10 5 11 metrics 13 1 MIB 7 11 10 7 MIB file 10 6 object names 6 9 polling application A 22 A 31 port number 5 12 protocol 6 1 request 5 16 retries 5 12 5 18 table structure 16 3 timeout 5 12 5 18 Tool 6 29 6 33 Tool option 6 31 Change Node option 6 29 New Key 6 36 Set Table 6 35 Set Variable 6 34 Type Discovery 5 11 5 14 5 17 5 18 type discovery 5 2 5 4 5 5 5 14 5 19 SNMP counter data 9 5 SNMP manageable 5 2 SNMP Tool option 6 7 sort 8 8 17 8 Sort option Data Manager 8 5 Sort Order 15 6 15 7 sort order TRENDsheet 11 14 Sort Order option 7 11 11 14 TRENDsheet 11 3 source node 12 3 13 3 source server 18 7 Specify Report Directory option 3 7 Specify Report directory option 2 11 Specify Report Directory Window 2 11 SQL 20 2 N
123. 1 40 00 PM 134 70 18 241 _wildcard ether2 ip 1 1 426 9 769 454 Nov 15 1997 1 40 00 PM 134 70 18 241 _wildcard ether2 ip 1 udp 1 426 5 884 674 Nov 15 1997 1 40 00 PM 134 70 18 241 _wildcard ether2 ip 1 udp domain 1 426 151 021 Nov 15 1997 1 40 00 PM 134 70 18 241 _wildcard ether2 ip 1 udpbootpc 1 426 0 Nov 15 1997 1 40 00 PM 134 70 18 241 _wildcard ether2 ip 1 udp tftp 1 426 0 Nov 15 1997 1 40 00 PM 134 70 18 241 _wildcard ether2 ip 1 udp who 1 426 1 616 Nov 15 1997 1 40 00 PM 134 70 18 241 _wildcard ether2 ip 1 udp ip xns rip 1 426 19 730 Nov 15 1997 1 40 00 PM 134 70 18 241 _wildcard ether2 ip 1 udp ccmail 1 426 0 Nov 15 1997 1 40 00 PM 134 70 18 241 _wildcard ether2 ip 1 udp nbt name 1 426 8 508 Nov 15 1997 1 40 00 PM 134 70 18 241 _wildcard ether2 ip 1 udp nbt data 1 426 3 528 Nov 15 1997 1 40 00 PM 134 70 18 241 _wildcard ether2 ip 1 udp nbt data smb 1 426 3 528 Nov 15 1997 1 40 00 PM 134 70 18 241 _wildcard ether2 ip 1 udp notes 1 426 0 Nov 15 1997 1 40 00 PM 134 70 18 241 _wildcard ether2 ip 1 udp snmp 1 426 5 264 118 Nov 15 1997 1 40 00 PM 134 70 18 241 _wildcard ether2 ip 1 udp snmptrap 1 426 0 Nov 15 1997 1 40 00 PM 134 70 18 241 _wildcard ether2 ip 1 udp sunrpc 1 426 127 224 A 54 Figure A 1 Example RMON2 Data Table If you want to know how many octets are recorded for a particular protocol this table suits your purpose Each row represents a tracked protocol Using TRENDsheet or TRENDgraph
124. 10 n DPIPE HOME bin mw collect i 15 n trendtimer 20 DPIPE HOME bin mw collect i 20 n 60 DPIPE_HOME bin mw_collect i 60 n 24 00 DPIPE HOME bin mw collect i 1440 n 24 00 1 00 DPIPE HOME bin trendit 24 00 3 00 15 DPIPE HOME bin keystats 24 00 4 00 DPIPE HOME bin db delete data 24 00 7 00 DPIPE HOMEJ bin lpr launch i 1 MO 7 00 DPIPE HOME bin lpr launch i 7 MONTH247 00 DPIPE HOME bin lrp launch i 31 eof In this example mw collect is run every 5 10 15 20 and 60 minutes and every night at mid night for data collection which is to be run once a day 1440 minutes TRENDit is run every day at lam keystats is run every day at 15 minutes after 3 a m db delete data is run every day at 4 a m pr launch is run daily at 7 a m weekly at 7 a m Monday morning and on the second day of every month at 7 a m 3 gt Options z D 3 gt S The path name of the schedule file The default is to read e DPIPE HOME lib trendtimer sched o Examples To start trendtimer using the schedule file tmp mytimer sched user trendadm exe cutes the command trendtimer s tmp mytimer sched A 93 A Man Pages TREND TREND B Database Administration The TREND Database and the Sybase SQL Server require special knowledge for proper set up and maintenance This appendix discusses some of the special issues associated with database administration in
125. 16 Setting X Axis Parameters 00 0000 cee eee 12 16 Setting Y Axis Parameters llle 12 17 Specifying Report Titles llle 12 19 Addo Gnd Hnes srair kuare ha DA REP bee DU 12 20 Drilling Down for More Detailed Data 12 21 Command Line Options irakien eee 12 22 Grade of Service solu GAN NADA NEL ex ee cee 13 1 Main Display and Menu System 00200 eee 13 2 Command Menu Summary 0000 eee eee 13 2 The Grade of Service Concept 13 5 Opening an Existing GOS Report 0 00 eee eee 13 6 Creating a New Grade of Service Report 13 7 Selecting Statistics eee 13 7 Selecting Data 2 eee 13 9 Selecting Different Data 00 llle 13 13 Displaying Data for a Different Node 13 13 TREND 3 6 User s Guide Displaying Data from a Different Database 13 14 Setting the Time Period to be Displayed 13 15 Setting Report Display Parameters iilii else 13 16 Changing GOS Bar Attributes 0 020 13 16 Setting X Axis Parameters llle 13 16 Setting Y Axis Parameters 2 0 0 0 cee eee 13 18 Specifying Report Titles cece eee ee eee 13 19 Adding Grid Lines iilii 13 19 Drilling Down for More Detailed Data llle 13 20 Command Line Options aaaea 13 21 Report Launcher leeseeeeeeeee eee eee
126. 17 TREND 3 6 User s Guide report application 7 16 report file directories add or remove 14 10 Report Instances box 15 5 Report Launcher 14 1 controls 14 3 Report Parameters Window 15 6 Report Scheduler 1 5 1 12 12 6 15 1 A 12 Add Drilldown Report option 15 3 Add New Report option 15 3 15 4 Print Now option 15 3 report instances 15 4 scheduled reports 15 5 setting report parameters 15 6 report time period 14 4 14 5 report title 11 8 12 19 report type 14 5 14 11 default 14 12 reporting 1 10 drill down 1 11 exception 1 11 GOS 1 11 1 12 interval 16 5 ReportPack 1 13 5 4 5 17 9 3 9 10 9 16 installing 4 1 reports automatic printing A 12 modifying existing TRENDgraph reports 12 5 scheduling with Report Scheduler 15 1 Reports list box restrict contents 14 4 reset 9 8 resort rows 8 8 retain original scale 12 19 RMON 5 5 5 16 7 15 9 1 A 23 A 31 C 19 agent 16 5 16 6 agents 16 1 collection policy 7 4 data 7 4 16 1 data polling 16 2 data report 17 4 data table 16 9 database tables 7 4 extensions 17 2 history group 16 3 history group polling 16 5 host group 16 3 host tables 16 7 matrix group 16 3 matrix tables 16 7 metrics 13 1 MIB 16 1 16 2 16 3 17 1 17 4 polling agent 16 1 16 2 polling frequency 16 4 polling process 16 6 probes 16 1 16 6 A 30 protocols 17 7 statistics group 16 4 table 16 3 16 9 17 2 viewing RMON tables in Data Manager 16 8 RMON support 16 1 RMON collect 17 6 rmon
127. 2 Valid values are 1 Rejects samples if a spike occurs 0 Does not reject samples The default is O Sets the debug output level for the TRENDpm process of mw collect Values of 0 1 2 or 3 are valid The higher the number the more detailed the information The default is O which means no debug output Debug output is written to standard out You should only use this option for testing in coordi nation with DeskTalk Systems Technical Support due to the additional overhead it places on mw collect Sets the minimum filter value The procedure rejects the sample if the delta value of a counter falls below this value The default value is 1 which means to accept the entire sample Sets the number of seconds the procedure needs to wait in order to acquire a lock on an upload table The default value is 10 which is 10 seconds TRENDpm zerror Sets the percentage level for valid data The value is the percentage of difference between the delta values of two Received Timestamps and two Sys tem Uptimes These statistics come from two con secutive raw data samples For example if rl and s1 are the Received Times tamp and System Uptime for the first sample and r2 and s2 are the Received Timestamp and Sys tem Uptime for the second sample then the calcu lation for the value is r2 r1 s2 s1 100 r2 r1 During processing if the calculated value for the samples exceeds the value set by this o
128. 2 Ag e 3 o 9 Me o Q Q EHE i e 3 To bring up the Specify Report Directory window select the Specify Report Directory option under the Configuration menu Enter the new path name in the Report Directory box and click on the OK button Typically the system administrator establishes a central report directory as read only for most users It is suggested that SDPIPE HOME reports is used as the direc tory with a separate directory setup for each copy of TREND software installed Multiple directories can be established for common report access by different groups Changing the Default Report Directory 2 11 2 TREND Configuration for trendadm Defining Keys Keys are selected using the Edit Descriptions command option from the Configura tion menu This action invokes the Description Editor window Description Editor of xt File View Data Table mib ll ifEntry Select Device dirty devil h Figure 2 7 The Description Editor Window Follow these steps to select a key 1 Choose Select Table from the File menu 2 Click on a data table name in the Select Table window and click OK 3 Select a device from the Select Device list using the downarrow 4 Select Exit from the File menu 2 12 Defining Keys TREND 3 6 User s Guide Setting Application Background Color Users with color monitors can set the background colors for the MIBwalker TRENDsGraph and Grade of Service window
129. 2 16 then delete it Deleting Tables 2 17 2 TREND Configuration for trendadm 2 18 Only the user trendadm can delete tables For all other users this option will be grayed out To delete one or more tables follow these steps 1 Select the tables to be deleted by clicking on their rows in the Table Informa tion area The selected rows will be displayed in reverse video 2 Select Delete table s in the File menu 3 For each table to be deleted this window will be displayed tablemanager Ea Q DELETE Table hour cisco100 Isystem perf sqiname H100 Isystem perf NOT UNDO ABLE Cancel Figure 2 11 The Delete Table Confirmation Window The window shows the user visible name and the internal database name of the table to be deleted 4 Click on the Delete button to delete the table 3 Click on Cancel to keep the table 6 Anew window will be displayed for the next table if more than one was selected Deleting Tables TREND 3 6 User s Guide License Key Administration TREND uses a licensing scheme for copy protection This section describes the license key process including the use of the DSI ELMHOST environment variable to assign a server host and improve performance TREND License Scheme TREND uses a licensing scheme to prevent unauthorized usage A license key con trols all licensing attributes for each application module in TREND A license key is a numeric string that resembles a ra
130. 22 12 22 20 category 10 7 create new report 10 5 10 6 expressions 10 12 groups 10 6 10 8 MIBs 10 6 TRENDcopy 1 6 1 14 18 1 C 12 TRENDodbview 1 6 Auditing Data and Generating Reports 22 20 Building a View 22 9 Define Expressions Window 22 7 Define Qualifications Window 22 8 Elements of TRENDdbview 22 3 How TRENDdbview Works 22 2 Limitations of TRENDdbview 22 3 Modifying an Existing View 22 19 Purpose of TRENDdbview 22 1 Select Columns Window 22 6 Select Data Tables Window 22 5 TRENDdbview Main Window 22 4 trendexport 1 6 TRENDexporter 18 1 21 8 Controls and Switches 20 3 Input Parameter File 20 6 Overall Process and Functionality 20 1 TRENDgraph 1 7 1 10 1 11 1 18 7 4 7 16 7 17 8 6 10 1 12 1 14 12 19 5 22 9 22 12 22 19 22 20 adding changing and deleting plot lines 12 16 adding grid lines 12 20 adding changing report titles 12 19 changing area and bar fills 12 15 changing line type color and width 12 14 changing point type color and width 12 15 command 12 2 drill down 12 21 modifying reports 12 5 selecting data 12 8 selecting keys 12 9 selecting statistics 12 9 setting the time period 12 13 setting X Axis parameters 12 17 setting Y Axis parameters 12 17 TRENDit 1 6 1 7 1 8 1 10 6 21 9 1 9 4 9 5 9 7 9 8 9 15 9 16 10 3 16 7 21 8 A 39 C 12 C 19 TRENDlabel 1 7 19 1 TRENDprint 15 9 TRENDproc 1 7 Creating and Applying 21 7 Possible Applications 21 6 Scheduling 21 10 TRENDrank 1 7 9
131. 25 Create Miblet Pair High low Counters Window 6 26 Pair High Low Counters Window 0 6 27 Counter Pairing Selection 0 2202000050 6 28 View of Applied Pairing 00 eee eee eee 6 28 Window Verifying Creation of Tables 6 28 Connect To Dialog Box pa siecia e nru cee eee 6 29 Community Strings Window 0000 a eee eee 6 30 MIBwalker with Variables of the ifEntry Table Selected 6 31 SNMP Tool Window selee eee 6 31 SNMP Tool with ifEntry Variable Displayed 6 32 Too Many Rows Error Message 6 32 Set Variable Window llle 6 34 Set Tool Window 0 0 0 cee eee 6 35 Collect Data Window llle eee 7 2 Collect Data Periodically Window 04 7 5 Change Database Window 00020200e eee 7 9 Collect Data Window llle 7 10 Order Requests Window aaan anaa 7 12 Data Manager Main Window sas saaaaa anaana aeaa 8 2 An Example of the Column Info Window 8 6 Show by Source Window 8 7 Show by Type Window 2 8 8 Sort Window esp db GNUN a BAAL paba 8 9 Data Aggregation and Manipulation Process 9 3 LAN Connectivity Executive Summary 9 13 LAN Connectivity Capacity Planning Report 9 18 TREND 3 6 User s Guide 9 4 9 5 10 1 10 2 10 3 10 4 10 5 10 6 11 1 11 2 11 3 11
132. 5 4 5 5 5 13 5 14 dsi protocols 17 9 17 12 17 13 17 15 table 17 6 17 7 dsi snmp nodes table 5 15 dsi tab alias 18 7 dsi trendalyzer 18 7 DSQUERY 14 5 14 7 A 12 A 26 E ecapVal qualifier 17 16 edit polling 7 13 type 2 10 Edit Current View option TRENDsheet 11 4 11 9 Edit Descriptions command 2 12 Edit Expressions Window 10 13 ee collect 1 8 1 9 7 15 7 17 element 14 14 description 14 17 name 14 17 End Time 11 12 12 13 13 15 14 19 Enlarging a Sybase Database B 3 env 2 22 environment variable 14 5 14 7 20 5 A 12 A 26 DSI ELMHOST 2 19 Ethernet 16 2 16 8 etherStatsEntry 16 3 exception reporting 11 4 11 12 Exit command 1 19 export 11 20 Export option TRENDsheet 11 3 11 18 EXPRESSION 10 14 expression 10 12 10 13 13 8 22 5 22 7 22 10 22 12 22 18 default 22 7 22 16 editing 10 12 length 10 12 text 22 17 expressions 11 6 12 7 13 8 A 68 GOS 13 8 Index 7 Index Index 8 field community string 5 10 data_db 7 8 DESCRIPTION 6 14 Group Name 6 25 group_name 7 8 Keep Data 2 14 Object Name 6 25 Roll Up Data Automatically 2 14 STATUS 6 12 type labels 6 14 file SYBASE interfaces 7 9 8 5 12 13 13 14 bmp 1 18 dis 5 5 5 14 gif 1 18 etc passwd 2 4 ASCII 11 1 Cshrc 2 22 interfaces 2 4 2 5 postscript 1 18 Profile 2 22 File menu 3 4 fill color 12 16 13 11 pattern 12 15 13 11 13 16 filter 14 15 16 9 18 11 filtering criteria 14 15 14 17 Find Object 8 5 Find option 6 7
133. 6 name 11 14 20 9 target device 17 12 target node 7 15 target server 18 7 tempdb B 4 B 8 template 11 5 The TRENDproc Application 21 2 threshold 9 17 threshold values 13 4 13 18 time categories 9 16 time interval 9 5 Time Period 14 18 time period 14 19 18 5 A 81 length 18 13 Time Period option 13 4 TRENDsheet 11 3 11 12 time range 11 3 12 3 13 4 time scale 12 4 12 16 12 17 13 16 13 17 timestamp 9 4 10 3 11 14 16 5 A 39 ta_period 9 4 virtual 9 4 title 13 19 title lines 12 19 tl_groups table 7 8 tl_users table 7 8 Token Ring 16 2 16 8 Token Ring Statistics A 34 tool Collect Data 6 29 Community Strings option changing 6 30 Discover 5 1 5 6 5 18 MIBwalker 3 5 3 8 Exit command 6 7 TRENDbuild 3 8 Tools menu 3 7 Top Level Report Directory 14 4 14 7 14 8 topN 16 3 group A 33 topology database 2 3 2 5 7 8 18 8 18 13 transaction log 8 4 translation 5 11 transport protocol 17 5 TREND 1 10 administrator 2 1 advanced features 3 1 control panel 3 1 high level description 1 1 TREND 3 6 User s Guide utilities 9 1 trend log file A 11 trend_discover 5 18 syntax 5 12 trend_export 20 2 20 5 trend_proc file 21 2 21 4 Defining 21 7 Syntax 21 9 trendadm 1 14 2 1 2 2 2 13 2 15 2 16 2 18 3 1 3 3 3 5 5 14 7 12 8 1 8 3 8 4 8 5 18 1 22 11 A 12 A 90 F 2 TRENDbuild 1 6 1 7 1 10 1 12 3 8 7 16 10 1 10 2 10 13 10 14 11 5 11 8 11 9 12 5 13 4 13 7 13 9 13 11 20 2 22 9
134. 6 User s Guide In general on slow processors the processor is the major performance bottleneck But as the speed of the processor increases disk and network I O becomes the limiting factor for the overall speed of TREND processing 9 O e 2h 3 O a 1 D el 3 Ce We recommend that you use the fastest available processor for the selected platform so that disk and network I O become the limiting factor Number of Processors Where possible employ a multiprocessor configuration two or more CPUs because TREND and Sybase can be configured to take advantage of multiprocessor environments and run parallel concurrent tasks This decreases the overall run time In most single system configurations two or more slow processors will run faster than a single fast processor For example a dual Pentium 166 MHz running Windows NT provides more throughout than a single Pentium Pro 200MHz or Pentium II 233MHz with proper Sybase and TREND configuration RAM To get the best performance out of Sybase you should obtain as much RAM as possible on each processor regardless of platform Sybase uses RAM mainly for disk cache and procedure cache buffers The more available disk cache the less I O a system does thus performance improves A guideline for selecting the amount of RAM for a given TREND system follows RE Application 256 512 MB Small to medium installations 1 000 4 000 interfaces 512
135. 7 6 note that the View Invalid button is enabled A get request can also return invalid protocols Why this happens and how to deal with it is described below DataPipe Components TREND 3 6 User s Guide Identifying and Renaming Invalid Protocols When you use the Protocol Definition Interface to send a get request it is possible even likely that the request will return at least some invalid protocols When this happens the View Invalid button is enabled You can click this button to see any protocols that have not been recognized and accepted by the Protocol Definition Interface Lj N E O 2 N o l Ko pi This is shown in Figure 17 7 Get Probe Protocol Data orinoco 4554420 8 Unknown Address 4 53 49 0 8 1 IP Address 4e 46530 16 1 0 0 1 0 0 8 0 0 0 0 1 7 0 0 8 1 4 0 0 Figure 17 7 Invalid Protocols Display The Protocol Definition Interface defines a protocol as being invalid for the follow ing reasons The protocol already exists in dsi protocols The device manufacturer has assigned a nonsensical name to the protocol OID The device manufacturer has not assigned any name to the protocol OID DataPipe Components 17 13 17 RMON Support 17 14 The protocol OID cannot be properly identified As explained in Using the Protocol Definition Interface to Populate dsi protocols on page 17 9 it is unnecessary even undesirable to include all available protocols
136. 70 A program to be run every 2 hours will run at midnight 2 a m 4 a m etc offset tells trendtimer to run the command on this line every interval plus lt off set gt hours and minutes so applications can be run at specific times It is specified in the format HH MM where MM must be 00 For example to run a program every night at 2 a m the schedule file entry s time would be 24 00 2 00 Since lt interval gt is figured from midnight Thursday January 1 1970 24 00 means every night at midnight Adding 2 hours to that means every night at 2 a m To run exactly at mid night use 24 00 24 00 Note that 24 00 0 00 is invalid To utilize MM a time interval of lt interval gt lt HH 00 gt MM is used in the schedule file The HH value can be a number of hours and the MM value can be a number of minutes A special case of lt interval gt allows a program to be run on a specific day of the week or month For a specific day of the week A 91 gt D v o Q o o trendtimer Fora specific day of the week use the first two letters of the day of the week English on which the command is to be run SU MO TU WE TH FR SA instead of the HH MM lt interval gt time format In this format an lt offset gt must be used For example for a program to run every Saturday night at midnight i e the midnight between Friday and Saturday the entry is SA 24 00 24 00 is used because the day of the week a
137. 8 2 Tab lar Display 5 NAK ERA fh beeen he AI RS RR REGURET EE 8 3 Command Menu Option Summary lisse 8 4 Pil way tine es BAGY DP DECLINE 8 4 VIEW md REE ieee atte a Sian Sas wate ERE a eat 8 5 TOONS af IKE 8 6 Obtaining Table Profile Information 00200 eee eee eee 8 6 Showing Particular Tables 0 0 00 cece eee es 8 7 Changing the Sort Order of the Data Manager Window 8 8 Monitoring the Rate of Database Growth 8 9 Data Aggregation and Manipulation 9 1 Raw Data Collection liliis 9 3 Data Cleaning iir KGG lan aeter Rei Ge Dx Rn a e her got 9 4 Transforming Counter Data into Delta Data 9 5 Validity Checking 0 cee In 9 8 Hollupz s Sec ERREUR AKN EL RIDE DEN 9 8 The Base Table reti DA ans PUTA DERRATE SY 9 11 Row by Row Conditional Processing 9 11 The Divide By Zero Issue 0 00 eee ee 9 12 Creating an Other Category for RMON2 Data 9 12 Assigning Grades to Ranges of Input Data 9 13 Other Data Aggregation Functions 00a eee eee eee 9 15 Forecasting mc wide denice TEE ND NAA 9 17 Baselines s nana stant tases akg eaten Ga alok GA aes 9 20 RANKING ie iie E IR Pact PAK ach Sa ied ates Rd 9 23 SUMMANY 3 65 6 paa ode mb np Ch Mor hbng ot ne hota Oe Gh rete PANDA 9 26 TREND 3 6 User s Guide TRENDbDuild eee ian ee vU eee
138. A 75 Option esr lm he ae kee A 76 TREND 3 6 User s Guide Descrlptioniu ss oie tr NAUNA Wate DA AA bes ah aoe A 80 TRENDsum Formulas 00000 eee A 80 TRENDsum By Variables lille elles A 81 Null Golumris 2 cero det op EH eme A 82 Rolling Baseline Table llle A 82 Input File Description llle A 83 Examples x eke ee Naba ha pues petenda A 84 Example 2 2a RAD NGANGA Haha KAREN PRA a ee RR A 84 EXAME Z erstes ene dedere KG KOREANA KA e Rn d A 85 Changes in TRENDsum 3 5 1 from 3 4 X A 86 trendtimier s ti sc el Suet ee dats EAE RN SENE ANE A 90 Description a a a E E a a E aaa A 90 The Schedule File 2 2 0 0 aaa A 91 Oir ee ae ee A nd a eed A 93 Examples ka BAS eats oka ank ana hina A 93 Database Administration cece eee eee eee B 1 Estimating the Size of Your TREND Database B 1 Enlarging the Database l l B 3 Starting Sybase from the Command Line B 9 Recovering from Sybase Error 1105 a B 10 Maintaining the Error Log 00 cece ee B 12 Fixing a Suspect Database 0 B 13 Backup Server 0 2 0 0 cee ee enne B 15 Installing a Backup Server for Sybase 11 9 2 Systems B 15 Backing Up the SQL Server Installation 040 B 18 Creating a Dump Device 0 cee eee ee ee B 19 Examples lle ee eee eee ree en ees B 19 Dumping the
139. Address ifAdminStatus ifOperStatus ifLastChange iflnNUcastPkts iflnDiscards F ifinErrors F ifinUnknownProtos ifOutNUcastPkts ifOutDiscards F ifOutErrors Figure 6 14 MIBwalker with Variables of the ifEntry Table Selected If you select a variable by mistake deselect it by clicking the right mouse button again To deselect all the variables click the left mouse button You may select an entire group of variables by clicking the group name that appears in bold However if you select some variables in the group and then select the group you are actually selecting the remainder of variables in the group If you select more than 237 variables for a MIBlet the following message appears mibwalker O Select fewer variables to create the MIBlet 6 22 Creating a MIBlet TREND 3 6 User s Guide 2 From the File pull down menu select Create MIBlet This option is only accessible when there are MIB objects selected for inclusion in a MIBlet Create Miblet Figure 6 15 File Menu With Create MIBlet Selected The Create MIBlet window appears Create Miblet o gt W 2 D Figure 6 16 Create MIBlet Window The name of your MIB file up to the first dot appears in the MIB Name Is field 3 Enter a group name in the Enter Miblet Group Name field This name is appended to the MIB name with an underscore between the names to crea
140. Assume for this example that you see the following configuration options displayed in the Status column select into bulkcopy trunc log on chkpt 4 Determine the device numbers that are already being used Do not use the numbers returned by the following query as values in the vdevno statements that follow 1 gt select distinct low 16777216 from sysdevices order by low 25 go 4 rows affected 5 Determine the device names that are already being used Do not use the values returned in the name column of the following two commands gt e 0 3D a uo Y o a0 om El Enlarging the Database B 5 B Database Administration B 6 1 gt select name from sysdevices 2 gt go name diskdump dpipe_disk dpipe_log master tapedump1 tapedump2 6 rows affected 15 select name from sysdatabases 25 go name dpipe db master model tempdb 4 rows affected Next create space for an extra database device The variables used in the command are name A name not already used by Sybase See the select name from sysdevices statement in step 5 for the names already in use physname Pathname of the new database file The file must not yet exist i e and must not be on an NFS mounted device vdevno An unused device number See the select distinct low statement in step 4 for the devices already in use size num MB to add 4512 If you are enlarging tempdb substitute tempdb2 for dpipe db2 a
141. DA ee IE A 37 Direct PONG cca cain x pe axe ee bet ect xa A 38 IRENDIbs SR rne ama antep Enc Ac en cce A 39 DOSCHPUONs becca cl pee event ftue e teach weed neds A 39 eL AA tee eae A 40 Examples BON edere ee ede mede e e em eee A 42 TBENDJDpI Basa itm odes pn dri Baa coo RR ed tide as PUE RT A 43 GUI eC A 43 pe ParameterS 00 cece ees A 45 TRENDrank corer iedee bes mee ao m4 DALA aes A 48 Oei M A 48 Description os ieee e ene A 48 Definition File Sirs eaa a E AENEA tee A 49 TRENDrank Sub Variables nnana a A 50 2 Clio CE Lc A 51 TRENDStep 2 eee ee m ee eme m e y c m e ey es A 52 Options si bs tese et eL ub A 52 Description csse slew e Re dye Re y UR eee A 53 Example kid echt icu e Spee hcl ied Rie one e eO A 54 TRENDstep Input File llle A 57 Column Definition llle A 63 Using Column Names in Expressions A 63 Mathematical Operators llle A 64 Logical Operators llle A 64 Handling Null Values in Expressions A 65 Final Block sec sull sed e bye ken A 66 Switch Blocks ims ie e espe Ra ee eee A 67 IF Blocks soo ehe Re d e A 68 String Comparison aa a a a a A 69 TRENDstep Output Table Creation A 69 Examples of Input Script Files llle A 69 Example aaa A 70 Example 2 a NAG KUDA LALA ah eee A 73 Example 3 aaa a ue PNG eee KY bh ees A 74 TRENDSUM nme Eit erue oe ee deeds A ee SE E
142. Database nauan ee eee ee B 20 Example of a Backup Script aa B 20 Loading a Database cece eee B 23 Installing a TREND Database On a Raw Device B 24 Shutting Down Sybase 0 2 ae B 24 Performance Tuning Issues cece ee eee eee eee C 1 Hardware Considerations 00 cece eee tee C 2 Processor Type and Clock Speed 22000000 C 2 xvii Contents xviii Number of Processors ees C 3 PAM bd aNG ds eS lad rs ba ate ahead Mo goed bau ONG C 3 Disk Drive Interfaces 0 0 cee eee C 4 Number of Physical Disk Devices 0a C 4 Operating System sees BRS KIARA AS Laat NDN ANG C 5 Sybase Configuration Parameters aana C 6 Total Memory si KAN na kka pi ee eee Po e QA yes C 7 Procedure Cache Percent eee eee ae C 8 Number of User Connections 0 cece eee eee C 9 Cache Pools 000 cee eee eee C 10 Online Engines a l eiss 0 0 0 0 ee C 12 Transaction Log Cache 0 0 0 eee C 12 Transaction Log Size C 14 Number of Locks isses eh C 15 Binding the Key Tables 0 00 cece ee eee C 16 Creating the Named Cache 0000e eee eae C 16 Binding the Individual Key Tables to the Named Cache C 17 Verify the Bindings Were Successful C 18 TREND Processing Options llle C 19 Data Collection llli C 19 TREND
143. Descriptions 2 12 Exit 1 19 Open 1 15 repeated 1 15 Save As 1 17 trend discover 5 12 TRENDgraph 12 2 Index 3 Index Index 4 common polling policy 7 7 community string 5 10 5 12 5 13 5 15 5 16 6 30 17 10 A 27 default 2 6 6 30 field 5 10 multiple 5 10 5 15 private 6 30 public 5 16 6 30 read 2 6 5 10 6 30 RMON 5 16 window 6 30 write 2 6 6 30 Community Strings option 6 8 changing 6 30 comparison operator 11 12 comparison value 11 12 compression routine 18 7 condensed 20 14 comma format 20 16 mode 20 16 Configuration menu 2 5 2 6 2 7 2 9 3 5 Define Users and Groups option 3 6 setting application color 2 13 Specify Report Directory option 3 7 configuring the system 2 1 Connect To Window 6 29 constants 10 3 Constraint 11 13 constraint keywords 20 9 Constraints command 11 4 control panel for TREND 3 1 rollup and aging 8 5 control parameter file 20 9 control table 16 3 16 4 16 8 17 4 A 33 Control Tables 7 4 copy protection 2 19 counter 6 14 9 8 10 3 10 4 10 5 data 9 9 SNMP 9 5 definition 1 10 inoctets 9 5 create rate tables 1 10 Create MIBlet 6 23 6 24 option 6 7 create new report with TRENDbuild 10 5 CSV 20 2 Current types 2 10 views 2 8 current data source 12 12 13 13 current directory 14 3 14 7 Current Hosts 6 29 D Daemon A 90 daily aggregation process 1 10 Data Set Reports 1 13 data aggregation 9 1 incomplete 7 17 collection 1 10 5 1 5 5 7 15 8 2 incompl
144. Dit to run 2 Second each satellite server should perform its delta data copy to the central server only after TRENDit has created the delta tables Therefore you must schedule TRENDcopy to execute after local TRENDit processing is complete Since the time needed for local TRENDit processing can vary be sure to fig ure in a buffer that allows TRENDit processing to complete before you sched ule the TRENDcopy start time 3 Finally schedule TRENDstep TRENDsum and TRENDrank to run on the central server only after the central server has received all delta data from the satellite servers that is after remote TRENDit and TRENDcopy processes are complete When TRENDcopy is Delayed D 6 Only one server invokes TRENDcopy and that server alone is responsible for com pleting TRENDcopy execution The server uses the Sybase interfaces file for com munication information regarding the other server Create the Sybase interfaces file on the server where TRENDcopy is executed See Creating the Sybase Interfaces File on UNIX Systems on page D 8 or Creating the Sybase Interfaces File on Win dows Systems on page D 15 as appropriate for instructions on creating this file If a problem on the satellite server makes delta data unavailable for copying there are no automated retries In this case the copy process is attempted on the next day at the scheduled time or you can invoke it manually If no delta data is copied to the central se
145. Document Overview llle XXV Related Documents 0000 ccc eee eee eee XXIX TREND Manuals lille XXX Sybase Manuals 0 cece eee XXX Standards Documents 00 e eee eee eee XXX Additional Publications 0000 cece xxxi Online User Manual Availability llle xxxi Introduction to TREND 4 1 1 TREND Overview 0 02 eee ee ee ES 1 2 A Brief Tour of TREND 0222 0c e eee ees 1 9 Collecting Data 1 ae 1 9 FROPOMING PT rc 1 10 Graphical Reports llli 1 11 Tabular Reports 0 0 00 eee 1 11 Grade of Service GOS Reports 22255 1 11 Custom Reports 1 12 iii Contents Report Launcher rer EEren ERE EEEE SEER AARRE RIRE 1 12 Report Ghalnlng 2s e pb tex NABA RR ERE RERERTRESET 1 12 TRENDweb Report Building and Viewing Facilities 113 Managing Data saa aa KALAW NGARAN LEA rea ep eon gd 113 Advanced TREND features 0 0 0 cece ee eee 1 14 Commonly Used Commands 000 e eee eee eee 1 15 The Open Command 000 c ee eee 1 15 S8V6 ola hha AA 1 17 SANG AS EET 1 17 AA AA 117 Brit kaka LA PA DD AR Ec 1 19 About TREND irem ed REX eee e Ca 1 19 TREND Configuration for trendadm a 2 1 Configuring the System va are eiae pran atA ET O eA 2 1 Defining Users and Groups 000 00 eee ees 2 2 The Sybase Interfaces File 0000 eeeeaee 2 4 Defining Nodos iie ctp NG
146. ENDexporten sons Seto tae i 44 IR ee ae 20 1 Overall Process and Functionality 00 0 0 e eee eee 20 1 Controls and Switches 2000 cece eee es 20 3 Input Parameter File llle 20 6 Qutp t Format 22x20 don KG iei en he dna RR RO ACER RR 20 14 Tab Separated Format Not Condensed 20 14 Comma Format Not Condensed 000 elles 20 15 Tab Format Condensed aa aaa anaa 20 16 Comma Format Condensed 0c eee eee 20 16 TREND Pro 7 x ER e eee bee ee eh er eee es 21 1 How TRENDproc Works essi esaeen eee ett 21 1 The TRENDproc Application llle 21 2 trend proc Files 2 0 cece eee e 21 2 xiii Contents Possible Applications of TRENDproc 00 0c eee eee 21 6 Creating and Applying a TRENDproc eee eee 21 7 Defining a TRENDproc rore einar k e eE ee 21 7 Things to AVOId zs sexe new da seat A bella A 21 9 TRENDproc File SYNAR rr aconta RAU EERE eee 21 9 Scheduling TRENDOOC s cew pedrera e aaa e eee 21 10 TRENDdbvieWw a na secs chaos Da ee NG NA a a a eed eee 22 1 Purpose of TRENDdbview 0 000 c eee eee 22 1 How TRENDdbview Works 0 00 0 cee eae 22 2 A Note on Limitations llle 22 3 Elements of TRENDdbview 0 000 eee eee 22 3 TRENDdbview Main Window selle 22 4 Select Data Tables Window 22 5 Select Columns Window 0 0000 e eee eee 22 6 Define Expressions Wi
147. ENDstep Column Type How to Reference output data Names a column in the output data table On the left side of an assignment expression use just the column SQL name otherwise use the ROTATE column_name syntax output key Names a key column in the output key table On the left side of an assignment expression use the KEY column name syntax On the right side of an assignment expression use the RKEY column name syntax Mathematical Operators The following mathematical operators are defined Expressions are evaluated from left to right All mathematical operators are equal use parentheses to ensure proper expression evaluation order Logical Operators The following logical operators are defined Operator Meaning Equal To l Not Equal To lt Less Than gt Greater Than lt Less Than or Equal To A 64 TRENDstep Operator Meaning gt Greater Than or Equal To amp amp Boolean AND Il Boolean OR Handling Null Values in Expressions The following table summarizes the rules for handling null values during arithmetic operations Default Result if Expression Result define IGNORE NULL s Statement is Used NULL ta NULL a NULL aor a NULL NULL a a NULL a or a NULL NULL NULL gt NULL a NULL NULL Ei v o Co o The following table summarizes the rules for handli
148. ET Address Bwildcard ether2 drp 8 1 Unknown Address Unknown Address wildcard ether2 sna th 8 1 0 0 1 0 0 128 213 2 0 0 IPX Address wildcard ether2 ipx 8 1 0 0 1 0 0 129 55 2 0 0 Unknown Address wildcard ether2 ipv amp 8 1 0 0 1 0 0 134 221 2 0 0 Unknown Address wildcard ether2 ns net 8 1 0 0 1 255 255 255 255 2 0 0 wildcard ether2 ip 1 icmp 12 1 0 0 1 0 0 8 0 0 0 0 1 3 0 1 0 wildcard ether2 ip 1 tcp 12 1 0 0 1 0 0 8 0 0 0 0 6 3 0 1 0 wildcard ether2 ip 1 udp 12 1 0 0 1 0 0 8 0 0 0 0 17 3 0 1 0 wildcard ether2 ip 1 ospf 12 1 0 0 1 0 0 8 0 0 0 0 89 3 0 1 0 wildcard ether2 ip 1 ns ip 12 1 0 0 1 0 0 8 0 255 255 255 255 3 0 1 0 IPX Address wildcard ether2 ipx 12 1 0 0 1 0 0 129 55 0 0 4 81 3 0 0 0 x Enter New Protocol Get Probe Protocols ange col Delete Protocol Figure 17 1 RMON Protocols Window The interface window displays all of the RMON protocols defined in dsi protocols The window appears empty the first time you launch the interface You can search and sort this display DataPipe Components 17 7 17 RMON Support To sort 1 Select Sort from the Edit menu The Sort By window appears Sort by Figure 17 2 Sort By Window 2 Select a sort variable and click Sort To search 1 Select Search from the Edit menu The Search window appears Search Current Address Type Current Protocol Name Current Protocol Oid Figure 17 3 Search Window 2 Selec
149. Edit Expressions Select Statistics Active Expressions delta_time gt Delta Time received_usec ifInUnknownProtos f El fl al frr Clear New Expression Add Delete LF Add to View Cancel Figure 10 6 Edit Expressions Window 10 12 Creating a New Report TREND 3 6 User s Guide The Edit Expressions window lists all the statistics in the current group in the Select Statistics box in alphabetical order All the statistics are available instead of just those activated in the Select Statistics window because any statistic may be needed for an expression and because an expression is treated like a selected statistic by TRENDbuild Note Although the Select Statistics box shows you the names of any assigned aliases the statistics actual names are used in expressions Build an Expression i E m z o ley Q Note It is important to follow these steps in order 1 Type in the expression s name in the New Expression box The name you give the expression will be used as its name in the report You cannot assign an alias to an expression however when you name the expression you are effectively giving it an alias 2 When the name is the way you want it click on the Add button The name will appear in the Active Expressions box Note that you cannot change an expres sion s name in the window so make sure you give it the name you want 3 To define the expr
150. GCODyi eer ax tte tu rtt e enr y Riese OUS C 20 TRENDit TRENDsum 0000000 ee C 20 db delete data echo a ek Roce REA era Rn C 22 Intermediate Reporting Tables lille C 22 Setting Up a Satellite Server Environment D 1 Data Management on the Satellite Server D 2 Using TRENDc0py ce eers oraina a RI 3 D 4 Timing TRENDit and TRENDcopy Operations D 5 When TRENDcopy is Delayed 0000 eee D 6 Node Lists and Polling Policy llle D 7 Firewall Considerations liess D 7 Creating the Sybase Interfaces File on UNIX Systems D 8 For Sybase 11 0 X ee eee D 8 For Sybase 11 5 rasuna sev aa eats cite BEA hei See au a Pa eS D 9 For Sybase 11 5 1 iea a a E e tees D 11 Creating the Sybase Interfaces File on Windows Systems D 15 TREND 3 6 User s Guide Discovery Considerations eee eee eee eee eee E 1 Relocating the TREND Database a F 1 ING OX ee ee ee ed Rae ee eho eee Index 1 xix TREND TREND Figures TREND Applications and Processes 1 2 The Select File Window eee 1 15 The Print Dialog Box e eee ee 1 18 The TREND Configuration Menu isses 2 2 The Define Users and Groups Window 2 3 The Define Nodes Window 2 222 2 6 The Define Views Window 0 2 2222 2 8 The Define Types Window
151. KEY dsi_target_name by_variable KEY dsi_data_source This specifies that the output table will contain one row for each target name interface combination for each ta_period in the input file Figure A 8 Example Input Script File with Comments TRENDstep Example 2 this file produces a break down of IP and Other packets in addition it grades the usage if the value of the Other packets column is gt 0 04 of ippkts then the grade bad define TIMEADJUST source_table Rmon2_statsdata_ destination_table rmon2_protocols switch KEY dsi_table_key case ip nt layer ip ippkts oldiststatspkts009 case 96 26 96 has layer above nt layer ignore it skip just physical layer case ip nt layer without IP must be other otherpkts oldiststatspkts009 endswitch The Switch Block below should properly be in a Final Block as it should only be executed once on the resulting rows It is shown as it is for example purposes only switch ROTATE otherpkts ROTATE ippkts 100 0 gt case lt 0 04 z grade GOOD D default grade BAD endswitch 9 by variable target name001 by variable KEY dsi data source A 73 TRENDstep A 74 Example 3 this example does something very similar to the one above but here we use if statements this report will produce separate c
152. MP Type Discovery can only be run from the command line Using the Command Line to Run IP Discovery As described in IP Discovery on page 5 2 IP Discovery pings to discover nodes in the range you specify When it finds an IP address that has a device IP Discovery sends an SNMP Get to determine whether or not that device is SNMP manageable This process can be accomplished from the command line Running Discover from the Command Line 5 11 5 Discover The syntax is trend_discover a lt delete ages default 10 7c community string for all SNMP GET C ping number of packets 1 default d debug level D suppress domain names E echo progress f community names file F type definition file h end IP range H help 1 sstart IP range o SNMP timeout in seconds O ping timeout in ms NT ONLY default 750ms gt P SNMP port number default 161 r SNMP retries gt s network Subnet gt p max entries per pdu default 20 S ping packet size default 1 byte t do type discovery u username v display version z activate host translation Of the options described above only 1 h and s are mandatory trend discover 1 lt start IP range n end IP range s network Subnet gt 5 12 Running Discover from the Command Line TREND 3 6 User s Guide Option Descripti
153. PeHOZ SOMA Identifies the source server Identifies the target server Identifies the tables to be copied Enter the name of the table as shown in the Object Name column of the Data Manager display See Data Man ager on page 8 1 for details Identifies the source topology database Displays a listing of all TRENDcopy options Specifies the user name Provides filtering by view Use of this parameter is optional It can only be used in conjunction with the U parameter Displays the TRENDcopy version TRENDcopy checks the dbstats tables to determine the current size of the database to be copied and does not run if the database used size exceeds the percentage specified in this parameter The default is 90 for 90 TRENDcopy Syntax TREND 3 6 User s Guide Usage Notes Using the s S and t Options To Copy Database Tables The S target server option is required The s source server is optional but we recommend you specify it for documentation purposes The source server specified by s is the server from which data and table definitions will be read Its default value is SDSQUERY The t option specifies which tables within the database will be cop ied If the t option is not used those non system tables listed in dsi_tab_alias will be copied a m z o t aS lt On the target server the entry in dsi_trendalyzer is created with all fields set to de fault If the correspon
154. R e ebur A 11 Ipr la nch s conem eee et ee ee eee ge A 12 Desenin DE A 12 eL A 13 Examples maang Hin nA oe bep Aha A A 13 mw collect see NAA welt ee eae Eh este ae Aha paan A 14 ODtIOFIS 274G pug KNA Gea ML BAE BAGAL NALATE tae Npa A 15 Description eee ee ee e m Re Res A 22 Fie EOCkS oh Ee xe RA ANAND DL QAM A 23 Local Storage of Data 2 2 eee eee A 24 Interval Poling 0 00 c eee ee eee A 26 Distributed Polling llle A 26 Direct Pollingi ERE Res A 27 Log File vis cee ev eme Een e Re Ree A 27 Cache Poler ere rr exce KR et A 28 Directory Structure A 28 Instructions for Loading the Cache Version of the Poller A 28 Examples se REL RE xd eee Eder EE us eed A 29 Interval Polling llle A 29 Direct Polling llle A 29 rmon collect css EL Ld uetke aeta e A 30 Synopsis naaa ae ek em we m eee EY o RR e ae A 30 Fomati aesae grece ee zen t dor A eto EE PRO A A 30 Format2 lues RA e KAN Ee e s A 31 DescHption iuis bedienen A 31 Polling Controls llle A 32 Supported RMON Groups eee A 32 Control Tables 0 00 e eee A 33 The History Table 00 cee ee ees A 33 The Matrix Table llle A 33 Token Ring MAC Layer and Promiscuous Tables A 34 Log File esee mem A 34 XV Contents lei PEE KAG ERER ND titan KNA TT A 34 Examples sod Ain sees e ELS Ge epbeRS PI ADD A 37 Interval Pollifigaszza rb ws RUPEE NA
155. Subnet Mask to be used in the Discover process S Specifies the packet size of a discovery ping The default is 1 byte t Causes Discover to run an SNMP Type discovery u Defines the TREND user name of the user conducting the discovery V Displays the version number of the product Z Specifies that Discover will attempt to translate a discovered IP address to a host name Notes About Some Discover Command Line Options If the u lt username gt option is not used Discover will try to get a user name from the USER environment variable If Discover cannot locate a USER envi ronment variable trendadm is used for the username If you use either the f or F option or both giving a file name without speci fying a file path Discover will look for the specified file in the SDPIPE HOME scripts directory If you use the t option to run SNMP Type Discovery but do not use F to spec ify a dis file Discover will look for dis files in the DPIPE HOME scripts directory If the dis files are missing an error message appears The a delete age option specifies how many consecutive attempts to ping a node must fail before the node is automatically deleted from the dsi nodes and dl type tables For the Start IP range 1 end IP range h and network Subnet s options if the value you input is not complete Discover will append zeros to the end of that value For example if you input the value 134 70 Discover will assume
156. Suspect Database B 13 gt Q 3 7 mp 5 A 3 UJ oa 0 A 0 o Pu 4 D B Database Administration 1 gt commit tran 2 gt go 4 Disallow updates to the system tables 1 gt sp_configure allow updates 0 2 gt go return status 0 1 gt checkpoint 2 gt go 5 Now shutdown Sybase 1 gt shutdown 2 gt go Server SHUTDOWN by request The SQL server is terminating this process CT LIBRARY error ct results Network packet layer internal net library error Net Library operation terminated due to disconnect CI LIBRARY error ct cancel Network packet layer internal net library error Net Library operation terminated due to disconnect 6 Now restart Sybase unix su sybase sybase cd sybase install sybase startserver f RUN_ lt HOSTNAME gt _SYBASE gt amp dev null amp where lt HOSTNAME gt _SYBASE is the name of your Sybase SQL Server or su root sh etc rc TREND DP up Note For NT make sure that the service for Sybase is started See the TREND Installation Guide for details B 14 Fixing a Suspect Database TREND 3 6 User s Guide Backup Server Dumps and loads are performed by the Backup Server program which runs on the same machine as SQL Server You can perform backups over the network using one Backup Server on a remote workstation and another on the local workstation Refer to the SYBASE SQL Server Syste
157. The port number for TREND servers installed on a UNIX system is typ ically 2052 The port number for TREND servers installed on a Win dows NT system is typically 5000 D 18 Creating the Sybase Interfaces File on Windows Systems TREND 3 6 User s Guide 9 Click OK The Network Address Attribute window reappears with your entry Network Address Attribute x ProtocolNetwork Address Up Down W 9 lt ED m 5 E s 3 3 3 anpiayes e dp bumas a 10 Click OK The DSEDIT1 InterfacesDriver window reappears FA dsedit DSEDIT1 InterfacesDriver File Edit Server Object Window Help Server Entry Version 1 Server Name BANYAS_SYBASE Server Service SQL Server Server Status 4 Unknown Ping icon SERGE Server Address TCP banyas 2052 For Help press F1 Creating the Sybase Interfaces File on Windows Systems D 19 D Setting Up a Satellite Server Environment 11 At this point test the connection to the specified server by clicking on the Ping icon the red lightning bolt icon below the menu bar The Ping window appears 12 Click the Ping button A message similar to the following appears in the dsedit window if you have successfully opened a connection to the specified server dsedit x Note If the message you receive indicates that the server is not recognized you have incorrectly specified the host name of the server o
158. The user visible name of a table in the database It is composed of the com bined names of the MIB and its subordinate table or schema and group that is the source of the data in the database table SQL Name The name by which the table is known to the SQL Server and the name which must be used in any SQL queries Type The kind of data stored in this table Possible values are data raw sample data rate inter sample delta values for counters original values for all other fields hourly daily weekly monthly and internal an internal TREND table that cannot be truncated or deleted Rollup Shows and sets whether or not the data in this raw data table will be aggregated into rate hourly daily and weekly summaries Possible values are yes aggre gate no don t aggregate and default aggregation determined by the value set for Roll Up Data Automatically in the Default Storage Time window See Changing System wide Rollup and Aging Defaults on page 2 14 Only the user trendadm can change this value Size Rows The number of rows in the table Size KB The size of the table in kilobytes Keep Data for Shows and sets the number of days the data will be kept in the table Data is discarded when it is aging 1 days old Possible values are forever never delete Days keep the data in this table for days and default determined o E o o 3 0 Q D Main Display and Men
159. Transact SQL kill command Restarting SQL Server If the long running transaction is due to a runaway query and the process with the open transaction has been identified use the kill command to stop the pro cess This clears the transaction and allows the log to be truncated If the kill command cannot stop the process restart SQL Server to resolve the problem Restarting SQL Server causes the database to go through normal recovery so any outstanding transactions are either committed or rolled back gt go 0 3D a uo OX m a0 om 3 Recovering from Sybase Error 1105 B 11 B Database Administration a Enter the following commands to restart Sybase SQL Server unix su sybase sybase cd sybase install sybase startserver f RUN_ lt HOSTNAME gt _SYBASE gt amp dev null amp where lt HOSTNAME gt _SYBASE is the name of your Sybase SQL Server or su root sh etc rc TREND DB up Note For NT make sure that the service for Sybase is started See the TREND Installation Guide for details Maintaining the Error Log Sybase writes messages to the error log each time the system is booted and each time Sybase error is encountered The name of the error log file on UNIX systems is SSYBASE install hostname SYBASE log for Sybase 11 9 2 On NT systems the name of the file is SSYBASE install errorlog To prevent the file from becoming unnecessarily large you should periodically delete all me
160. User s Guide TREND 3 6 Desk Talk TREND 3 6 User s Guide DeskTalk Systems Inc makes no representation or warranties with respect to this manual or except as specifically stated in the applicable user agreement or warranty notice with respect to any hardware firmware or software described in this manual DeskTalk Systems Inc spe cifically disclaims any expressed or implied warranties or merchantability title or fitness for a particular purpose Furthermore DeskTalk Systems Inc reserves the right to make revi sions or changes to any and all parts of the manual hardware firmware or software at any time without obligation to notify any person or entity of the changes DeskTalk DeskTalk Systems the DeskTalk logo and TREND product names are registered trademarks or trademarks of DeskTalk Systems Inc Other brand and product names are reg istered trademarks or trademarks of their respective holders Copyright 1992 2000 DeskTalk Systems Inc All rights reserved No part of this publication may be reproduced photocopied stored in a retrieval system transmitted or translated into any language without the prior written permis sion of DeskTalk Systems Inc TREND 3 6 User s Guide Publication No UG361 2000JUN16 DeskTalk Systems Inc 19191 South Vermont Avenue Ninth Floor Torrance CA 90502 310 630 1000 TREND Contents FIQGUICS Hi eee Tr pU XXi Preface a 0b BER dein DANE AG BEA SR faded Se XXV
161. Using the c and C Options To Filter Data by Column 18 10 Using the e f and g Options To Filter Data by Date 18 11 Example 1 2d d eem ES PP ERE ev ee 18 11 Example Z sep aeui mn Mana DP KALAN PRAY 18 12 Example Shas cesses e e e er ee Re ros 18 12 TREND 3 6 User s Guide Using the v and n Options To Filter by View and by Type 18 13 Using the b and I Options To Copy Data 18 13 Example aasa es Aha ma inane TANDA LA 18 14 Example 2 vade Ka WAR xg ened caw KAN halin bana 18 14 TRENDl bel aa NAT KG ae eee Seen oa PNPA PA NADYA 19 1 SYNOPSIS maan i Rade Mele us NGA BKA pu eoi KAG LUN ed doa 19 2 OPOS ee aa Ka a e ms kl eed BANA Baan AN GNG LO aah t ELEME 19 2 Usage Notesi e oua Dee eX Dx bga alee wide beak ae 19 4 Populating the dsi descr Column lslesselesslssns 19 5 Locating the Target Row in the Source Data Table 19 5 Ensuring Key Table Data Table Compatibility 19 6 Update Restrictions llli 19 6 Extracting Substrings from Column Values Ls 19 6 Concatenating Column Values ee eee 19 7 Example NENA edu pe Ra amag nos p RE tung UNG 19 8 Example lapid pada Na NANANA ene epo dd epa 19 8 Exainple 2 sedie Pret orb teet e wate bd was 19 8 Example 3 4 uateuteteenategubpE ACIE RE Anata RE Ua 19 9 Example 4 sip pee tigre AA pi ets edd tes 19 9 Example Seen Sedes deben daa ei he wedepe daa be taal 19 10 TR
162. XENT_SYBASE trendadm N E m 032 AB no fo 9 Nio o a Q Q EHE at e 3 Figure 2 2 The Define Users and Groups Window Follow these steps to create a new group Enter the name of the new group in the New Group box The name can contain a maximum of eight characters If you enter more than eight characters the entry is truncated to eight characters 2 Next enter the name of the SQL Server which will store the group s node con figuration information in the Topology Database box 3 Enter the name of the SQL Server which will store SNMP data collected through polling setup by a group member in the Data Database box see The Sybase Interfaces File on page 2 4 4 Click on the Add Group button to register the group 5 Click Cancel to quit out of the window The user will do the collecting Follow the steps below to create a new user Defining Users and Groups 2 3 2 TREND Configuration for trendadm 1 Enter the name of a user to be added by typing their UNIX system username in the New User box Entries are case sensitive and must exactly match an entry in the etc passwd file or in the NIS user database The name can contain a maximum of eight characters If you enter more than eight characters the entry is truncated to eight characters 2 Next click on the downarrow of the Current Groups box and select the name of the group of which they will be a member 3 Click on the Add User
163. You can access a different SQL Server by clicking on the Change Database option under the File menu to bring up the Change Database window Change Database PATUXENT SYBASE LOK Cancel Figure 7 3 Change Database Window mg e ol o o D r 5 Note A report may not display the same information for both databases The name of your current database is displayed in the Change to box To change to a different server click on the down arrow in the Change to box to see a list of the Sybase SQL Servers known to this machine The values presented here are read from the SYBASF interfaces file Select the name of the server you want to access in the report Setting Up New Polling Requests 7 9 7 Collect Data Click OK to change to the selected database Click Cancel to quit out of the window without changing servers Managing Polling Requests It is important to keep track of your data collection requests and to be able to modify or delete them The Collect Data window displays a list of all data collection request definitions stored in the database Collect Data G x File View Tools User REST ER En Delete Entry Node View Type User Frequen Mib router trendadm 15Minutes mib ll ifEnti router trendadm 15 Minutes cisco100 Isyste router trendadm 15 Minutes cisco100 ifentry proto router cisco trendadm 15Minutes framerelay dte rf frCirc orinoco rmon tr rg 15 Min
164. _collect 1 5 1 8 1 9 7 15 7 17 16 2 16 6 C 12 C 19 utility A 30 RMON2 9 1 9 2 9 12 16 1 agent 17 6 17 11 agents 17 1 17 9 collector 17 5 data creating an other category for 9 12 DataPipe 17 1 17 2 17 4 17 7 MIB 17 1 17 2 17 4 polling agent dpipe_rmon2 17 6 protocol definition interface 17 6 protocol naming convention 17 12 protocols 17 7 17 9 17 11 Index 15 Index Index 16 support 17 1 Roll Up Data Automatically 8 3 rolled up table 11 10 rolling baseline A 82 rolling baseline period 9 22 rolling baseline aggregation 9 20 rollup 1 7 2 13 8 3 9 1 9 8 16 7 Rollup column 9 10 roll up table 10 4 ROTATE notation 9 14 S samples 16 5 samplings fail 9 8 satellite database servers 1 14 Save As option 1 17 option 1 17 saving collection request definition 7 9 scalability 1 14 score 13 5 13 10 Scoring area 13 10 Scoring box 13 11 scoring level 13 11 SDT 1 13 Search Lists window 14 16 Select Group button 10 7 Key 6 35 select class of table 8 7 existing type 2 10 key 2 12 Select Alias box 10 11 select columns 22 15 Select Data option 12 11 13 3 13 9 Set 12 8 12 10 12 12 12 14 Set window 13 3 13 9 13 10 13 13 13 18 window 13 4 Select Data Set window 7 17 select devices and elements 14 4 Select File window 1 16 Select From box 10 9 Select Group button 13 7 Select Key 13 11 Select Keys button 7 6 Select Mib Window 10 6 Select Node 3 5 Select Node option TRENDsheet 1
165. _key gt appended to its name in addition to the timestamp described above The condense keyword causes trend_export to remove trailing blanks from each col umn included in the SQL select statement If the max keyword is used then the begindate finishdate begintime and finishtime constraint keywords are ignored Also any imbedded start or end date constraints in the original qss file will be ignored If the frequency keyword is used daily weekly or monthly then any other date constraint is ignored That is if in the same report specification block the max keyword and or the begindate and or the finishdate keywords are used then they will be bypassed If you plan on running the same report on a daily weekly or monthly basis then use the frequency keyword for the required time period That way no modification to the control parameter file will be needed By specifying the daily weekly or monthly frequency value the following actions are done Input Parameter File 20 9 N j D m Z o o bad o s D 20 TRENDexporter 20 10 for daily gt the previous day s data is queried for For example if the cur rent date is 6 1 97 then the report would query for data from 5 31 97 00 00 to 5 31 97 23 59 No matter when the report is run the previous day s data is queried for for weekly gt the previous week s data is queried for For example if today was Wednesday May 28th 1997 then
166. a period value are Nov the value JonesNov is used as the dsi descr column value in Ksi dbstats if the existing value is null or blanks If the existing dsi_descr value is nonnull or nonblank the value is not changed The target source row is identified the same way as described in example 1 Example 4 The command trend label k Ksi dbstats t Rsi dbstats o 4 r dsi descr user name r dsi descr ta period 1 3 populates the dsi descr column in the Ksi dbstats key table with a concatenation of the value for the user name column and the first three characters of the ta period value in the target source table row the same ways as described for example 3 with the exception that the existing value of dsi descr is replaced by the concatenation even if that value is nonnull or nonblank The r option used in example 4 causes unconditional replacement the e option used in example 3 replaces only if the existing value in the target key table record is null or blanks Ce J m z o ley 82 Furthermore the target record in the source data table is the dsi key id record where the ta period value is 04 00 00AM on the previous day Examples 19 9 19 TRENDlabel Example 5 The command trend label k Ksi_dbstats_ t Rsi dbstats o 4 e user name e ta period 1 3 r dsi eurodate ta period 5 3 r dsi eurodate ta period 1 4 r dsi eurodate ta period 8 4 takes data from Rsi dbstats where ta period for the dsi
167. a ta period value of today s date at 10 00 00 AM is used If there is no record for 10 00 00 AM TRENDlabel searches for a record for 09 00 00 AM and so on backward in 1 hour increments until a match is found the maximum ta period value that is equal to or less than the current hour minus 1 hour Example 2 19 8 The TRENDlabel command trend label k Kib ii ifentry t mib ii ifentry e ifdescr010 c speed ifspeed013 c type iftype011 populates the dsi descr column in the Kib ii ifentry key table the same way described in example 1 In addition it creates if the column does not already exist and populates the key table column named speed with the value for ifspeed013 in the target source data table row and the key table column named type with the value for iftype011 in the target source data table row if the key table column values are null or blank If the value in either column is nonnull or nonblank the current value is not replaced The target source row is identified the same way as described in example 1 Examples TREND 3 6 User s Guide Example 3 The command trend label k Ksi_dbstats_ t Rsi_dbstats_ e user name e ta period 1 3 populates the dsi descr column in the Ksi dbstats key table with a concatenation of the value for the user name column and the first three characters of the ta period value in the target source table row Thus if the user name value is Jones and the first three characters of the t
168. a0797 1 hour cisco100 Isystem perf c cisco100 View qw1234567890 month cisco100 Isystem perf d cisco100 Isystem perf system perf e day cisco100 Isystem perf Data Table K100 Isystem perf A a Kcam hlhostdata B Key Tab able gt Data fables N N D m z o ex o ok Cancel Figure 22 2 Select Data Tables Window Complete the Select Data Tables window first Before you can assign qualifications apply expressions or choose data columns for a view you must first define its data and key tables Each of the other definitions is referenced to the tables you select For this reason you must choose your data tables before you can specify columns or assign either expressions or qualifications If you try to access another window before you specify at least one data table the following message appears Elements of TRENDdbview 22 5 22 TRENDdbview Select Columns Window Use the Select Columns window to specify the data columns to appear in your view cisco100_View_Misha0797 1 a a target_name cisco100 Isystem perf b a received ts Kcam hlhostdata A c a request ts K100 Isystem perf B se a delta time 2 a user name a table key a report type Figure 22 3 Select Columns Window Access the window by clicking the Select Columns button in the TRENDdbview main window 22 6 Elements of TRENDdbview TREND 3 6 User s Guide It is not necessary to define which columns from the selected data tables
169. ability to provide min max and avg values is especially significant as applied to reporting on RMON History groups While min and max reveal short term changes over hours avg provides a view of long term variations over days Note TREND will only poll for a single history instance for each data source i e probe network interface or probe network interface coupled with Frontier Domain If multiple Histories are configured TREND will select one on a best fit basis The best fit can be found by multiplying the number of buckets by the bucket width to calculate the total span for History The span selected is the smallest one greater than the polling interval The RMON polling process rmon_collect will analyze each history table instance and put information in the file DPIPE_HOME Polling Frequency TREND 3 6 User s Guide tmp trend log informing you if the polling intervals defined for your his tory tables are appropriate For example if an instance of the history table has 70 buckets representing one minute each that instance spans across an hour period actually 70 minutes There is no reason to poll it more frequently DeskTalk recommends this polling scenario Conversely if the history table has an instance with 50 buckets of 30 seconds each only 25 minutes are represented and polling once an hour will lose data collected in the other 35 minutes Even if only polled once per day rate tables will contain data sample
170. able containing that variable 2 Bring up the SNMP Tool window Figure 6 27 if it is not up already 3 Click the Get button in the SNMP Tool window If you selected the specific variable in the MIB map only that variable appears in the SNMP Tool window In the event there are multiple instances of the variable all instances appear If the variable is writable the Set Variable button is highlighted when the variable value is returned If you selected and got an entire table select the instance of the variable you want to set If the variable is writable the Set Variable button now is highlighted 4 Click the Set Variable button This brings up the Set Variable window Figure 6 30 Note that the SNMP Tool window will be frozen while the Set Variable window is up Set Variable Figure 6 30 Set Variable Window 5 Enter the new value for the variable on the Set Variable line 6 Click the Apply button to set the variable If the set is successful you will see a box with the following message on your screen set successful 7 Click the OK button in the box to continue 8 Click the Cancel button to quit out of the window 6 34 Utilities TREND 3 6 User s Guide Setting Multiple Variables in a Table 1 Select the table name in the MIB map that contains the variable you want to set 2 Bring up the SNMP Tool window if it is not up already 3 Click the Get button in the SNMP Tool window The Set Table bu
171. ach hash mark on the X Axis represents 1 week Click the OK button to activate the new scale Click Cancel to retain the original scale Setting Report Display Parameters 13 17 13 Grade of Service Setting Y Axis Parameters The vertical or Y Axis represents the overall health range based on the statistics plot ted You can assign labels to the threshold values To add or change the Y Axis param eters select the Y Axis option under the View menu to bring up the Y Axis window Axis Label Quality Axis Threshold s 0 5 Threshold 5 00 Label Poor Threshold 4 00 Label Fair Threshold 2 00 Labek Good Threshold 1 00 Label Excellent Threshold Labe tt Threshold tabek Cancel Figure 13 10 Y Axis Window The window displays the current labels assigned to the Y Axis To add a new label click in a blank Threshold box and enter the threshold number corresponding to the points Pts in the Statistics To Graph box of the Select Data Set window which is accessed from the View pull down menu see Figure 13 6 Enter the appropriate threshold number and then enter a corresponding label in the Label box to the right To change a threshold or label highlight an existing entry and type in the new one Click the Apply button to activate the new labels Click Cancel to retain the original labels 13 18 Setting Report Displa
172. acing the first character of data table name with a K Multiple tables can be specified on a single command line If for example you want to copy the tables mib II ifEntry mib II system and rmon history from the server PRIMARY SYBASSE to the server BACKUP SYBASE the following command is used trendcopy t mib II ifEntry t mib II system t rmon history s PRIMARY SYBASE S BACKUP SYBASE Using the T Option To Specify a Topology Database TRENDcopy allows you to override the default topology database when you copy data If you do not specify a topology database TRENDcopy uses the topology da tabase defined for DSQUERY The T option is only used to specify a different to pology database For example assume you want to copy data tables from PRIMARY SYBASE to BACKUP SYBASE PRIMARY SYBASE has the default topology database Chicago you want to apply the topology database Dallas The command is trendcopy s PRIMARY SYBASE T dallas S BACKUP SYBASE 18 8 Usage Notes TREND 3 6 User s Guide Using the U Option To Specify a User Name When you copy data using TRENDcopy you can specify the defaults that apply to a particular TREND user If you do not specify a user TRENDcopy uses the defaults associated with the default group For example assume you want to copy data tables from PRIMARY_SYBASE to BACKUP_SYBASE but you want to apply the topology database that belongs to the user Joe Bloggs If Joe Bloggs
173. ad want to create expressions Assigning Aliases and Creating Expressions on page 10 10 defines the units of measurement for statistics in raw and roll up tables Select specific statistics to include by clicking on their names in the Select From box Selected names will be highlighted You can deselect a name by clicking on it a second time You can select all the statistics at once by clicking on the Select All button You can deselect all the selected statistics at once by clicking on the Clear button Once you have selected the statistics you want include them by clicking on the Add button Their names will appear in the Active Statistics window Creating a New Report 10 9 10 TRENDbuild Note Their names will continue to appear in the Select From box even after they are selected If you want to exclude one or more active statistics click on their names in the Active Statistics window to highlight them then click on the Delete button Their names will be removed from the window Add and delete statistics until the list in the Active Statistics box is the way you want it Apply their names to the report by clicking on the OK button Quit out of the window without saving your changes by clicking on the Cancel button if you OK prior to clicking Cancel changes are effective The names of the selected statistics will appear in the Statistics box in the main TRENDbuild window You can scroll through the names in
174. after the pg option Registers the procedure Specifies the database server name Specifies the table name and the procedure application type and optionally specifies the procedure class and procedure type Displays the version stamp for TRENDpm pe Parameters bArchive Enables archiving of raw data A value of 1 archives the raw data A value of 0 does not archive the raw data The default is 1 bCheck_index Specifies whether to use existing indices on the upload table or to drop existing indices and then recreate them The value 1 means that the existing indices on the upload table are used The value 0 means that the existing indices are dropped and then recreated The default is 0 bDelta time Specifies which clock to use to calculate Delta Time The value of 1 directs the procedure to use System Uptime to calculate Delta Time The value of 0 directs the procedure to use the Agent Clock Column for the calculation The default is 0 A 45 gt D v o Q o o TRENDpm A 46 bSuppress_spike debug_level line_suppress_value retry interval Specifies whether to reject samples if there are spikes A spike occurs when the value of any counter suddenly goes too high TRENDpm detects a spike when the value of any counter in the first sample is greater then the value of that counter in the second sample or the difference exceeds the spike threshold The value of the spike threshold is
175. ains the same header and footer columns that appear in the input_table plus all the data columns that appear on the left hand side of expressions i e column name expression The columns are added in the order of the column statements The new key table contains the standard key table columns plus all of the KEY columns that appear on the left hand side of expressions i e KEY column_name expressions TRENDstep tries to determine the type of the output column from the type of the expression If constants are used as part of the expression a floating point number indicates that the resulting column is of counter type an integer number indicates that the resulting column is of gauge type Examples of Input Script Files Three examples of TRENDstep input script files follow You may see additional examples by reviewing the files with the rot extension that are available with the installed ReportPacks in the SDPIPE HOME packages directory A 69 gt D v o Q o o TRENDstep A 70 Example1 Figure A 7 is an example of an input script file It is used to generate the output data table in Figure A 2 Example TRENDstep Output Table on page 55 source_table Rmon2_statsdata_ destination_table rmon2_protdist define IGNORE_NULL KEY if_speed KEY ifspeed if KEY nt layer ip 1 amp amp KEY ap_layer www http www octets diststatsoctets010 other ip octets 0 0 diststatsoctets010
176. ame such as frame25 or a partial word string such as frame in the text box Select the Partial Word check box if you specify a par tial word string The filtering criteria is applied to the Device Name column in the database the device entries that are retrieved for display in the Select Device list box are restricted to those that match the specified value Device Description The Match Value window for device description appears if you select this filter This window looks simi lar to the Match Value window for an element descrip tion see Figure 14 11 The filtering criteria is applied to the Device Description column in the database the device entries that are retrieved for display are restricted to those that match the specified value 3 Even after filtering criteria is specified the list of entries in the Select Device list box can be lengthy In this case you can search the list for entries that Selecting Devices and Elements 14 15 E gt D o mg m o i 3 7 7 mg 14 Report Launcher match a specified string by selecting the Search Lists option from the View menu see Figure 14 5 The Search Lists window appears if you select this option see Figure 14 8 4 Inthe Search Lists window select the Devices radio button and specify a search string in the Search text box If the string is a partial word select the Partial Word check box 5 Click Search to locate the
177. ame 8 3 9 10 name 16 8 queries 8 3 query 10 2 20 1 select statement 20 9 Server 7 8 7 9 8 3 11 3 11 11 12 12 13 14 20 2 B 1 B 15 B 18 Name 8 2 server 12 3 statement 10 12 WHERE clause 14 13 stable element 9 24 stacked bar chart 9 13 Start button 5 7 Start Time 11 12 12 13 13 15 14 19 statistic 10 5 11 13 13 10 13 11 13 13 statistics 9 2 9 10 9 12 9 15 9 16 10 3 10 8 10 9 11 6 12 7 13 8 16 2 16 8 names 10 10 statistics group RMON 16 4 Statistics to graph box 13 18 Statistics to graph box 13 13 Statistics to graph table 13 12 Status 6 4 STATUS field 6 12 Stop button 5 8 storing data data_db 7 8 Structure of Management Information 6 11 Subnet 5 10 5 12 5 14 address 5 3 mask 5 3 5 9 5 14 summarization 1 13 summarization of data 9 1 9 8 9 15 summary tables 9 8 Index 17 Index Index 18 Sybase 11 21 B 1 C 3 C 6 interfaces file 2 4 Sybase error 1105 B 10 Sybase transaction log 2 16 2 17 sysObjectID E 3 system default values 2 15 system wide default values 2 14 T ta period 9 4 9 17 10 3 A 39 ta samples 9 17 10 3 tab format 20 14 table control 16 4 data 16 4 dsi node list 17 6 dsi protocols 17 5 17 6 key 16 4 maximum width 6 21 Table Creation Time 17 2 Table Display area 15 3 table element A 81 table keys 11 14 table names prefixes 9 10 table key 9 8 tables 10 2 deleting 2 17 truncating 2 16 tabular report 11 2 11 6 11 7 11 12 12 21 target 17
178. ame size The type statement defines the length of a column that contains string data The default size is 20 bytes the maximum size is 255 bytes The characters enclose the size and are required Example type ifdescr 128 The byvariable input_table and output table statements from the 3 5 x and earlier versions are still available Please note that they may be discontinued in future releases since TREND has standardized on the new labels for all rollup programs A 61 gt D v o Q o o TRENDstep A 62 The following statements are required in the input file source_table table_name destination_table table_name column namezexpression5 9 9 9 by_variable expression Every command in an input file must end with a semicolon Blank spaces in the input file are ignored unless they are part of a character string These statements and their components are described in detail below The length of the table_name in the destination_table statement should have a max imum of 31 characters otherwise if the name is longer TRENDstep creates an in valid key table If you have inadvertently specified a table name that is greater than 31 characters in the destination_table statement and already executed TRENDstep to create the in valid key table perform the following steps to remove it 1 Kill the TRENDstep process after it has created the base table and invalid key table
179. an write the data shown in the TRENDsheet report to an ASCII file on disk To export the data in your report to an ASCII text file on disk select the Export op tion under the File menu This will bring up the Export window Figure 11 10 Export Window If every column in the report will be written to the file click the All Columns box If you only want specific columns exported click the Selected Columns box Select the columns you want exported by double clicking on their column heading cells in the report window The selected columns will be highlighted Exporting Data TREND 3 6 User s Guide Enter the directory path where you want the file written on the Path line The default is your home directory Enter the name of the file to be written on the Name line The default is export txt Click the Apply button to export the data from your report to the file The Apply button will remain grayed out while the export is running Click the Reset button to change the values in the window back to what they were the last time you did an Apply or to their default values if you have not clicked on the Apply button since bringing up the window Click the Cancel button to quit out of the Export window Format of a TRENDsheet Export File An export file contains one line for each row in the report The fields are written to the file in the same order in which they appear in the report The fields are sepa rated by tab characters Th
180. anager display See Data Manager on page 8 1 for details This option is mutually exclusive with the f and s options If you omit the f s and t options all data tables are aged which is the default db_delete_data Specifies when to update the index statistics page Valid values are 1 Before running db_delete_data 2 After running db delete data 3 Before and after running db delete data The default is no update statistics If you specify this option you must also specify the U option Specifies the day of the week to update the database statistics page Valid values are SU Sunday MO Monday TU Tuesday WE Wednesday TH Thursday FR Friday SA Saturday There is no default If you specify this option you must also specify the u option Displays the version stamp for db delete data A 7 gt D E v o Q o o db_delete_data Examples A 8 The following command deletes data from each TREND database table one at a time according to the aging criteria specified for each table in the database db delete data The following command Runs concurrent copies of db delete data to delete data from all TREND data base tables Five tables are processed concurrently Runs update statistics on Sunday after db delete data completes db delete data c 5 u 2 U SU The following command ages the data in the mib II ifEntry table deleting data ac cording to the aging value s
181. ancel button to quit out of the window Adding a New Key You can add a new instance to the current table with the Set Tool window if the current node allows it for the current table 1 Enter the value for the new key on the New Key line then click the Add button If the operation is successful you will see a box with the following message on your screen set successful 2 Click the OK button in the box to continue 6 36 Utilities TREND Collect Data The Collect Data application s main functions are to define an enterprise polling pol icy for regular collection of management data and to administer the storage of data in a relational database It is easy to use because requests are set up through a point and click GUI and it is powerful because a single collection request can be pointed at a group of nodes or at a type of node and all appropriate devices are polled auto matically The list of devices is dynamic as nodes are added to or removed from the group polled devices change accordingly The user can modify or delete current polling definitions as necessary Note To add MIB variables not currently in your TREND database see MIBwalker on page 6 1 7 1 N e o D oO D e 5 7 Collect Data Collect Data The application is invoked by clicking on the Collect Data button in the main TREND window Collect Data Op x File View Tools User Ai users pal Edit Entry De
182. and Manipulation 9 6 The following diagram illustrates this processing Raw Table Rate Table Non Non ta_period Counter Counter ta_period Counter Counter Delta_Time Item Item Item Item ty V1 V1 ty C SV9 V2 900 VI ty V2 V2 Likewise the sample for period t is compared to the sample for t For counter data types the value v3 v5 is stored in row ty For noncounter data the value v3 collected at period t3 is stored in row t The delta time for row t is again 900 Data Cleaning TREND 3 6 User s Guide The following diagram summarizes this processing over the course of a day Raw Table Rate Table Non Non ta_period toute Counter ta period Counter Counter Delta_Time Item Item Item Item ti V 1 V 1 t 1 Cc 17 V5 900 VIVI t2 V2 V2 t2 C95 V3 900 V3 V2 t3 V3 V3 t3 C35 V4 900 Va V 3 ty V 4 V 4 to 5 Co 57 V96 900 V967V95 tog V96 V96 Thus during the cleaning process pairs of rows from the raw data tables are compared each row of the output rate table is expressed as delta data that represents the result of the comparison At the end of day 1 then the raw tables contain 96 rows of data one row for each sampling period However the rate tables contain only 95 rows of data at the end of day 1 since the sampling taken at period t is lost For the second day the rate output tables contain all 96 rows becau
183. ange Database 7 9 11 3 11 11 12 12 Change Node 6 29 Collect Data Periodically 7 3 7 5 Column Format 11 4 11 16 11 17 Community Strings 6 8 Create MIBlet 6 7 Define Nodes 2 5 Define Users and Groups 2 3 3 6 Define Views 2 7 Delete table s 8 4 descriptions 5 13 Do Hostname Translations 5 11 Edit Current View 11 4 11 9 Export 11 3 11 18 Find 6 7 6 9 Load Mib 6 6 Print 1 18 Print Now 15 3 Refresh 8 5 11 4 Save 1 17 Save As 1 17 Select Data 12 11 13 3 13 9 Select Node 11 4 11 9 Set Default Storage Time 8 5 Set Table 6 35 Set Variable 6 34 Show by Source 8 5 Show by Type 8 5 SNMP Tool 6 7 6 31 Sort 8 5 Sort Order 7 3 7 11 11 3 11 14 Specify Report Directory 3 7 Time Period 11 3 11 12 13 4 Truncate table s 8 4 View Collection 6 7 X axis 13 4 Y Axis 13 4 Order by box 7 12 Order Requests window 7 11 Other Host 6 29 outoctet 9 14 P packet 5 12 18 5 parallel 21 3 21 5 C 19 parameter 13 12 15 4 18 13 file 20 7 IP address range 5 9 Read Community String 5 10 Subnet mask 5 8 parameters 13 5 13 11 15 5 20 3 22 18 Parameters button 15 5 parameters of discovery 5 8 TREND 3 6 User s Guide parent collector 17 6 parent protocol 17 16 parmVal qualifier 17 16 Partial Word 6 10 pattern 12 14 pdu 5 12 5 13 5 19 performance level 13 1 performance statistics 7 15 periodic collection 1 4 plot line 12 12 12 16 point color 12 15 type 12 15 width 12 15 Poll by Type 7 14 For box 7 5 Fro
184. appears dsedit Directory Service Session D 12 Creating the Sybase Interfaces File on UNIX Systems TREND 3 6 User s Guide 4 Highlight the Sybase server from the list of Available Servers and click the Add New Server Entry button The Server Entry Editor dialog appears ie i o lt iv o m 3 Z S 3 D ko 3 amperes e dn bumas a 5 Type in the name of the secondary Sybase server not to be confused with the server s host name and click on the Add New Network Transport button Note The Sybase server name on a UNIX platform is typically hostname_SYBASE The Sybase server name on an NT platform is typi cally the same as the host name In this example we are adding a UNIX server named BANYAS_SYBASE to the Sybase Interfaces file Creating the Sybase Interfaces File on UNIX Systems D 13 D Setting Up a Satellite Server Environment A secondary dialog entitled Network Transport Editor appears dsedit Network Transport Editor Be a pes o EIE 1 0k Cancel Help 6 Select tli tcp as the Transport Type enter the secondary server s host name BANYAS in this example enter the port number that is configured for Sybase connections on the secondary server and click OK Note The port number for TREND servers installed on a UNIX system is typ ically 2052 The port number for TREND servers installed on a Win dows NT system is typically 5000 This returns
185. arameters C 6 With Sybase 11 and above many configuration parameters can be set to improve over all Sybase throughput The most effective ones are related to memory and cache pools When performance tuning any system or application you change only one parameter at a time If you change more than one at a time and the net change results in degraded performance you will not be able to easily identify the change that caused the negative effect Here are some suggestions for determining and changing the values of some Sybase configuration parameters 1 First obtain a configuration listing from an operational Sybase system by issuing the sp configure command isql Usa P sp configure go 2 Look for the information discussed below in the output Sybase Configuration Parameters TREND 3 6 User s Guide Total Memory Find the line for total memory in the Cache Manager group in the configuration output listing Group Cache Manager Parameter Name Default Memory Used Config Value Run Value memory alignment boundary 2048 0 2048 2048 number of index trips 0 0 0 0 number of oam trips 0 0 0 0 procedure cache percent 20 32934 20 20 total data cache size 0 126978 0 126978 total memory 7500 180000 90000 90000 By default Sybase allocates 7 500 pages 15MB of RAM which is insufficient for any implementation In this example 90 000 2K pages are allocated to Sybase which amounts to 180MB of RAM allocated to Sybase For
186. arged See the output from the sp_helpdb command issued in step 2 In this example dpipe_db and tempdb require that you set the select into bulkcopy option If you are enlarging tempdb substitute tempdb for dpipe_db in the command 1 gt sp_dboption dpipe_db select into bulkcopy true 2 gt go Run the CHECKPOINT command in the database that was changed return status 0 Note Skip step 10 and step 11 if you are enlarging tempdb 10 Reset the trunc log on chkpt option for dpipe_db This step is only 11 required if this option was not originally set on the database being enlarged 15 sp dboption dpipe db trunc log on chkpt true 25 go Run the CHECKPOINT command in the database that was changed return status 0 Run the checkpoint command for dpipe db after performing step 9 or step 10 15 use dpipe db 25 go 15 checkpoint 25 go Enlarging the Database TREND 3 6 User s Guide 12 You can see the enlarged database size information from inside Sybase by typ ing the following commands If you enlarged tempdb substitute tempdb for dpipe_db unix isql Udsi dpipe Pdsi dpipe 15 sp helpdb dpipe db 25 go 13 Restart TREND unix su Password root sh etc rc TREND DB up root sh etc rc TREND timer up Starting Sybase from the Command Line To start Sybase from the command line execute the following commands unix su sybase sybase cd sybase install sybase
187. artial listing of entries received for a GET on the ifEntry variable i Set Variable Set Table Abort T Group by Instance Cancel Variables under ifEntry Entries 44 INTEGER INTEGER 2 STRING MS TCP Loopback interface STRING Fast Ethernet Adapter INTEGER softwareLoopback 24 INTEGER ethernet csmacd 6 INTEGER 1500 INTEGER 1500 GAUGE 10000000 GAUGE 10000000 Figure 6 28 SNMP Tool with ifEntry Variable Displayed If you select a specific variable only that variable is polled but you get the value of that variable for all instances of the table it is in If you select a table MIBwalker polls for every variable in every instance of that table including the contents of any tables under the selected table A counter on the left side of the window below the Get button increases in increments of 50 as data is received up to 2000 entries If the Get operation produces more than 2000 entries for your selection the following screen appears mibwalker x N Operation aborted too many rows Figure 6 29 Too Many Rows Error Message The 2000 rows collected are still shown in the SNMP tool window Utilities TREND 3 6 User s Guide Once the GetNext completes the number of variables returned is displayed in the message Variables under lt object gt Entries lt number gt where lt object gt is the name of the object selected to be collected in the MIB ma
188. ates that data is collected for specified hosts E ca 02 i xXx indicates the interval to repeat the poll 2 O gs PYY specifies the maximum number of objects that 253 appear in the collection request 2 O 35 E Z 3 The following message shows a specific example for an mw collect process initiation Process 1234 collects data for mw collect using a specified list of hosts at 5 minute intervals with a maximum of 10 SNMP variables per packet Pid 1234 usr TRENDsnmp bin mw collect n i 5 p10 An example of the complete log entry follows trendtimer Fri Apr 10 09 20 03 1998 Pid 1234 usr TRENDsnmp bin mw collect n i 5 p10 Process Completion Message Every time a process completes normally a message appears in the log The format of the messages follows Process id nnnn terminated x sec where nnnn is the process id number and x is the number of seconds the process took to complete Message Handling 2 27 2 TREND Configuration for trendadm For example the following message shows that process 5678 completed processing in less than 1 second Process id 5678 terminated 0 sec An example of the complete log entry follows trendtimer Fri Apr 10 01 00 02 1998 Process id 5678 terminated 0 sec Process Error Message Any time a process produces an error a message appears in the log with the date and timestamp the process records the message The format of the message follows progra
189. atively small delta rank indicates stable element performance TRENDrank ranks the columns of a daily summary table that TRENDsum has cre ated and stores the results in a ranking table in the TREND database The ranking table has one row per day for each element that is being ranked and the following seven columns Timestamp for day1 day2 day3 and so on ID of the element being ranked Rank 1 2 3 and so on Daily summary value of the metric being ranked Rank for the element on the previous day Delta rank This item represents the difference in the element s rank between yesterday and today That is if the element was ranked 8 yesterday and 5 today delta rank is 3 Previous delta rank Ranking TREND 3 6 User s Guide Ranking Data that is ranked by TRENDrank appears in the Top Ten reports in the TREND ReportPacks Figure 9 5 shows an excerpt from the LAN Connectivity Top Ten oes K TAL X report LAN Connectivity Wine and poorest heath Change rankings indate LAN maraoes T Top Ten pana m m E N p feat aao 1 aa feat 2 rese 2 2 am sumas 7 a 5 CT 1 menm 3 meus 1 emos 5 apo 1 eene 6 6 s sumas 2 Bc 7 9 ame Aa 6 9 apo z ese 0 02 Press 2 asar io el o2 kaaa mare 76763240728 77446520437 1 1 C remwen 3 aa TI NEN aar z 20 72725
190. ator expression This statement defines a case within the switch block Subsequent statements in the case are applied to all rows from the input file that match the specified expres sion The next case default or end switch statement signals the end of the case column_name expression This statement defines a column in the output table The order of the column statements is the order of the columns in the output table if created This statement is required TRENDstep Statement Description comment Comments begin with the characters and end with the character Examples this is a comment this is a comment type ifdescr 128 comment can be on the same line as another statement too default This statement defines the default case within the switch block Subsequent statements in the default case are applied to all rows from the input file that are not processed by any preceding case in the switch block if expression elseif expression else endif These statements define the processing in an if block One or more column expression statements can follow the if and elseif statements switch expression endswitch These statements enclose the statements for the switch block The statements in the switch block are executed for each row of the input table that satisfies the expression given on the switch statement type column_n
191. atrix table kept to the minimum number of meaningful rows TREND RMON Reports All TREND reports have been developed using TREND reporting tools with no special code having been written Some of these reports however required the con struction of a database view to extend the database schema Viewing RMON Tables in Data Manager 16 8 When you first look at the RMON tables in Data Manager there appear to be pairs of raw data tables However a closer look will reveal some differences The tables whose Object names end in data and whose SQL names end in data are the actual data tables The other similarly named table is the control table For example the table with Object name rmon history and SQL name rmon hist is the control table for the history table The table with Object name rmon history data and SQL name rmon histdata is the data table for the history table The Rollup column value for the control tables is set to no This must not be modified The ethernet statistics table and token ring statistics tables share a common control table rmon ethernet statistics The ethernet history table and token ring history tables share a common control table rmon history TREND RMON Reports TREND 3 6 User s Guide Domains The RMON probes made by Netscout software implement domains that allow the user to create subsets of the RMON tables which only contain data that pertains to the defined domains Domains can fi
192. ave mw_collect execute all collection requests with a 5 minute interval on the machine on which mw collect is installed putting a maximum of 20 MIB variables in each packet trying to poll the node a maximum of 3 times and waiting 2 seconds between polling attempts invoke mw_collect with the command line mw collect n i 5 p 20 o 2 r 3 Interval polling entries for mw collect are in the following file SDPIPE HOME lib trendtimer sched Direct Polling To have mw collect poll a node called foo Desktalk com for the MIB II ifEntry ta ble use the command gt D v o Q o o mw collect h foo Desktalk com t mib II ifEntry i 59 In this example the i option is the Collection ID which means that mw collect will not read from the database polling policy that corresponds to interval 59 A 29 rmon_collect rmon_collect Polls RMON probes for RMON MIB data Synopsis A 30 Format 1 rmon_collect B number of minutes lt default 60 gt NT version only e max number of processes C number of minutes default 30 gt NT version only d debug level i interval k n o timeout in seconds p max entries per pdu P snmp port number lt default 161 gt r number of retries s round factor sec default 3005 u V w high water mark default 90 gt rmon_collect Format 2 rmon_collect Description B number
193. aved as a postscript ps file You can specify an extension of bmp or gif to save the out put in one of these formats instead Command Line Options 13 21 13 Grade of Service TREND 13 22 TREND 14 Report Launcher B D r o i 3 7 IT D The Report Launcher allows you to run a TREND installed or user defined report against data from any available database on an ad hoc basis This chapter describes how to Invoke the Report Launcher Select the database against which the report is run Select the report Select the devices and elements to include Specify the report time period Q 9 9 2 Generate a report 14 1 14 Report Launcher Invoking the Report Launcher 14 2 You invoke the Report Launcher by clicking on the Report Launcher button in the TREND main window TREND Figure 14 1 TREND Main Window In response the Report Launcher window appears Invoking the Report Launcher TREND 3 6 User s Guide Report Launcher BEI Top Level Report Directory Up one level A poj 1 m o c 3 1 I Figure 14 2 Report Launcher Window The controls in the Report Launcher window are Control Description Look in edit box Identifies the current report directory The current directory determines the list of reports displayed in the Reports list box Invoking the Report Launcher
194. bbreviation requires an lt offset gt and an offset of 0 00 is invalid To run every Saturday at 1 00 a m the entry is SA 1 00 For a specific day of the month use the letters MONTH where x is the day of the month An offset is required to specify the hour For example MONTH2 7 00 means to run the command every second day of the month at 7 00 a m MONTH29 1 00 means to run the command on the 29th day of the month at 1 00 a m By default trendtimer runs the commands in the schedule file according to local time You can have it run according to Greenwich Mean Time GMT by putting the lowercase letter g in front of lt interval gt lt offset gt This can be useful if you are con cerned about daylight savings time The hyphens in the syntax are place holders for future arguments They must be included lt command line gt is the entire command to be run by trendtimer This is the rest of the line in the file It does not need to be inside quote marks Environment variables can be used in the command line by placing them inside and omitting the lead ing in the UNIX variable name or leading and trailing in the Windows variable name Comment lines in the file begin with a TRENDtimer will only schedule the first 50 entries in the trendtimer sched file A sample trendtimer sched file is trendtimer sched 5 10 15 A 92 DPIPE HOME bin mw collect i 5 n DPIPE HOME bin mw collect i
195. below Note Bar and area graphs present cumulative displays of data Time periods must intersect Selecting Different Data Displaying Data for a Different Node When a TRENDgraph report is created plot lines must be defined Statistics or ex pressions must be identified as well as which device key pair is to be the source of the data To change the data source node for a report select the Select Data option mh N D m Z o Q r 5 o Selecting Different Data 12 11 12 TRENDgraph under the View menu This will bring up the Select Data Set window Figure 12 4 Follow these steps for each plot line 1 Click on a row in the Lines to graph box The window will highlight the defini tion for that line 2 Set the desired values for the Select Device and Select Key boxes 3 Click the Update Line button When the lines are set the way you want Click the Graph button to display the data from the new source Note that you select the device key pair for each plot line in dependently You can have different nodes providing data to the same report You do not have to use a single node as the source for all data You may want to save the report again so the next time it runs it uses the current data source Displaying Data from a Different Database 12 12 You can access a different SQL Server to display its version of the information pre sented in this report Click the Change Database option under t
196. can use wild cards as part of the val ue on the line By default the value in the Filter box is so all files in the current directory are displayed Once you are in the directory in which the report you want to run is stored you can run it by either double clicking on its name in the Files window or clicking once on Commonly Used Commands TREND 3 6 User s Guide its name to select it which will cause its name to appear in the Select File box then clicking on the OK button The type of report defined in the file is determined by the file name extension as follows Extension Report Type gos Grade of Service Report qgr Graphical Report qss Table Spreadsheet Report Save The Save option is used to save a previously saved report Save As The Save As option is used to save a newly created file to disk or to change the name of a previously saved report Selecting the Save As command brings up the Select File window which allows you to define the name and path of the file Print By default TREND uses postscript to provide a hard copy of the results of a report The printout will give you the entire report as it appears in the display area e g the Commonly Used Commands 1 17 L ET pia mg f ex 7 E 3 1 Introduction to TREND area inside the box ina TRENDgraph window on your screen If there are multiple pages they will be numbered To get a
197. cations and processes can be installed on a single machine or on sep arate machines Function Installed on Applications Polling Agents polling station mw collect rmon collect Rollup Aging Processes client machine or data trendit base server db delete data Client Applications client machine licensed TREND on an X term or host TRENDbuild user display basis TRENDgraph TRENDsheet Grade of Service Data Manager MIBwalker TRENDcopy TREND includes an additional licensing mechanism to track a purchased TREND database size against the database s daily usage Initially you must call DeskTalk Systems Technical Support to obtain a license key for this new feature If daily da tabase usage exceeds the purchased database size a daily warning message is dis played on the screen and in trend log You continue to receive these warning messages until you purchase an expanded TREND database and a new license key is generated 2 20 License Key Administration TREND 3 6 User s Guide License Composition TREND uses a license key for copy protection and authorization Each license key contains a composite of features The TREND features numbered from 30 through AO are as follows Q 0 0 8 8 0 0 0 0 8 9 30 31 32 33 34 35 36 37 38 39 40 Server GUI Applications Poller GUI Applications expert i e MIBwalker 2 Poller RMON S TRENDcopy 3 Unused at this time
198. ces tmp critical_interfaces_list format comma style individual end Example 4 demonstrates the use of the exception keyword In the first report the ex ception keyword has been turned on This causes the system to generate a target name table key listing in the file tmp critical_interfaces_list The generation of this list is based upon whatever criteria was specified in the exception report The output from this exception report is then used as an instances input list for the next report Note that the second report is set to style individual which means that for each target name table key pair a separate output file will be generated Also note that the frequency daily keyword was used This means that both the exception report and the drilldown report will only query for the previous day s data Input Parameter File 20 13 N j D m Z o o bad o s D 20 TRENDexporter Output Format As stated earlier the ASCII files can be generated in two formats tab or comma sep arated and both formats can be generated either in a condensed or uncondensed mode Here are examples of how the output would look in the various possible com binations of these methods They are based on a TRENDsheet report which utilized the following columns Column Name Size in Database dsi_target_name 64 bytes dsi_table_key 128 bytes ifinoctets018 8 bytes Tab Separated Format Not Condensed The tab format consists of
199. chart in the Executive Summary reports supplied with the TREND ReportPacks Figure 9 2 shows an excerpt from the LAN Connectivity Executive Summary report This report provides ClOs and other executives an overview of the PR BP A TOA E OS LAN Connectivity erformanwe of corporate LANs Each chart shows kay matos paa d petiam Sm Pme T R E N D Executive Summary Grade of Service All LAN Interfaces Tues Jun 30 0 00 Tues Jun 30 23 00 Wi utilization Al Devices 0 D Discards Al Devices 0 Bl grors Al Devices 0 00 02 00 04 00 06 00 08 00 10 00 12 00 14 00 16 00 18 00 20 00 22 00 Hourly Scale Figure 9 2 LAN Connectivity Executive Summary The Base Table 9 13 o 5 2 D O E D 5 uonebaibby ejeq 6 9 Data Aggregation and Manipulation Here is an excerpt from the TRENDstep script file that is used to produce this report The excerpt illustrates how a column value from an input row is graded utilization ifoutoctets024 ifinoctets018 8 delta_time004 ifspeed013 1 100 0 if ROTATE utilization gt 40 utilization_score 4 0 utilization40to 100 1 0 else if ROTATE utilization gt 20 amp amp ROTATE utilization lt 40 utilization_score 3 0 utilization20to40 1 0 else if ROTATE utilization gt 10 amp amp ROTATE utilization lt 20 utilization_score 2 0 utilization 10to20 1 0 else utilization_score 1 0 utilizationOto10 1 0 endif In this exampl
200. collect makes log entries in the file SDPIPE HOME tmp trend log Options A 34 rmon_collect has the following options B Specifies the number of minutes that each BCP process can run The system kills the BCP process if it runs longer than the specified num ber of minutes The default is 60 minutes This option is only available on the NT version c The number of concurrent collection processes the master collection mw_collect will start When mw_collect starts it will start small chil dren processes that actually do the collection default is 5 C Specifies the number of minutes that each collector process can run The system kills the collector process if it runs longer than the speci fied number of minutes The default is 30 minutes This option is only available on the NT version rmon_collect i k n Set a debug output level Values of 0 1 2 or 3 are valid The higher the number the more detailed the information The default is 0 which means no debug output Debug output is written to standard out This option should only be used for testing in coordination with Desktalk Systems Technical Support due to the additional overhead it places on rmon collect The name of the host to be polled This name will be saved in the data base along with the data returned from the node so in order to be con sistent with other data use the same node name format used when entering nodes in the database The polling i
201. columns of data aligned with the next in a series of stan dard equally spaced tab stops That is the actual amount of separation between col umns is the number of spaces from the end of the column to the next tab stop Because this example is not condensed the amount of space occupied by each col umn is based on the field size defined in the database Below is an example of a tab separated ASCII output file with the headers option set to yes 20 14 Output Format TREND 3 6 User s Guide 64 bytes 128 bytes f 8 byes dsi target name dsi table key ifinoctets018 your router 1 123456 your router 2 123456 your router2 1 1234567 The output generated if the headers n keyword was specified would be the same data in the same format but without the column headings your router 1 123456 your router 2 123456 your router2 a 1234567 Comma Format Not Condensed The comma format consists of columns of data separated by commas Each column occupies the number of bytes specified for it in the database Note that whenever a comma delimited file is requested all character based data is enclosed in quotes With headers y set the output would look like this dsi_target_name dsi_table_key rifinoctets018 your router 4 7 17 7123456 your router E LI 123456 your router2 5 Ka 71234567 Without headers the same data would be generated but without the column headings N your router ULT 123456 your router pra 7
202. command line option If you omit the n option the default is 50 percent Rolling Baseline Table Use command line option y to specify the number of days in the rolling baseline pe riod If you specify the y option the e option is not required Once the summari zation process has started you cannot change the name of the source table used for the rolling baseline or the number of days in the baseline period Use of the y option places restrictions on the destination table name This name al ways begins with SD42 assuming a 42 day baseline Normally the remainder of the name is just the source table name For example when the source table name is SDifexceptions the destination table name becomes SD42SDifexceptions You may use the e option to append the specified value to the name For instance the following option string y42 t SDifexceptions e fore yields SD42SDifexceptionsfore as the destination table name TRENDsum A CAUTION Alternatively you may use the a option to specify a name to use instead of the source table name Here the option string y42 t SDifexceptions a mytable yields SD42mytable as the destination table name Use of the e option and the a option on the command line at the same time gener ates the following error message Use of baseline name precludes use of destination name A rolling baseline is inherently a daily summary Therefore no by variable greater than day is acc
203. countless devices and sources Critical pa rameters for a given device are rolled up into an overall indicator that reflects that device s performance level 13 1 13 Grade of Service Main Display and Menu System Select the Grade of Service button on the TREND Main window to invoke the Grade of Service application Figure 13 1 Grade of Service Report Window Command Menu Summary File 13 2 Main Display and Menu System TREND 3 6 User s Guide The New command is used to create a new Grade of Service report see Creating a New Grade of Service Report on page 13 7 o Q o Q D h W D S 7 D Use the Open command to retrieve a pre defined GOS report see Opening an Ex isting GOS Report on page 13 6 and The Open Command on page 1 15 The Save option stores previously saved GOS reports Save As stores new reports or copies existing reports Graphs can be printed directly to a postscript printer by selecting the Print command see Print on page 1 17 The Change Database command provides access to data tables in other databases see Section Displaying Data from a Different Database on page 13 14 Exit quits the Grade of Service application see Creating a New Grade of Service Report on page 13 7 and Section Exit on page 1 19 About provides TREND version information see About TREND on page 1 19 View HENA Edit elect Da
204. cription field 6 Click Next to continue 7 The Create Miblet Pair High low Counters window appears Figure 6 19 Description Pair high low counters if needed and then click the Finish button This will cause the Miblet to be created mib it Miblet Group Name is Foo l Group Property Key Table Name is Kib 1I Foo II Group Pair High ow Counters T Enable trendit summaries lt Back Finish Figure 6 19 Create Miblet Pair High low Counters Window MIBwalker automatically creates the raw and rate MIBlet tables with all other required SQL objects to support raw to delta processing 8 Tocreate the legacy summary tables from TRENDit select the Enable trendit summaries check box 6 26 Creating a MIBlet TREND 3 6 User s Guide 9 If you do have 64 bit counters then click Pair High Low Counters button This button is accessible when the selected MIB objects are 32 bit counters The Pair High Low Counters window appears Pair High Low Counters o gt Ww x D Figure 6 20 Pair High Low Counters Window a Select one entry from Select High Part to act as the high counter and one entry from Select Low Part to designate the low counter The selection appears as the following example Creating a MIBlet 6 27 6 MIBwalker b Select high part Select low part D g sitCtrSplitl Low Add Counter Pair Delete Counter Pair High
205. cuments from the Internet Engineering Task Force Definitions for enterprise MIBs come from the manufacturers of the nodes Use MIBwalker as a tool to perform the following functions Learn about the contents of a MIB file Select sets of managed objects to define data collection tables Conduct GET and GET NEXT requests to ensure that devices can be polled When you execute MIB walker it displays a MIB file as a graphic MIB map tree diagram that shows the MIB s structure It allows you to analyze the MIB s content You can use MIB walker to create a data table containing database entries from a group of selected variables which is a MIBlet See MIBlets on page 6 19 6 1 6 MIBwalker MIBwalker also provides a MIB browsing application offering GET GET NEXT and SET capability MIBwalker can poll a node for a MIB group or managed object and display the response on the screen or write it to the database MiBwalker Process Flow Figure 6 1 shows the process for using MIBwalker to create a database table in preparation for collecting and aggregating data The process consists of obtaining the MIB file for the vendor device node preparing the MIB for MIBwalker by applying typographic and syntax rules and validating object types selecting variables for the raw table and then creating the database table O 9 Original Prepare MIBs for MIBWalker Select Create MIB file for _
206. d TRENDgraph on page 12 1 If you requested a TRENDgraph the Select Data Set window will be displayed in order to define the line styles for the fields being plotted Command Line Options The TREND polling processes ee_collect mw_collect and rmon_collect are launched from the command line See Man Pages on page A 1 for a more detailed description of these applications Incomplete Data Collection or Aggregation Data collection the running of ee_collect mw_collect or rmon_collect and aggre E gation the running of TRENDit TRENDsum and other aggregation utilities place S bcp i e bulk copy and other temporary files in the following default directories ng o For UNIX Amp S For Windows NT c tmp For Windows NT the temporary files are placed in c tmp by default You can change this default location for Window NT by assigning the desired path to the variable DPIPE_TMP as follows 1 Sign on to the local domain as Administrator 2 Select Start Settings Control Panel System Environment Command Line Options 7 17 7 Collect Data 3 Assign the desired path to DPIPE_TMP For example DPIPE_TMP tmp If the tmp directory in UNIX or the DPIPE_TMP or default c tmp directory in Windows NT does not exist the TREND data collection and aggregation opera tions are not performed successfully No error message is issued If you are having problems with incomplete data collection or aggregat
207. d Creating Expressions on page 10 10 Note If an alias is set and plotted in GOS and then changed in TRENDbuild the grapher will still show the original alias because TRENDbuild does not automatically update the lines added to the report file by GOS You must resave and reopen the file after modifications have been completed Main Display and Menu System TREND 3 6 User s Guide The Grade of Service Concept 2 Q o Q m 02 D S 7 Dm The following example explains the basic principles of Grade of Service reporting EXAMPLE Assume that you wish to compute the grade of service for a particular LAN segment You determine that the significant parameters contributing to your LAN GOS are Utilization Collisions Errors and Overhead with relative contributions as follows Parameter Weighting Utilization 4090 Collisions 25 Errors 25 Overhead 10 You next decide that you will include four categories of service in your GOS report Excellent Good Fair and Poor In this case a score of 1 for an individual parameter or weighted combined parameters will fall in the excellent range Similarly scores of 2 will denote good 3 will be fair and 4 will be considered poor performance You must also determine what ranges of scores for individual parameters constitute excellent good fair or poor service Parameter Excellent Good Fair P
208. d monitor TREND tables TREND 3 6 User s Guide Chapter 9 Chapter 10 Chapter 11 Chapter 12 Chapter 13 Chapter 14 Chapter 15 Data Aggregation and Manipulation Provides an overview of TREND data aggregation and manipulation processes and introduces the utilities that per form these functions TRENDit TRENDstep TRENDsum and TRENDrank Relates data aggregation concepts to the TREND ReportPacks TRENDbuild Describes how to create and manipulate query files which the TRENDgraph TRENDsheet and Grade of Service applications use TRENDsheet Describes how to use TRENDsheet to create manipulate and print tabular data displays and export their data to an ASCII file TRENDgraph Describes how to use TREND graph s point and click graphical user interface to create custom graphical reports Grade of Service Reporting Describes how to use the GOS application to create reports on system health in a stacked bar format Report Launcher Describes how to use the Report Launcher to run reports on an ad hoc basis Report Scheduler Describes how to use the Report Scheduler to schedule peri odic printing of reports xxvii Preface xxviii Chapter 16 Chapter 17 Chapter 18 Chapter 19 Chapter 20 Chapter 21 Chapter 22 Appendix A RMON Support Describes the setup and polling policy definition for RMON probes including support for RMON extensions RMON Support Describ
209. d to save the view then run it 3 Once you ve successfully run a view you can use TRENDbuild to generate a report file either qss or qgr and view the data in either TRENDsheet or TRENDgraph The processes for defining data tables selecting columns and assigning expressions and qualifications are detailed in the procedures below Define the Data and Key Tables in a View 1 Click the Select Data Tables button The Select data Tables window appears 2 Inthe MIB Combo Box choose a MIB from the list of those available on the current database Its tables appear in the Available Tables Box 3 Select one or more tables from the Available Tables Box and click Add The selected tables appear in the Selected Data Tables Box and their associated key tables appear in the Key Tables to Selected Data Tables Box Note At this time TRENDdbview will only support a maximum of two key tables If you highlight a table by mistake you can deselect it by selecting another table in the Available Tables Box by selecting a different MIB in the MIB Combo Box or by clicking Clear Working with TRENDabview TREND 3 6 User s Guide If you add a table by mistake highlight the table in the Selected Data Tables Box and click Delete If you decide you don t want any of the tables you ve added click Delete All The Delete All Button clears the Selected Data Tables Box To ascertain which key table references a selected data tabl
210. d to the right of the column on top of which you released the broken lined box Formatting the Report 11 15 11 TRENDsheet 11 16 If the report is wider than the TRENDsheet window and you want to move a column to a position not displaying on the screen move the column to the last position in the window scroll the report over in direction you are moving the column and move the column again Column Order and Report Sort Order You can make your report much easier to understand if you put the left hand col umns in the report in the same order as your sort order For example if you are sort ing a report on line utilization by utilization level and time of day Received Time put the Utilization column in the left hand column of the report and Received Time in the column to the right of Utilization by dragging each column to the left of the fixed Time Period column Note that you cannot move a fixed column see Fixed Columns on page 11 16 To put a different column s data in the first column drag that column to the left of the fixed column Fixed Columns Fixed columns are data columns at the left most side of the report which cannot be moved to new positions nor be replaced by other columns They also continue to dis play in the window if you scroll the report to look at different columns Numeric Column Precision By default numeric columns are displayed with four decimal places of precision You can change this
211. data Same as Cache version BCP gate way Failed loading data for other reason than above Removes the entire input file Same as Cache version A 25 gt D E v gD e D 7 mw_collect A 26 Interval Polling mw collect is mainly used to poll for SNMP data at a regular intervals When invoked mw collect reads the polling control table in the database for the list of instructions whose intervals match the value of the i option on the mw collect command line Each entry in the list specifies a MIB table to be collected and the device group spe cific device or specific instance from a device to be polled for the table This polling policy information is entered automatically when you install a Report Pack You can also enter or modify polling policy information through the Collect Data Periodically window Device groups are the View and Type lists defined through the TREND Configure Views and Configure Types windows See the TREND ReportPack Guide for information about using Autopilot See Collect Da ta on page 7 1 for information about specifying your own polling policies mw collect queries the database for the list of fields in the table to be collected Then for each node to be polled it obtains the node s SNMP Read Community string from the database as configured in the TREND Configure Nodes applica tion see Collect Data on page 7 1 for details and polls the
212. data regardless of timestamp DEFAULT 0 The name of the ASCII output file To insure that existing files arent over written the actual output file name will have a timestamp appended to it in the format YYYYM MDDHHMM If o myfile txt is specified and trend_export is run on April 10 1997 at 3 30pm the output file name will be myfile txt 19970410153 The complete path name of the ASCII out put file to be generated 2of3 20 4 Controls and Switches TREND 3 6 User s Guide Flag Description Argument p The maximum number of Any positive integer DEFAULT 5000 lines to be generated in the output file q The name of the SQL The name of the SQL Server DEFAULT Server from which the Current value of the users DSQUERY data will be extracted environment variable r The name of the The complete path name of any TRENDsheet qss query TRENDsheet query file to be used 3 of 3 Note that whenever the i option is used all other command line options are ignored except the q option In this case the named input parameter file controls ASCII file generation exclusively The utility needs one environment variable set DPIPE HOME This variable con tains the path name of the directory where all TREND system components reside The utility writes all of its messages to the trend log file and looks for the export con trol parameter files in DPIPE_HOME lib
213. de basis By default TRENDit will run once each night rolling up all data tables For each raw data table TRENDit will create or add rows to arate table an hourly table a daily table a weekly table and a monthly table based upon the amount of valid data available for aggregation TRENDbuild equates a data table to a group Las i E m 2 g ley E Q Note Raw data tables contain values as collected from the polled devices Most performance related values are counters Consequently most perfor mance reports are based on roll up tables not raw tables All TREND data tables have a similar construction consisting of header data and footer fields All headers contain identifiers for the source of the data a received timestamp a requested timestamp and a delta time value Data fields correspond to objects in the associated miblet The footer fields include indexing information and for hourly daily weekly and monthly tables a timestamp marking the beginning of the aggregation period ta_period and the number of raw data samples collected during the aggregation period ta_samples Note All data tables contain a delta time value In raw data tables delta time is the length of time since the data source was rebooted in units of one hundredths of a second i e sysuptime In rollup tables the delta time value is the time between samples in units of seconds For reporting pur poses a column called ta_sysuptime
214. des a Report Builder you can use to build the following types of reports ET d Q 7 E 3 apa TRENDview PRO Reports TRENDview PRO Data Set Reports SDT TRENDview HTML Reports You can use the TRENDview PRO Client to modify and view your own reports or the ones in the TREND supplied ReportPacks using a Java enabled web browser The TRENDview HTML Client enables you to use any web browser to view reports that have been stored in HTML format These facilities are described in the following user manuals TRENDweb Installation and Administration Guide Guide to Building TREND Reports Guide to Viewing TREND Reports Managing Data The Data Manager is an administrative tool that controls data rollup allows aging criteria to be changed and displays total database statistics Aggregation and aging criteria can be viewed by all users but only the trendadm user has the authority to change these criteria Once data has been collected TREND manages the aggregation and summarization of the data automatically You can use Data Manager to check the size of the database manually purge data from tables and control the retention times for specific data ta bles The rate hourly daily weekly and monthly tables are the source for reporting A Brief Tour of TREND 1 13 1 Introduction to TREND TRENDcopy provides incremental periodic filtered database replication service to a backup se
215. diately collect a particular MIB table s values from a par ticular node on demand by executing mw collect with the t option from the command line Log File mw collect makes log entries in the file SDPIPE HOME tmp trend log A 27 gt D v o Q o o mw_collect Cache Poller A 28 Directory Structure Temporary files data files and cached polling policy files that mw_collect uses are stored under the following directory structure COLLECT HOME feeder name collection id BCP where SCOLLECT HOME Is the directory that the SCOLLECT HOME environ ment variable points to typically SDPIPE HOME collect feeder name Always has the value MW for instances of mw collect collection id Identifies the collection ID which is a number corre sponding to the value of the i option on the mw collect command BCP Is the directory where data that is ready to be loaded into the database is stored Instructions for Loading the Cache Version of the Poller Perform the following steps to load the cache version of the poller 1 Install new license key 2 Rename the standard poller mw collect to another name such as mw collct nocache 3 Rename the cache poller mw collect cache to mw collect mw_collect Examples Interval Polling To have mw_collect execute all collection requests with a 10 minute interval invoke mw collect every 10 minutes with the command line mw_collect i 10 To h
216. dify Discovery files E 1 module 1 4 Collect Data 1 4 Data Manager 1 4 db_delete_data 1 4 Grade of Service 1 5 MIBwalker 1 5 mw_collect 1 5 Report Launcher 1 5 Report Scheduler 1 5 rmon_collect 1 5 TRENDbuild 1 6 trendcopy 1 6 TRENDodbview 1 6 trendexport 1 6 TRENDgraph 1 7 TRENDit 1 7 trendlabel 1 7 trendproc 1 7 TRENDrank 1 7 TRENDsheet 1 8 TRENDstep 1 8 TRENDsum 1 8 trendtimer 1 8 monochrome printer 12 15 13 16 13 19 multiple databases TREND installations 7 8 multiple sort keys 11 15 mw collect 1 5 1 8 1 9 7 15 7 17 C 12 C 19 utility A 14 mw collection table 5 15 N nested sort 7 12 NETscout software Domain View exten sions 16 2 New Key adding a 6 36 NIS 2 6 NIS user database 2 4 No Filter 14 12 14 15 14 16 node 5 1 5 14 5 16 5 20 6 8 7 1 7 6 7 7 11 10 13 13 15 4 16 4 17 6 Define 3 6 define 2 5 definition 6 1 import 3 4 name 7 12 new 2 6 Select 3 5 type 2 2 2 6 3 6 D 7 views 2 2 Node Type option 2 6 Index 11 Index Index 12 Node View Type 7 10 node_name 3 4 nodes 2 2 2 6 2 10 nowait 21 10 nowait block 21 3 21 4 number of retries 17 10 numeric column precision TRENDsheet 11 16 O Object name 6 25 6 33 8 3 16 8 Type 6 4 6 14 OBJECT IDENTIFIER 6 11 obsolete data A 4 OID 17 12 17 13 E 3 OID string 17 6 Open command 1 15 operators 12 7 option Add Drilldown Report 15 3 Add New Report 15 3 15 4 aging 2 16 Aliases Expressions 13 4 Ch
217. ding key table does not exist it is created and the dsi_key_tables table updated accordingly In addition to the standard syntax listed above the t option can be used in multiple ways The destination key table and destination table names can be eliminated from the command string as follows trendcopy t source table s source server S target server If you want to omit the key table name and retain the destination table name how ever the syntax becomes trendcopy t source table destination table S Source server S destination server When the destination table name is specified and no entry exists for this name the compression routine creates a table name If source or destination table names in clude the key words rate hour day week or month the first letter of the created table name is modified to the corresponding roll up level TRENDcopy provides an error message when the destination key table name of the t option is improper or invalid It continues the copy process using a default naming Usage Notes 18 7 18 TRENDcopy convention for the destination key table unless the destination data table and desti nation key table already exist For example TRENDcopy creates the destination data table called basic_test_info_ and the destination key table called Kasic_test_info_ from the following syntax t basic_test_info_ ktest_info basic_test_info_ It creates the destination key table name by repl
218. do this Click the Save As option under the File menu This will bring up the File Se lection window allowing you to write the current report definition to a new report file on disk The new file can be modified to create a different report then saved to preserve the changes Creating a New Graph To create a new graph select New from the File menu of the TRENDgraph window This command invokes the TRENDbuild application where new reports are defined N E m O Q o ko T Creating a New Graph 12 5 12 TRENDgraph untitled Figure 12 2 TRENDbuild Window Selecting Statistics to be Graphed 12 6 The first step in creating a new graph is to select the category MIB and group of the collected data to be reported upon Use the Select Mib and Select Group buttons in the Report Scheduler window EXAMPLE Selecting the Hourly MIB II ifEntry Group To create a report based on hourly mib II ifEntry data first click on Select Mib and select mib II from the Select Mib pop up box Click OK Next click on Select Group and hour_mib II_ifEntry from the Select Group pop up box Click OK The next step is building expressions to be graphed using the available statistics Click on Assign Expressions box in the Report Scheduler window to invoke the Assign Statistic Aliases window Figure 12 3 Creating a New Graph TREND 3 6 User s Guide Assign Statistic Aliases Figure 12 3 Assign Statistic Alias
219. done using TRENDbuild and the procedure for doing so is described in the chapter on building TREND reports Working with TRENDabview TREND A Man Pages age_files age files Allows you to set up the automatic deletion of files from a specified directory age files h i days 1 p directory s V Description This utility enables you to delete files automatically from the specified directory and its subdirectories if you also specify the s option Options age files has the following options h Displays the command line options help i Specifies the number of days to keep a file Files whose creation date is older than the number of days specified for this option are deleted l Lists the contents of the target directory but does not delete the files p Identifies the directory to age The default is the current directory A 2 age_files V Removes files meeting the aging criteria specified for the i option from all subdirectories of the directory identified in the p option Displays the current version and build number of the age_files utility A 3 gt D E v o Q o o db_delete_data db delete data Ages obsolete data out of the database db delete data c of child processes d debug level f feeder h i fime index m type s sqlname t tablename u update statistics U day of week v Description
220. dow The time period you have specified is displayed after the Time Period label in the Report Launcher window If no time period is given for the report in the database or the report definition file you cannot specify a time period The Set Time Period window is inactive and the follow ing text is displayed after the Time Period label in the Report Launcher window Not Specified in Report Specifying the Report Time Period 14 19 14 Report Launcher You can quickly request a report for the previous period by clicking on the Prev Hour Prev Day Prev Week or Prev Month button to the left of the Set Time button in the Report Launcher window The button is labeled Prev Hour Prev Day Prev Week or Prev Month respectively depending on whether an hourly daily weekly or monthly report is selected in the Reports box In the example see Figure 14 12 the button is labeled Prev Day be cause a daily report is selected in the Reports box Generating the Report After you have specified the report device s element s and Start and End Times as appropriate click the View Report button to generate the report Note The View Report button appears in the Report Launcher window see Figure 14 12 and in the Devices window see Figure 14 9 You can launch the report by clicking on the button in either window If you want to produce the same report for different devices or elements you might find it handy to leave the Devices
221. e A new column named utilization in the output table is computed from an expression that uses inoctet outoctet delta_time and ifspeed values from the rate input table The 1 in the expression ifspeed013 1 is used to prevent a potential divide by zero condition The ROTATE notation in the next state ment is used to refer to the new column in the output table A new column named utilization score is computed and assigned a graded value as follows 4 0 for utilization percentages in the range 40 100 3 0 for utilization percentages in the range 20 40 2 0 for utilization percentages in the range 10 20 1 0 for utilization percentages less than 10 9 14 The Base Table o TREND 3 6 User s Guide So ao Led o mop TRENDsum then uses this graded value in the utilization score column to compute ES a single average GOS value allgos across all LAN interfaces combined as shown lo in the following formula E on 30 column allgos utilization score 0 60 discard pct score 0 25 error_pct_score 0 15 avg The allgos value appears in the LAN Connectivity Executive Summary report shown in Figure 9 2 Other Data Aggregation Functions Once TRENDit has cleaned the raw input data and TRENDstep has optionally conditionally processed the rows of the resulting rate table TRENDsum can be used to summarize the data by any defined element property by time or by element property and t
222. e TRENDsheet display is filtered to include the objects and instances associated with the selected bar graph This feature is particularly useful in the grade of service ap plication Instances of performance that are below a certain level can be analyzed in detail by drilling down Si d O a o o 3 o Tabular Reports By executing the TRENDsheet module from the main menu you can browse and se lect from a list of pre formatted spreadsheet style reports Like graphical reports TRENDsheet reports are based on query files Tabular display of selected statistics can be sorted and displayed with a number of options Output can be printed to post script printers or saved to a file like TRENDgraph TRENDsheet also supports ex ception reporting Exception reporting allows you to define constraints using Boolean expressions as conditions for executing a particular tabular report Grade of Service GOS Reports GOS reporting is a valuable tool for network and systems health reporting and Ser vice Level Agreements SLAs The GOS algorithm converts historical performance statistics into instantly meaningful graphical displays using a weighted stacked bar chart format GOS allows the user to select any set of SNMP or RMON metrics and assign a grading scale to each selected metric based on relative importance An over A Brief Tour of TREND 1 11 1 Introduction to TREND all grading scale e g excellent good fair
223. e from the Select View box and one value from the Select Type box The values you select will appear on the Select View and Select Type lines Select the desired Polling Mode from the box then select the values for the mode as described above The values displayed on the Node View and Type lines are se lected from the corresponding boxes as described above not as input lines N e o o o D r 5 Specifying How Often to Poll for the Data TREND polls for requested data at regular intervals as specified in the collection re quest definition To select a polling interval for the current request click on the down arrow next to the Poll Interval box then click on the desired interval If you select Off this request will be ignored Setting Up New Polling Requests 7 7 7 Collect Data Specifying Which System Should Poll for the Data It can be more efficient to poll from a system closer to the intended targets than from the one on which the request was defined This approach reduces network overhead and distributes the polling load By default the system on which polling is defined also does the polling You can override this by entering a different client name on the Poll From line The node name entered here must be configured to run TREND polling agents Collections Involving Multiple Database TREND Installations 7 8 From the user s perspective TREND s use of multiple databases for data collection is c
224. e highlight the table in the Selected Data Tables Box and click the Data Table gt Key Table Button An information window pops up showing the associated key table N N D m z o ex o To ascertain which data table is referenced by a selected key table high light the table in the Key Tables to Selected Data Tables Box and click the Key Table gt Data Table Button An information window pops up showing the associated key table Kii_snmp_overhead_ gt day mib II snmp overhead d 4 By default the first key table that appears in the Key Tables to Selected Data Tables Box also appears in the Key Table for View Box This is the key table by which the view will be referenced To change the key table used for the view double click the other key table in the Key Tables to Selected Data Tables Box The new key table will appear in the Key Table for View Box in place of the default key table Working with TRENDdbview 22 13 22 TRENDabview 22 14 5 When the information shown in the Select Data Tables window is correct click OK The Select Data Tables window closes and you return to the TRENDdb view main window Once you have specified the tables for your view you can save and run it It is not necessary to define either expressions or qualifications If you choose not to specify which columns to include your view will include only the thirteen default columns from the primary tab
225. e port number that is configured for Sybase connections on the secondary server after the prompt 7 i o lt i o m Z E 3 iy ko 3 apjayes e dn bumas a 9 Press Ctrl a to accept your entry 10 Enter y to the prompt Is this information correct and press Return 11 Press Ctrl a to accept your entry 12 Enter y to the prompt Write the changes to the interfaces file now and press Return 13 Press Ctrl x to exit the current screen 14 Press Ctrl x again to exit the sybinit utility For Sybase 11 5 Use the sybsetup utility to install the Sybase interfaces file for Sybase 11 5 on UNIX systems Perform these steps Creating the Sybase Interfaces File on UNIX Systems D 9 D Setting Up a Satellite Server Environment 1 Bring up a terminal window 2 Bring up the sybsetup utility by typing the following command SSYBASE bin sybsetup 3 Select Specify Directory Services Information from the Configure Sybase Servers menu 4 Highlight Sybase Interfaces File on the Select a Directory Service dialog ver ify the location of the interfaces file and press OK 5 Highlight the Sybase Server from the list of Available Servers and select Add New Server Entry on the Directory Service Session dialog 6 On the Server Entry Editor dialog type in the name of the secondary Sybase server not to be confused with the server s host name and select Add New Network Transport This bri
226. e 6 14 Command Menu Option Summary 6 6 File l3 View Tools Load Mib Create MIBIED Exit About Figure 6 4 File Selection Window The Load Mib option allows you to load a different MIB into MIBwalker The name of the currently loaded MIB file appears in the header bar See Loading a MIB on page 6 10 Main Display and Menu System TREND 3 6 User s Guide The Create MIBlet option allows you to create a raw collection table in the database comprised of selected managed objects from a MIB See MIBlets on page 6 19 The Exit command quits the MIBwalker tool The About option provides version information for TREND See About TREND on page 1 19 View o W 2 2 Figure 6 5 View Selection Window The Find option is used to locate an object in the MIB map provided you know all or part of the object s name The View Collection option brings up the Collect Data window See Collect Data on page 7 1 Tools Tools SNMP Tool Figure 6 6 Tools Selection Window The SNMP Tool polls the current node for the current values of selected MIB objects and displays the results on the screen See Polling Data to Screen on page 6 31 Main Display and Menu System 6 7 6 MIBwalker The Change Node option allows you to change the currently polled node MIBwalker uses the current node as the target when polling for current MIB values See Changing the Cur
227. e Discovery is typically run automatically usually once a day It can only be initiated in two ways By the TREND Autopilot when you use it to install a TREND ReportPack From the command line Note The Discover GUI accessed from the TREND Main window is only used to launch IP Discovery The purpose of SNMP Type Discovery is to identify the nature of the devices in the dsi_nodes table What Discover Does TREND 3 6 User s Guide In particular SNMP Discovery determines the Type of a discovered device SNMP Type Discovery interrogates nodes to determine what kind device they are and records the information in the TREND database This identification allows TREND to target a specific type of device for data collection a o i o o lt 2 How SNMP Type Discovery Works Discover uses special files identified by the extension dis to determine whether or not a device is of a certain type Each dis file contains the necessary protocols to test a node for a single specific type Note dis files are discussed at greater length in Running Discover from the Command Line on page 5 11 In SNMP Type Discovery the Discover tool reads the dsi_nodes and dl_type tables to identify the nodes that require type validation SNMP Type Discovery Table Population There are two database tables that are affected by SNMP Type Discovery dl_type and dsi_nodes For each node on your network the dl_type tab
228. e Report Launcher the name of the Top Level Report Di rectory that was in effect when you last closed Report Launcher is used as the initial Top Level Report Directory and is also made the initial current directory You can designate a different Top Level Report Directory as follows 14 8 Selecting the Report File Directory TREND 3 6 User s Guide 1 Inthe Report Launcher window select Report Directories from the View menu Report Launcher Top Level Report Directory Up one level Cisco Interfaces RMON ethernet Daily RMON Ethernet Statistics EE A D Dm pap m o i 3 7 IT D Figure 14 5 Report Launcher Window View Menu Selecting the Report File Directory 14 9 14 Report Launcher 14 10 The Report Directories window appears Report Directories x Directories c trendsnmp reports c trendsnmp reports ethernet Remove Directory Current Report Directory c ttrendsnmpjreports ethernet Apply Cancel Figure 14 6 Report Directories Window Use this window to add or remove report file directories The list of directories you can work with appears in the Directories list box 1 To add a directory to the list type the full path name of the directory in the edit box and click Add To remove a directory from the list click on the directory name to high light it and then click Remove Directory To designate the directory to be u
229. e SDifexceptionsfore output table application interface reporting version 3 5 date 1998 06 11 Copyright 1998 Desktalk Systems Inc All rights reserved trend_sum proc to create days to threshold calculations from the daily summary table column utilcurrent P95utilization per95 column inutilcurrent P95waninutil per95 column oututilcurrent P 95wanoututil per95 column errorcurrent AVGerror_pct avg column discardcurrent AVGdiscard_pct avg column utiltrend P95utilization dtt 40 30 60 90 column wanutiltrend P 95wanutil dtt 40 30 60 90 column waninutiltrend P 95waninutil dtt 40 30 60 90 column wanoututiltrend P95wanoututil dtt 40 30 60 90 column discardtrend P95discard pct dtt 5 f30 f60 f90 column errorstrend P95error pct dtt 10 f30 f60 f90 by variable keyid gt D 5 v D e D 7 A separate row is created in the SDifexceptionsfore output table for each keyid rep resented in the input table A 85 TRENDsum A 86 Changes in TRENDsum 3 5 1 from 3 4 x This section compares 3 4 x and 3 5 1 TRENDsum processing 1 TRENDsum requires that all non time by variables except for the first which becomes dsi_target_name appear as columns in the destination key table All components that are used to make up the key are required for possi ble future TRENDsum operations The following table points out the TRENDsum formulas in 3 4 x that differ from t
230. e area and stacked bar formats are also supported TRENDgraph will print the displayed graph to a postscript printer or a file TRENDit Meaningful information is produced by aggregating raw data into periodic summaries The TRENDit process cre ates hourly daily weekly and monthly rollup tables from polled data Each summary provides average maximum minimum and total values for counter and gauge fields and carries forward character data and constants trendlabel A command line application that allows TREND users to label various column headers trendproc A simple language for process chaining TRENDproc makes it possible to run various TREND processes sequentially in parallel or both TRENDrank A command line application that builds a ranked table on selected metrics from a summary table TRENDrank enables you to rank and compare selected metrics from the current period with the appropriate day of the week from the baseline summary 4 of 5 TREND Overview 1 7 1 Introduction to TREND Module Description TRENDsheet TRENDsheet reports display rows and columns of infor mation selected from the data repository or calculated from stored values The data can be sorted on any combi nation of fields and columns can be resized and reordered directly on the screen Printouts of reports can be requested or the data can be exported to an ASCII format file Report definition parameters are stored
231. e clicking on the category name Click on the Cancel button to quit out of the window without making a selection The name of the category you choose will appear in the Category box in the main TRENDbuild window The second step in defining a report is to select the group which contains the statis tics to be reported on To select a group click on the Select Group button in the TRENDbuild Main win dow Figure 10 1 This will bring up the Select Group window Select Group Eg Select From Il SD sertfa day mib ll SD type p day mib ll SD type s day mib ll ifEntry day mib ll snmp over day mib ll sysuptime day mib ii ifentry su hour mib ll ifEntry hour mib ll snmp ove hour mib ll sysuptime mib ll STATUS F 0K Cancel Figure 10 3 Select Group Window A group is the specific type of data to be used in the report It is a collection of sta tistics which are stored in a single table in the database Just as a category can be thought of as an SNMP MIB a group can be thought of as a set of MIB objects Ini tially the database contains a single table or group corresponding to the raw MIB object data As data is aggregated a set of roll up tables or groups is created Creating a New Report 10 7 mh i E m z g ley Q 10 TRENDbuild Note EXAMPLE If the only group of statistics being collected based on the MIB II file was the ifEntry table once rollup had occurred you would
232. e count of input samples used to compute the row Forecasting TRENDsum also enables you to forecast resource usage for 30 60 and 90 day periods and compute the number of days until a specified usage threshold is reached days to threshold forecasting The 30 60 and 90 day forecasts answer the question Where will the property be ing forecast e g Utilization be in 30 60 or 90 days The days to threshold DTT forecast answers the question In how many days will the property being forecast be at the threshold value Forecasts are used in the Capacity Planning and Forecast reports supplied with the TREND ReportPacks Figure 9 3 shows a sample LAN Connectivity Capacity Plan ning report In this report TRENDsum formulas for computing 30 60 and 90 day forecasts are used Forecasting 9 17 9 Data Aggregation and Manipulation gt Designed network planners network OaesntTaten LAN Connectivity Capacity een esa details the pasali alamin LANs indicating opportunities for load balancing to improve service R E N D Capacity Planning levels without additional investment Drill down folders provide detail for each selected interface aaa Eee eru Figure 9 3 LAN Connectivity Capacity Planning Report In the Capacity Planning and Forecast reports all TREND forecasting is based on 95th percentile values computed on daily summary input tables that TRENDsum has created In these reports the 95t
233. e currently selected managed object item 4 To see the definition of a particular managed object click it in the MIB map If you click the name of a group such as mib II the managed object information area is blank Eligible fields displayed for each managed object are Label Object ID Oid Object Type Access Status 6 4 The name of the selected managed object The path through the MIB tree to the selected variable represented as a string of numbers separated by periods The last value in the variable s object identifier is the selected variable s position in its table The SNMP data type syntax assigned to the selected variable The Object Type specifies the database storage definition The aggregation process references the object type to determine the method for processing See Table 6 1 Valid types are INTEGER OCTET STRING OBJECT ID GAUGE COUNTER COUNTER32 COUNTER64 TIME TICKS OTHER The means by which a variable may be accessed Valid access modes are read only read write not accessible The implementation support required for the variable Valid access modes are mandatory current optional depreciated obsolete Main Display and Menu System TREND 3 6 User s Guide Default Value The default value for the variable if one is specified in the MIB Description A pop up window containing the description for the selected variable Click on the Description button to display Figure 6 3
234. e data is likely to change in the interval Other factors affecting polling frequency decisions include whether you are likely to miss anything if the polls are farther apart the load placed on both the polled and polling systems and on the network by the data collection especially if target nodes are at remote sites and on the amount of database space you have for storing the data Remember that changing polling frequency from 10 to 5 minutes doubles the volume of data collected Typically performance statistics are collected every 10 or 15 minutes while very large tables such as RMON host and matrix are collected hourly N e o o ap o D r 5 Additionally keep in mind that TREND is not intended for reviewing real time cur rent statistics to solve an immediate problem That is the role of dashboard or real time applications which collect data at a high frequency rate generally measured in seconds display the most recent data and discard it when new data arrives or soon afterwards Such real time applications are focused on troubleshooting with a few Some Considerations When Organizing Polling 7 15 7 Collect Data key statistics collected from a few devices TREND polling is aimed at providing a baseline to support the troubleshooting activity and at identifying patterns generally occurring over hours days and weeks Viewing SNMP Data 7 16 TREND lets you display the data you ve collected on your scre
235. e file Figure 6 13 illustrates a basic MIB scheme Definitions statem ent Comments Top of MIB tree internet MIB header directory experim ental private Block 2 enterprises Block 3 Block 1 Block 4 Figure 6 13 Viewing a MIB in Distinct Blocks 4 Using the TREND interface select File gt Load Mib If this file loads properly you have confirmed that this portion of the MIB is in compliance with the rules 6 18 Preparing MIBs for MIBwalker TREND 3 6 User s Guide 5 Now add the Block 2 text to the file and repeat the procedure Do this for the remaining blocks in the file By process of elimination you should be able to isolate the block s that contain rule violations MIBlets MIBlets can provide a data source for custom designed reports A MIBlet is a composite set of MIB variables selected for data collection MIBlets are useful in that they allow you to focus your collection on data actually needed for reports This increases polling efficiency speeds up data collection and reduces the amount of database space used More importantly MIBlets allow you to generate a single report that includes variables from separate MIB tables o ius 2 x D Utilizing 64 Bit Values SNMPv2 introduced 64 bit counters however SNMPv1 agents only support 32 bit counters One common way that SNMPv1 agents return the higher precision values stored in 64 bit counters is to overlay a pair of 32 bit coun
236. e first line in the report lists the names of the fields in the file The names which can contain embedded spaces are also separated by tab char acters Here is sample data from a TRENDsheet report as it looks in an export file Received TS Dat Received TS Tim Packets Broadcasts Multicasts Collisions Octets m 10 17 94 08 15 00 1122 1003 H 84 0 559808 2 10 17 94 08 10 00 1134 1009 87 0 565696 10 17 94 08 05 00 1078 959 85 0 528192 10 17 94 08 00 00 1158 1026 Exporting Data 11 19 11 TRENDsheet 11 20 91 10 17 94 92 10 17 94 84 10 17 94 88 10 17 94 87 10 17 94 88 10 17 94 90 07 07 07 07 07 07 554 50 45 40 35 30 00 00 00 00 00 00 567104 1195 597760 1126 560576 1019 496704 1134 563200 1205 598336 1130 563136 1069 1007 897 1007 1082 1006 If you export a field with a data type of datetime such as the Received TS column its values are divided into two fields a date field with the format MM DD YY and a time field with the format hh mm ss For example the field called Received TS in the report shown in the TRENDsheet window is written to the export file as the fields Received TS Date and Received TS Time Numeric fields are written to the file using the same precision with which they are displayed in the report If you look at the export file i
237. e named cache to be visible For example to add the named cache with the name of the cache equal to keycache and the size of cache equal to 1 MB enter the following commands isql Usa P 1 gt sp_cacheconfig keycache 1M 2 gt go Remember to shut the Sybase SQL Server down and restart it See Cache Pools on page C 10 on how to configure the cache you just created Sybase Configuration Parameters TREND 3 6 User s Guide Binding the Individual Key Tables to the Named Cache The next step is to bind the key tables to the named cache You can select the key tables of your most frequently used data tables key tables of your largest data tables or any key table that would help the performance by reducing disk access Enter the following commands 9 O Si e 3 3 O D el 3 Ce isql Usa P 1 use dpipe db 2 gt go Use the following command to return the names of all data tables with their corresponding key table names 1 gt select dsi data table dsi key table from dsi key tables 2 gt go Once you have identified the names of all the key tables that you wish to cache repeat the following command for each key table name 1 gt sp_bindcache keycache dpipe_db dsi_dpipe key_tabl_nam 2 gt go where key_tabl_nam is the name of the key table to cache Note that you must enter the quotes For example to bind the key tables Kib ii ifentry Ky interface perf and K5 frcircuitentry to t
238. e number is retained if the name is truncated since it is truncated from left to right o W z o Troubleshooting a MIB Current MIB parser technology used in MIBwalker requires strict adherence to typographical and syntax rules as described in Preparing MIBs for MIBwalker on page 6 11 If the parser rejects the MIB a popup box is output by the parser with one or more error messages These messages identify an error message the line where the rule violation is located and a reason for the error See Figure 6 12 mibwalker Figure 6 12 Example of an error message The rejected MIB file must then be modified to correct these errors Make the necessary changes and reload the MIB file Preparing MIBs for MIBwalker 6 17 6 MIBwalker Recommended Procedure DeskTalk recommends the following procedure for troubleshooting a MIB that will not load 1 Make a copy of the rejected MIB and rename the file 2 Open a text editor and examine the MIB contents 3 View the entire MIB in terms of separate vendor objects and logical blocks The objective is to first divide the MIB into manageable and proportionate blocks then to diagnose each block until you have isolated the area that caused the MIB to be rejected a b C Retain text from the beginning of the file down to the end of Block 1 Cut all text from the beginning of Block 2 to the end of Block 4 Retain the END statement Save th
239. e of allowing data to be aggregated and aged according to system wide defaults or setting specific values on specific tables To set or view the system wide default values follow these steps 1 Select Set Default Storage Time from the Data Manager s File menu to bring up a window that shows you the current default settings Default Storage Time Figure 2 9 The Default Storage Time Window 2 Set the desired value for each field To see the choices available click on the downarrow button by the field you are setting The fields to be set are Roll Up Data Automatically Determines if sample data in a data table will be aggregated Keep type Data Changing System wide Rollup and Aging Defaults TREND 3 6 User s Guide The period of time data in all tables of the type of table indicated will be kept When a row has been in the table lt save period days gt it will be deleted from the table 3 Click OK to save the newly set values 4 Click Reset to re display the values set the last time the OK button was clicked 5 Click Cancel to exit Only the user trendadm can change the system default values Other users will be able to change the values displayed in the window but will not be able to make the new values effective because the Apply button will not work it is grayed out in the display N E m 032 ag no 3 o 9 Mie o fo Q ERE P Ly e D Changing Rollup on a Specific Table
240. e operation is not performed You may consider the 2 00 sample to be a false max in that the sample covers twice as much time as the 1 15 sample All rate counter expressions such as utilization in 3 4 x TRENDsum files need to be adjusted in 3 5 1 because the avg function in 3 5 1 does not divide by delta_time However avg calculations are more problematic than min and max calculations due to the time weighting issue Consider the example in the table above TRENDsum 3 5 1 calculates the avg function over the 60 minutes as 107 4500 or 0 023777 while TRENDsum 3 4 x calculates the function as 0 09665 4 or 0 02416 The discrepancy results from the weight given the 30 minute sample To remove the discrepancy completely and maintain compatibility in 3 4 x and 3 5 1 avg calculations you can take one of two actions 1 Expressions such as the following in the 3 4 x TRENDsum files column inutilization ifinoctets018 8 ifspeed013 100 avg max can be changed as follows in 3 5 1 TRENDsum files column inutilization ifinoctets018 8 ifspeed013 delta time004 100 0 max column inutilization ifinoctets018 8 ifspeed013 100 0 avg A 88 Then you need to adjust the expressions in the qgr qss and gos files by multiplying them by ta samples delta time 2 Alternatively you can replace the avg function in the 3 5 1 TRENDsum files with the wav function which weights the values based on delta time in 3 5 1 If you choo
241. e table In addition to aggregating raw data into rate tables TREND rolls up rate data into hourly daily and weekly tables The process that occurs between each level of rollup is virtually identical In each case hourly daily weekly and monthly four rows are created in the corresponding table These four rows correspond to the aggregation that has been performed on the data average avg minimum min maximum max or total tot Hourly tables are calculated from rate tables Daily tables and weekly tables are calculated from hourly and daily tables respectively using the total row from that table Monthly ta bles are calculated from daily tables About Data and Data Tables in TREND TREND 3 6 User s Guide Statistic Counters Gauges Other Types Average avg Sum of values divided by total delta time i e units per second Average value of samples in interval Value from the latest timestamp row in interval Minimum min Value from sam ple with mini mum rate in units per second Minimum value of samples in interval Value from the latest timestamp row in interval Maximum max Value from sam ple with maxi mum rate in units per second Maximum value of samples in interval Value from the latest timestamp row in interval Total tot Total of values for all samples in interval Average value of samples in interval Value from t
242. e the new variable name such as dump database master to MASTER DEVICE file master dmp with init B Use diskfiles defined at run time 1 Change S TAPE DEVICE on the respective dump command to be the full path name such as dump database master to Full Pathname Trend Backups master dmp with init If you desire to have multiple backup copies available at any instance you may create these files with a date designation by using the syntax dump database master to Full Pathname Trend Backups master date t y m d H M dmp with init 2 Remove the segment starting with ile and ending with the double quote from the respective dump command If you modify the dump command for the master database it is important to keep the with init segment Second make sure that the proper access permissions exist on the S YBASE install directory for the script to create the SDSQUERY log ga 0 3o 0 g TO D D O o 3 Backing Up the SQL Server Installation B 21 B Database Administration B 22 Third review the script s log file SYBASE install SDSQUERY 1og ona reg ular basis to identify any error situations that may arise for a specific database Fourth verify the backup server exists and is running before invoking the script See Backing Up the SQL Server Installation on page B 18 Note If all the databases result in a tape volume change then the system will be waiting
243. e_number Is the page of the report shown in this file A companion description file is also created that contains information about gif file contents The description file is named xyz identifier dsc where identifier Is the same identifier used in the gif file name TRENDprint File Naming The extension to the output file name determines the format in which the report file is saved ps gif or bmp 15 8 Output to a File TREND 3 6 User s Guide TRENDprint can accept only one value for the name of the output file or files gen erated by a scheduled report request If you define a TRENDprint drilldown se quence that sequence will generate at least two different reports and probably more Of those the 2nd through nth report will have as many output files as there are de vice instance pairs identified in the exception report step of the drilldown sequence For example if a drilldown sequence includes three drilldown reports and the ex ception report identifies ten objects to report in more detail 31 reports are generated All 31 reports will have the same name For example the name xyz identifi er page number gi is assigned to 31 files Moreover 31 dsc files are created at the same time ol D Dm 3 wn 7 TI a o Output to a File 15 9 15 Report Scheduler TREND 15 10 TREND 16 RMON Support RMON stands for Remote MONitoring a means to observe traffic o
244. earch to scroll through the entries that match the search string Click Clear to clear the search string specified in the Search box Click on the desired report name to select it The selected name is high lighted Then click View Report to create the report Alternatively you can launch the report by double clicking on its name in the Reports list box Selecting Devices and Elements The default is to include all devices and elements in a report but you can restrict re port contents to selected devices and elements To do this 1 Click on the Select Devices button before you launch the report The Devices window appears Devices F No Filter Iz __View Report L Cancel Figure 14 9 Devices Window 2 Select the device on which you want to report from the entries in the Select 14 14 Device list box All Devices is the default Selecting Devices and Elements TREND 3 6 User s Guide To make your job of choosing a device easier you can restrict the entries on the device list by selecting a filter in the Device Filter list box You can choose one of the following filters Filter Retrieves No Filter No filtering criteria is applied to the list the default Device Name The Match Value window appears if you select this filter Device Name Partial Word BB Clear OK Cancel Figure 14 10 Match Value Window for Device Name You can specify a device n
245. ed For example a 3 4 x expression such as the following column inutilization ifinoctets018 8 ifspeed013 100 avg max needs to be changed as follows in 3 5 1 TRENDsum files to maintain 3 4 x compatibility column inutilization ifinoctets018 8 ifspeed013 delta time004 100 avg max You should consider the possibility of obtaining a false max or min if you do not divide a counter expression by delta time in 3 5 1 For example assume the poll rate is every 15 minutes and a poll at 1 45 is missed The following table shows the sample values Value ofthe Rate ized Value Function Time Counter value delta time of the Delta time Expression Counter Expression 1 00 20 0 02222 900 gt 1 15 25 0 02777 900 1 30 22 0 02444 900 y 2 00 40 0 02222 1800 9 max 40 0 02777 1800 tot 107 0 09665 4500 avg 26 75 0 02416 0 09665 4 1125 wav 29 40 0 02377 107 4500 900 Computed as the total delta_time 4500 divided by the number of samples taken 4 1125 A 87 TRENDsum Computed as the total delta_time 4500 divided by the number of samples taken plus the sample for the missing period 5 900 With 3 4 x TRENDsum processing the max function selects the 1 15 sample because of the hidden divide by delta_time operation that 3 4 x performs on the counter expression With 3 5 1 the max function selects the 2 00 sample because the hidden divid
246. eee eee 14 1 Invoking the Report Launcher 00 eee eee eee eee 14 2 Selecting the Database ee 14 5 Selecting the Report File Directory 22200005 14 7 Designating the Top Level Report Directory 14 8 Selecting a Report 0 0 0 cee tt eens 14 11 Selecting Devices and Elements 0 202000eee 14 14 Specifying the Report Time Period 0 eee eee 14 18 Generating the Report 14 20 Refreshing the Report List llle 14 20 Report Scheduler eeeeeeeeeeeeeeeennne 15 1 Printing a Repon cs naa AA NG Cap na te peels age Hao hEPEGVAISV 15 1 Menu Command Summary sasaaa aaua aaaea 15 2 Scheduled Reports sierosa anene a eee eae 15 3 Report Instances 5 deme du een RERE EERTE REES 15 4 Adding New Reports 020 cece teen teens 15 4 Report Chaining Adding Drilldown Reports 15 4 Defining Scheduled Reports 00 ee eee eee eee 15 5 Defining Report Instances isses 15 5 Setting Report Parameters llle 15 6 Qutputto a File cem tet meh ent ue eee eds 15 7 Example de es deer ta eet need eO dub 15 8 TRENDprint File Naming 2 00 BIS 15 8 xi Contents Xii RMON S pport c cec ERE RD E Ram PRE ALAN AA 16 1 RMON Group Support 0 0 00 cee nh 16 2 Tables and Data Tables 0 0 0 cece eee eee 16 3 Polling Fredguencyc oes rete epe DR RR ice ERAT Ri 16 4 Histor
247. efore preclude the need for the standard topN group col lection In addition since they are not appropriate for the long term performance profiling done by TREND alarm filter capture and event groups are not polled Furthermore because TREND does not distinguish between the Source Destination and Destination Source versions of the Matrix table the table in the database can be sorted and read in any order it is not necessary to collect both versions of the table from the probe Except for etherStatsEntry the supported RMON tables use structures which are unlike the regular SNMP table structure Most SNMP tables are simple lists of variables where each instance of a table is one copy of the set of variables e g each instance collects the variables for one interface on a router Each instance of RMON table is a set of instances grouped by common attributes For example a history table instance may consist of a set of history table variable lists Each of these lists con tains the values of variables collected for a specific amount of time e g 50 instances or buckets of the history table variables each representing 30 seconds of collected data for a total of 25 minutes Note Do not load an RMON MIB into MIBwalker and define periodic data collection through the MIB map and MIBwalker If you do invalid data will be collected Tables and Data Tables The RMON History Host and Matrix groups contain control tables and data ta
248. egation Type Cancel dirty devil gt Javerage zj Sort Order Figure 13 5 Select Key Window Select a key by scrolling down and clicking on the appropriately numbered row Use the scroll bar to see all the available keys In the Select Data Set window click Add Line Continue to select keys and aggre gation types and add statistics until all statistics have been specified for the graph Creating a New Grade of Service Report TREND 3 6 User s Guide EXAMPLE Creating an Interface Health Report An Interface Health report is defined with the following parameters and weighting Kar 5 Q o Q D w D S 7 m Parameter Weighting Excellent Good Fair Poor Utilization 50 lt 25 25 35 36 50 51 75 Input Errors 10 lt 1 1 2 2 3 3 4 Output Errors 10 lt 1 1 2 2 3 3 4 Receive Pkt Discards 15 lt 1 1 3 3 5 5 10 Sent Pkt Discards 15 lt 1 1 3 3 5 5 10 Having selected the five statistics parameters in TRENDbuild call up the first sta tistic to be added to the GOS report e g Percent Utilization in the Select Statistic box of the Grade of Service Select Data Set window Click Select Key to invoke the Grade of Service Select Key window and set the device key and aggregation type desired see TRENDbuild on page 10 1 for ad ditional details about TRENDbuild and the Select Key function On the View pull down menu in the Scor
249. el for debug output This option is primarily for development purposes Waits for the specified amount of time in seconds for activity when a connection is committed to the database The default is 1800 seconds Lists the command line options help Displays the current version and build number of the utility log_backup log_backup Backs up the trend log file every day to a new file whose name is based on the day of the week log backup Description An entry to execute this utility is stored automatically in the trendtimer sched file during the TREND installation procedure The back up file names created by this command are trend log monday trend log tuesday trend log wednesday trend log thursday trend log friday trend log saturday trend log sunday gt D v o Q o o lor launch Ipr launch Prints TREND reports configured for automatic printing lpr launch d debug level i interval V Description Ipr launch prints the reports configured for automatic printing by Report Scheduler Ipr launch is normally invoked by trendtimer once a day to run daily reports once a week to run weekly reports and once a month to run monthly reports Daily reports use the midnight to midnight period of the previous day as their time period Weekly reports use the most recent Sunday midnight to Sunday midnight period Monthly reports use the period from midnight on t
250. elect Data Tables window Define Qualifications window Select Columns window Define Expressions window Each of these elements serves a specific purpose some essential others optional The function of each element is described below Elements of TRENDdbview 22 3 22 TRENDdbview TRENDdbview Main Window 22 4 The TRENDdbview main window appears when you launch TRENDdbview untitled joisco100 a PUTAS BE LS Figure 22 1 TRENDdbview Main Window The TRENDdbview main window is the connection to each of the other elements and is the only path by which they can be reached In addition the TRENDdbview main window contains a File pull down menu with the New Open Save Save As Run Change Database Exit and About commands Once you have defined a view you will use the TRENDdbview main window to de termine the name under which the view will be saved The procedures for accom plishing these actions are described later in this chapter Elements of TRENDdbview TREND 3 6 User s Guide Select Data Tables Window You access the Select Data Tables window by clicking the Select Data Tables button in the TRENDdbview main window The buttons and fields in the Select Data Ta bles window allow you to define the data and key tables that make up your view and the key table to be associated with it d cisco100 View MISHA0397 A C 2 day cisco100 Isystem perf b cisco100 View Mish
251. elect Node window Specifying Data to be Included in the Report 11 9 ok ok J m z 7 T D et 11 TRENDsheet 11 10 Select Node Lok Select ai nodes Aggregation Type total Node airty devil Sort Order OK Cancel Ke Description INN E EE Figure 11 4 Select Node Window The Node box in the window lists all nodes that have data in the table being reported on in the database It also gives you the option all nodes which is the default If you choose all nodes you automatically see all keys for all nodes in the report The Aggregation Type box is used to select the minimum maximum total and av erage of all types of aggregation for rolled up tables not raw tables Click the Sort Order button to bring up the Sort Order window which allows you to select the order that the TRENDsheet data is sorted by see Sorting the Data in the Report on page 11 14 for more detailed information on how to sort If you select a particular node in the Node box by clicking on its name the Key box will display a list of all keys in the table for the selected node Select the desired key from the Key box When you have chosen the node key pair you want click the OK button to have the report only display values for that pair Click Cancel to quit out of the window to without changing the current node key pair values The node key pair you choose will be saved as part of the repo
252. element property table_key level for the time period to produce four summary tables hourly daily weekly and monthly as follows Rate Table Hourly Table Daily Table Weekly Table Monthly Table 9 8 Data Cleaning TREND 3 6 User s Guide During this roll up process the following four rows are created in each summary table for each element property table_key 1 2 3 4 A row with the minimum value for the time period dsi agg type 2 A row with the maximum value for the time period dsi agg type 3 A row with the total value for the time period dsi agg type 4 A row with the average value for the time period dsi agg type 5 For all data types except counter and gauge data TRENDit stores the last value collected for the item as that item s value in each summary row because summaries of data types other than counters or gauges are not sensible For example a device s IP address is the same in each collection sample averaging multiple occurrences of the IP address is a meaningless operation The value of the delta time column in each summary row is the sum of the delta time values in the input rows The summary rows are created as follows for counter data which has been transformed into delta data and gauge data Summary Counter Delta Value Gauge Value Computed Row Computed As As Minimum Minimum of the counter data Minimum of the input values divided by their asso va
253. eline output The full name of the destination table will be SDnn plus this name where nn is the number of days used for the value of the y option You may not use both the a and e options The specification for the grouping by variable see TRENDsum By Variables on page A 81 which has the format alias Jcolumn or keyword where column keyword keyid ta_period hour day week month quarter year day_of_week day_of_week_by_hour hour_of_day Is the name of a column in the source table Has one of the following values Group by the column named dsi_key_id_ Group by ta_period Group by hour Group by day Group by week Group by month Group by quarter Group by year Group by day of week for baseline table Group by day of week by hour for baseline table Group by hour of day for baseline table TRENDsum C The specification for a column in the output table where column spec has the format alias column expr statisticl statisticn where statistic at least one of the following avg Average ent Count of samples dtt value Number of days until value is reached where value is the threshold value the user specifies is 30 Projected value in 30 days f60 Projected value in 60 days f90 Projected value in 90 days mad Delta time when the maximum value occurred mat Time when the maximum value occurred in the source table max Maximum value med
254. eline value because 42 days of data is used to compute it is sig nificantly more refined than a value where only one day s samplings are sum marized to compute the baseline statistic for day 1 5 When the number of days in the rolling baseline period is reached processing continues in the fashion described above However the rolling baseline value that is computed from the preceding 42 days of samples is appended to the table The existing baseline is no longer replaced Thus rows are added to the rolling baseline summary table as follows Day Baseline Statistic Computed from 1 ay Days 1 42 2 az Days 2 43 3 a3 Days 3 44 4 ay Days 4 45 n an Days n n 42 The Executive Summary reports in the TREND ReportPacks compare a resource usage for each hour of the day to an hourly baseline that has been computed for the resource The concept for computing an hourly baseline is the same as the one for computing a daily baseline However the hourly baseline table has one row for each hour of each day computed for a sliding 42 day baseline period as described above 9 22 Baselines TREND 3 6 User s Guide Figure 9 4 shows an excerpt from the Frame Relay Executive Summary report This report provides CFOs CIOs and other managers an overview ERA BA ea eS Frame Relay of the performance of corporate Frame Relay networks Each chart T R E N D shows key metrics aggregated for all PVCs Key
255. ell as specific SNMP and RMON mod ules TREND is a complete application that any member of the network staff can use to collect appropriate data and produce meaningful reports without networking or database expertise Its extensible architecture also provides an application toolkit that the network expert can use to extend TRENDS functionality without program ming or SQL scripting TREND is a true client server scalable SNMP application for enterprise network management TRENDS report building capability allows system operators and net work managers to create custom reports and report libraries The RDBMS data re pository can be centralized or distributed across multiple databases while the report builder and table administration tool eliminate the need for RDBMS expertise 1 1 L ET pia fo Q j 7 E 3 apa 1 Introduction to TREND TREND Overview As shown in Figure 1 1 TREND Applications and Processes TREND is comprised of user visible application modules and user transparent processes that carry out the required tasks pamm CD PaesenTaTON INFORMATION DATABASE MGMT TS Collect Data TRENDbuild TRENDsheet Data Manager Report Scheduler TRENDweb Report Launcher onfigure Users and TRENDgraph Grade of Service lt gt End User Applications Data Repository dpipe db Background Applications
256. else and case constructions that you can use to conditionally process input to check for this condition Appendix A Man Pages describes the TRENDstep language in detail Another way to avoid a potential divide by zero problem is to use ifspeed 1 in the formula that computes utilization The 1 increases ifspeed a negligible amount Creating an Other Category for RMON2 Data 9 12 RMONQ agents provide statistics on a large number of individual communication protocols that various device interfaces use In cases where you are only interested in the statistics for a few protocols and you want to group the statistics for all other protocols into an Other category you can use TRENDstep to conditionally The Base Table TREND 3 6 User s Guide process each column of data in the input row create a new column called Other in the output table and place the sum of the data for all other protocols in that column In this case the output table contains the same number of rows as the input table but the columns are different The section on TRENDstep in Appendix A Man Pages gives an example of this TRENDstep function Assigning Grades to Ranges of Input Data The TRENDstep language also enables you to conditionally examine a selected column value in each row of input data determine if the value falls within a particular range and assign a grade Grading is used to produce the Grade of Service GOS stacked bar
257. en This can be data recently polled from the target node or it can be data previously stored in the data base in raw or aggregated form Data is viewed via TREND s spreadsheet style re porting application TRENDsheet or line graph style report application TRENDegraph Normally the TRENDbuild application is used to select what data to include in a report You can quickly create a simple report of the data in a database table by invoking TRENDsheet or TRENDgraph from the Tools Menu in the Collect Data window Select the collection request definition that polls for the table you want to look at Select the type of display by clicking on the arrow in the TRENDsheet or TRENDgraph button This will display a pop up menu of the available types of data Click on the type value you desire The following options are available View Raw Data Displays raw sample data View Rate Data Displays inter sample delta values for counters and original values for all other fields View Hourly Data Displays hourly summary information View Daily Data Displays daily summary information View Weekly Data Displays weekly summary information Viewing SNMP Data TREND 3 6 User s Guide These default reports include all statistics in the selected table without aliases or ex pressions As with all TREND reports a query file is created These reports can therefore be manipulated like any other graph or table report see TRENDsheet on page 11 1 an
258. ent TRENDproc block Second Level TRENDsum runs against the data tables created by trendit Again you can create multiple blocks each launching one TRENDsum against one set of data tables Third Level TRENDexporter exports the aggregated data to the primary TREND database Fourth Level A series of reports is run against the new data Having established more or less what you want to accomplish you can define your TRENDproc file to execute the proper commands In this case you might create a single block for each satellite database you wanted to run trendit and TRENDsum against Each block would have three commands Run trendit 2 Run TRENDsum 3 Run TRENDexporter If your TREND network had six satellite servers this would mean creating six sep arate nowait blocks one for each server You would then create a seventh block to run the desired reports against the primary database 21 8 Creating and Applying a TRENDproc TREND 3 6 User s Guide Things to Avoid bs mh m 2 jo o e 2 You need to avoid trying to access the same database table at the same time For ex ample you would not want to have two TRENDsum processes trying to summarize data from the Rib ii ifentry table at the same time TRENDproc File Syntax A TRENDproc file is a text file with the extension pro It must contain at least one block and the blocks must adhere to a particular format A block begins with statement begi
259. ented in the graph To change the report titles select the Graph Title option under the View button to bring up the Graph Title window Graph Title Dx TURIN TIT a IET EET Title1 SINA SESE S Titlez Houry III 0K Conce Figure 12 10 Graph Title Window The window displays the current report title values By default Titlel is the name of the report file without the qgr extension Change the titles by typing in new values Click the OK button to activate the new titles Click on Cancel to quit out of the window without changing the original values mah N J m 2 O Q o o T Setting Report Display Parameters 12 19 12 TRENDgraph Adding Grid Lines 12 20 Grid lines extend the Y Axis hash marks across the graph area to make it easier to determine the actual values of data points in the graph The user can turn grid lines on and off and change their color Grid lines will appear in a printout of the report Keep in mind how they will look on paper while you set them up especially if you have a color screen and a monochrome printer To change the grid line settings select the Grid option under the View menu This will bring up the Grid window Show Grid Grid Color emm Black OK Cancel Figure 12 11 Grid Window The window shows the current grid display settings To turn grid lines on or off Click the down arrow of the Show Grid box and select Yes to turn t
260. epted However you can specify a lower level summarization level such as hour_of_day to get 24 summary rows of data per day You can omit the day by variable if you specify the y option In this case day is automatic When using the optional day of week or day of week by hour summarization level it is strongly recommended that the number of days for the baseline period be evenly divisible by seven otherwise some summary rows will summarize a different number of days than others For example one summary row might be for seven Mondays while another row might be for six Tuesdays Input File Description The preferred method for providing parameters to TRENDsum is to use an input file The input file includes statements with keywords that correspond to specific options on the command line The following keywords introduce these statements Keyword Corresponding Command Line Option source_table T t destination table column C by_variable b baseline days y A 83 gt D 5 v D e D 7 TRENDsum Keyword Corresponding Command Line Option null percentage n replace table r Use of this option is very unusual If omitted TRENDsum appends rows to the destination table which is typically what you want first day W The by variable source table and destination table keywords from the 3 5 x and earlier versions are still available Please n
261. equests TREND 3 6 User s Guide the table to highlight that request s row then click on the Edit Entry button alterna tively you can double click on the line This will bring up the Collect Data Periodi cally window Modify the polling parameters as described in Setting Up New Polling Requests on page 7 4 The new parameters will be displayed in the table Deleting Request Definitions Existing requests can be deleted by the user who created them as shown in the User column or by the user trendadm To delete a request click on the request s line in the table This will highlight that request s row Next click on the Delete button The row will be removed from the table Exiting the Collect Data Window To quit out of the Collect Data window select Exit from the File Menu Some Considerations When Organizing Polling N e o o o D r 5 TREND makes it very easy to setup data polling that will collect a great deal of in formation from a large number of nodes even from just a few polling requests It is worth taking a moment to plan your polling strategy Polling Groups TREND s ability to poll user defined groups of devices from a single data collection request is one of the application s most powerful features When defining device Some Considerations When Organizing Polling 7 13 7 Collect Data 7 14 groups there are various strategies that can be used to maximize efficiency and sim
262. er To pull the data from the satellite server onto the central server That is run TRENDcopy on the central server according to the scheduling information given in the trendtimer sched file residing on the central server Either method is effective To implement one of these methods place the following command in the appropriate trendtimer sched file trendcopy S source server S target server f R This command copies raw and delta tables serially which may result in excessive run times See Performance Tuning Issues on page C 1 To improve throughput and reduce run times you can execute multiple TRENDcopy processes with the t option concurrently For example trendcopy S source server S target server t table name This command copies only the table specified by table name to the central server Include a trendcopy command line with the t option in trendtimer sched for each delta table you want to copy If your servers are set up in a UNIX environment you can download the following scripts from the DeskTalk Systems FTP site and use them as an alternative to the trendcopy command TrendCopy sh TrendCopyLoop sh To use these scripts Using TRENDcopy TREND 3 6 User s Guide 1 Place the downloaded scripts in the SDPIPE HOME bin directory on the server where you want to execute TRENDcopy 2 Replace the trendcopy commands in the appropriate trendtimer sched file with one of the following commands
263. ers long cause an error mes sage TRENDit is unable to construct the temporary work table names as lt table name gt _ lt processid gt lt date gt Sybase limits the string to its maximum Sets the minimum value against which all gauges and counter deltas for each RMON host and matrix table row will be compared to determine whether that row should be rolled up If the values of ALL these fields in a given row are less than or equal to the v value the row is not in cluded in the rollup Although this can lead to gaps in the rollup data of the RMON host and matrix tables these gaps are periods when these hosts or pairs of hosts had no activity By omitting these no activity rows considerable space can be conserved in the rollup tables The de fault value is 0 Displays the version stamp for TRENDit Only rollup occurs between a specified time period If using w 1 3 the rollup process only rolls up data from Monday to Wednesday this only affects the daily table Valid values are 0 Sunday 1 Monday 2 Tuesday 3 Wednesday 4 Thursday 5 Friday 6 Saturday A 41 gt D v o Q o o TRENDit Z The percentage difference between the sysuptimes returned with polled data and the local received timestamps of two consecutive raw data samples which TRENDit will allow and still consider the data for the interval valid If the percentage difference between the changes in these two values is greater than the value
264. es Window The Assign Statistic Aliases window provides Selection of statistics that will be included in expressions to be graphed upper left Mathematical operators to be used in expressions beneath Select Statistics box Creation of expressions from statistics and operators bottom Naming and adding deleting expressions and Q 0 2 9 Listing of Active Expressions assigned upper right EXAMPLE Creating an Input Utilization Report vs an Output Utilization Report To specify input utilization for a device first click in the New Expressions box and enter a name for the expressions to be created e g Input Utilization Click Add to place the expression in the Active Expressions box Click the new expressions name in the Active Expressions box to continue make sure it is highlighted Select the following combination of statistics and operators to create the expression for Input Utilization at the bottom of the Assign Statistic Aliases window GfInOctets 8 ifSpeed delta time 100 0 N J m Z O Q o o T Creating a New Graph 12 7 12 TRENDgraph It is important that the left and right parentheses are paired Click Add to View to save the expression Create an expression for Output Utilization in the New Expressions box e g Output_Utilization Click Add Select Output Utilization from the Active Expres sions box Enter the following formula in the Output_Utilization input box Ci
265. es the RMON DataPipe TRENDcopy Describes TRENDcopy which allows you to copy informa tion from one database to another TRENDlabel Describes the command line driven TRENDlabel applica tion which you can use to populate specific columns of key tables TRENDexporter Describes the TREND exporting module which adds versa tility to data export TRENDproc Describes TRENDproc which allows you to streamline the reporting process TRENDdbview Describes TRENDdbview which generates SQL database views Man Pages Contains the latest Man Pages published by DeskTalk Systems Inc TREND 3 6 User s Guide Appendix B Appendix C Appendix D Appendix E Appendix F Database Administration Provides the TREND Administrator trendadm with useful information for maintaining the TREND database Performance Tuning Issues Discusses how to improve TREND performance Setting Up a Satellite Server Environment Discusses considerations for processing performed on the central server and satellite servers Discovery Considerations Describes how TREND stores nodes and how you can mod ify Discovery files Relocating the TREND Database Explains how to move the TREND database from one host to another Related Documents The following documents may also be useful XXix Preface TREND Manuals TREND Installation Guide TREND User s Guide TREND ReportPack Guide TRENDweb Installation and Admin
266. espe cially if you have a color screen and a monochrome printer Besides the shades of gray available only line type and width will be distinguishable on a monochrome printout Changing Point Attributes On a line graph points show where actual data points fall on the plot line They make it easy to see how far apart your data samples occur The Type box is a list of the available point types e g dot or box Select the point type for the current line by clicking on the Type box s down arrow button then click ing on the desired type Select None if you do not want points included on the current line The Color box is a list of the available point colors Select the point color for the current line by clicking on the Color box s down arrow button then clicking on the desired color Use the scroll bar to see all the available colors The Width box is a list of the available point widths in points Select the point width for the current line by clicking on the Width box s down arrow button then clicking on the desired width Use the scroll bar to see all the available point widths Defining Fill Attributes Area and bar chart fill patterns and colors can be changed in the Fill Attributes area of the Select Data Set window N J m Z O Q o o T Setting Report Display Parameters 12 15 12 TRENDgraph Select the fill pattern by clicking on the pull down arrow on the Type box then click ing on t
267. ession you just named click on its name in the Active Expressions box Click in the wide box near the bottom of the screen to enter the expression 4 Build the expression by clicking on the statistic names and arithmetic operator buttons and entering numbers as necessary You are responsible for correct syntax in the expression TRENDbuild will not check your work Note Values added to an expression by clicking on statistic names or operator buttons will be added to the end of the expression regardless of the cur Creating a New Report 10 13 10 TRENDbuild sor position The use of parentheses is recommended for all expressions Ensure that right and left parentheses match for example ifInOc tets 8 fSpeed 100 0 5 When the expression is defined the way you want it click on the Add to View button to add it to the report definition Do this before beginning to define another expression You can modify an existing expression Click on its name in the Active Expressions window You will see its name appear below the arithmetic operator buttons and the expression text will appear in the box at the bottom of the window You can delete the entire expression by clicking on the Clear button Make any changes to the ex pression text then click on the Add to View window to apply the changes When you finish defining expressions quit the window by clicking on the Cancel button The names of the expressions will appear in
268. et for this table in the database db delete data t mib II ifEntry The following command deletes data from the TREND database tables based on time period db delete data m ta period Note To enhance performance in TREND installations using Sybase make the following change in the 70DPIPE HOME lib trendtimer sched file Change 24 00 5 00 DPIPE HOME bin db delete data to 24 0045 00 DPIPE HOME bin db delete data 24 00 24 00 5 DPIPE HOME bin db delete data c X u 2 U Sa where X is the result of the number of system processors multiplied by two child processors Systems having additional resources could incre ment the number of child processors in the equation db_delete_data The starting time is moved to an earlier time 5 minutes after midnight prior to the early morning aggregations of ReportPacks so that tables are processed that is reduced in size and aggregation utilities will per form quicker gt D v o Q o o keystats keystats Performs an update statistics command on each key table in the database keystats d debug level O sql timeout 7u V Description The purpose of this utility is to maintain the index of each key table in order to max imize performance when collectors and data aggregators select against the key tables Options keystats has the following options d O u V Enables you to specify a debug lev
269. ete 7 17 request definitions 7 10 database 7 8 objects 17 1 22 7 22 9 points 12 20 13 19 repository 1 7 1 8 1 9 1 14 rollup 9 1 source 10 3 16 3 16 4 16 6 node 13 13 structure 16 1 summarization 9 1 9 15 type 6 4 9 10 11 20 22 17 Data Manager 1 4 2 13 8 1 22 19 definition 1 13 Delete table s option 8 4 TREND 3 6 User s Guide Refresh option 8 5 Set Default Storage Time option 8 5 Show by Source option 8 5 Show by Type option 8 5 Sort option 8 5 Truncate table s option 8 4 viewing RMON tables 16 8 window 9 10 B 4 Data Type box 10 10 10 14 data_db 7 8 value 7 8 database changing 7 3 full 8 2 monitoring size with Data Manager 8 1 performance 16 5 size 2 20 space 8 9 summary information db used 8 2 tables 10 2 view 22 1 viewing tables with Data Manager 8 1 Database Summary 8 9 DataPipes 9 1 9 3 9 4 date range 12 19 13 19 day range that wraps 18 12 db used 8 2 8 9 db_delete_data 1 4 1 8 C 19 utility A 4 dbstats tables 18 6 Default Storage Time Window 8 3 8 4 Value 6 5 default 11 10 12 16 12 18 13 15 13 16 14 5 14 14 15 4 18 7 18 9 18 13 22 7 columns 22 15 directory 1 10 1 16 14 7 expressions 22 7 22 16 path 3 7 port number D 8 public 5 15 qualifications 22 8 22 9 22 14 report directory 2 11 topology database 18 8 Default Report Directory changing 2 11 default table length 11 8 Define Nodes 3 6 window 2 5 Types 3 6 window 2 9 Users and Groups option 2 3 3 6
270. ewing lists of values in displayed in different orders or grouped for comparison Tabular formats are especially useful for excep tion reporting showing for example exceeded thresholds TRENDsheet provides these functions and also allows selection of the node to re port on and which database to query for the data TRENDsheet reports can be print ed and data can be exported to an ASCII file for transfer to spreadsheets or other applications ok a m z o 7 7 i rec Main Display and Menu System You invoke TRENDsheet from the Tools menu of the TREND main window The TRENDsheet window appears initially with a blank display area and a menu bar pro viding access to TRENDsheet functions see Figure 11 1 11 1 11 TRENDsheet TRENDsheet _ Of x Figure 11 1 TRENDsheet Main Window Menu Command Summary File Menu Use the New command to create a new tabular report see Creating a New Tabular Report on page 11 5 11 2 Main Display and Menu System TREND 3 6 User s Guide The Open command brings up pre defined tabular reports see Opening an Exist ing Tabular Report on page 11 4 The Save option is used to store a previously saved report Save As is used to store new reports or to copy or move existing reports Tabular reports can be printed to a postscript printer using the Print command see Print on page 1 17 Export TRENDsheet data to word processors a
271. fOutOctets 8 ifSpeed delta_time 100 0 Click on Add to View to save the expression When all expressions to be included in the graphical report have been created click Cancel to exit the Assign Statistic Aliases window Once you have finished defin ing the basic report in Report Scheduler you must save the query file you have cre ated The Save Save As and Exit options on the Report Scheduler File Menu will each invoke the File Selection window prompting you to store the file under a name of your choice The file will then be automatically loaded into TRENDgraph After saving the report select Exit to quit Report Scheduler You will automatically return to TRENDgraph Selecting Data For new reports TRENDgraph will automatically invoke the Select Data Set win dow allowing you to define plot lines The Select Data Set window can also be brought up manually by clicking on the Select Data option under the View menu 12 8 Creating a New Graph TREND 3 6 User s Guide Select Data Set Set DOCS ey Description Line TyplLine Cel Line WicPoint Till In Util Cia a a a Red 3 None en a oe NEU Ml Figure 12 4 Select Data Set Window Each plot line displays the values of a statistic or expression for a particular device and key over time Type color and width of lines and points can be modified here As new lines are defined their definitions are displayed in the Lines to graph box in the Select Data Set windo
272. file d Delete procedure A 43 TRENDpm A 44 ot pe Set trace parameters Note that there is no space between the trace group and the trace level Valid values for the trace group are traces the application traces the database traces the library traces all groups a Valid values for the trace level are 1 5 where 1 is the lowest ver bosity and 5 is the highest verbosity Executes the procedure synchronously Add the a option to exe cute the procedure asynchronously Generates the code Display the command line format help Name of generated or output file TRENDpm places the header information into the CodeGeneratedFile and overwrites it if it exists Adding the c option causes TRENDpm to check for the CodeGeneratedFile if it exists TRENDpm exits with an error Specifies the target table name Specifies the execution parameters Do not use any spaces in the parameter list however there is a space after the pe option Valid parameters are bArchive turn_off_archiving bCheck index check index bDelta time delta time bSuppress spike suppress spikes debug level debug level pm line suppress value min filter value retry interval retry interval zerror clock error value See pe Parameters on page A 45 for the descriptions of the parameters TRENDpm Pg Specifies the generation parameters Do not use any spaces in the parameter list however there is a space
273. for a license on the network until it finds it or it times out The license manager will normally locate the license server host automatically You may explicitly specify the server host s via the environment variable DSI ELMHOST It is strongly recommended that the vari able DSI_ELMHOST be set in the application environment env to the server where the license resides This will allow better control of license server access and will also improve system performance The files to be modified are Cshrc and Profile lo cated in DPIPE_HOME lib Example If the license server resides on the host Desktalk setenv DSI_ELMHOST Desktalk will broadcast setenv DSI_ELMHOST Desktalk will not broadcast Once DSI ELMHOST is set TREND applications request licenses from Desktalk as well as broadcasting a request on the local IP Subnet If there are multiple TREND installations where each is running a separate License Manager and users connect to each system directly i e not using remote clients to display TREND GUIs then an implementation is available to provide further lock down on license allocation by performing these steps 1 Create a resource file License Key Administration TREND 3 6 User s Guide UNIX Systems a Create a text file using VI and enter the following line SIPACCEPT IP Address of the License Host b Save the file in the directory usr trend_license with a name such as dsi_resource Windows NT a Crea
274. ggregating it The TRENDit process performs data aggregation in TREND Data aggregation is performed using a series of steps typically leading to longer retention of more highly aggregated data Raw data tables in TREND are populated with timestamp values constants and sta tistics Statistics may be counters integers strings or gauge values A counter is a continuous raw count of a statistic since a device was rebooted A gauge is an instan taneous measurement that is only meaningful at an instant in time Rate tables are created by selecting data from raw data tables and calculating delta values between successive points Periodic aggregations rolls up average total minimum and maximum values by the hour day week and month Obviously data tables cannot be created until sufficient time has been allowed for data to be collect ed aggregated data hourly daily weekly and monthly will not be available until after TREND s daily aggregation process runs In general you must wait until at least the next day after beginning data collection to start producing reports Reporting TREND reporting is based on the creation and manipulation of query files These files contain all of the information necessary to define display and print tabular and graphical reports TREND reports can be created using the TRENDbuild applica tion which is accessed by both the grapher TRENDgraph spreadsheet TRENDsheet and Grade of Service applications
275. gh the etc hosts file Domain Name Service or NIS SNMP GET and SET commands require passwords known as community strings The Read Community String is used for GETs or polling The Write Community String is used for SETs The default Read Community String is public and the de fault Write Community String is private You can override the defaults for the cur rent node by entering new values in the Read Community String and Write Community String boxes in the window The strings are case sensitive and must match the strings expected by the device exactly The Node Type is the kind of device you are entering e g SNMP or RMON Select the node type from the list available by clicking on the downarrow next to the Node Type box You can poll groups of nodes that share a common type Use the Define Types option from the Configure menu to develop a list of type values in advance to Defining Nodes TREND 3 6 User s Guide avoid inconsistencies see Defining Types on page 2 9 RMON must be used only for RMON devices all others are SNMP Note that if you want to collect two kinds of data from a device set it to RMON To add a node once its properties are entered the way you want them click on the Add Node button To delete the current node click on the Delete Node button To quit out of the window click on the Cancel button Defining Views N E m Sz z no Eie o Nio fy LI fo Q EHE P i e 3 A view is a
276. ght See Using the e f and g Options To Filter Data by Date on page 18 11 l Length of the time period to copy This option can only be used with the b option length can be any integer and unit can have one of the follow ing values h hour d day m month zm y year E See Using the b and 1 Options To Copy Data on page 18 13 G Ir m Minimum threshold for a bulk copy packet bcp transaction The NG default is 100 rows M Maximum threshold for a bulk copy packet transaction The default is 100 rows n Provides filtering by node type If this option is used the U option must also be given p Specifies the id for a defined list of instances to be copied The id refers to a specific collection policy based on direct instance poll ing Information about this type of collection policy is stored in the dsi polling list table r Specifies copying TRENDit related tables Valid values are Raw tables Rate tables Hourly roll up tables Daily roll up tables Weekly roll up tables Monthly roll up tables All TRENDit roll up tables zZzomm TRENDcopy Syntax 18 5 18 TRENDcopy R S S t T u U ay V W 18 6 Specifies copying TRENDsum related tables Valid values are Raw tables Rate tables Hourly roll up tables Daily roll up tables Weekly roll up tables Monthly roll up tables Quarterly roll up tables Yearly roll up tables All TRENDsum roll up tables
277. group of the collected data to be reported upon Use the Select Mib and Select Group buttons in the TRENDbuild window EXAMPLE Selecting the Hourly MIB II ifEntry Group To create a report based on hourly mib II ifEntry data first click Select Mib and se lect mib II from the Select Mib pop up box Click OK Next click Select Group and hour mib II ifEntry from the Select Group pop up box Click OK Creating a New Grade of Service Report 13 7 13 Grade of Service 13 8 The next step is building expressions to be graphed using the available statistics Click the Assign Expressions box in the TRENDbuild window to invoke the Edit Expressions window Figure 13 3 Select Statistics Active Expressions delta_time gt Delta Time E E received usec ifAdminStatus iflnDiscards iflnErrors iflnNUcastPkts iflnOctets iflnUcastPkts iflnUnknownProtos El FH asa eee New Expression Add Delete Add to View Cancel Figure 13 3 Edit Expressions Window The Edit Expressions window provides Selection of statistics that will be included in expressions to be graphed upper left Mathematical operators to be used in expressions beneath Select Statistics box Creation of expressions from statistics and operators bottom Naming and adding deleting expressions and 9 9 9 Listing of Active Expressions assigned upper right EXAMPLE Creating an Expression Input Utilization To spec
278. h percentile values for a particular element property or expression such as Utilization over time are used for forecasting because percentile values eliminate spikes from the collection samples 9 18 Forecasting TREND 3 6 User s Guide TRENDsum forecasting calculations are based on a least squares linear regression which means that the calculations attempt to draw a straight line through a set of data points collected over time as shown in the following diagram v 5 2 D O E D en 5 uonebaibby eleg 6 Time p Each data point represents a daily 95th percentile value for the property being fore cast You want to look at peak utilization for planning purposes so TRENDsum uses a daily 95th percentile value which is a near maximum value For a DTT calculation you have these cases You will hit the threshold in the calculated number of days You are already beyond the threshold You are never going to hit the threshold the line slopes downward away from the threshold value You are a long distance from the threshold If the threshold is too far away the forecast is meaningless Forecasting 9 19 9 Data Aggregation and Manipulation If the slope is negative decreasing sloping downward the result returned for DTT is a null value Likewise if the threshold is more than 1 000 days away the result returned is null The formulas that TRENDsum uses to compute all forecasts are gi
279. h the function is applied The val is considered to be counter data if more than one column appears in the expression or if a single col umn expression is of counter data type The val is considered to be gauge data only if the single column expression is of gauge data type gt D E v D e D no TRENDsum By Variables A by variable indicates a time period or a table element such as dsi_key_id At least one by variable is required to specify the aggregation level If both a table element and a time by variable are given the time by variable must be the last one specified A 81 TRENDsum A 82 If a time by variable is not specified TRENDsum aggregates the selected data into one row within any element by variable specified Each non time by variable must exist as a column in the source data or key table Each non time by variable after the first one must also exist as a column in the des tination key table The first by variable is always assigned to dsi_target_name If the column name differs between the source and destination tables use the alias fea ture For example by variable h key frcircuitnumber In this example the name of the column in the source table is frcircuitnumber and the name of the column in the destination key table is h key Null Columns You can specify a percentage of source table rows that must be null for TRENDsum to include a null value in the summary row of the output table See the n
280. hatever character string you input with the c option will be used as the iden tifying community string for the GET requests If you want to specify multiple community strings for the SNMP GETs you can use the f option This option can be used in two ways 5 15 5 Discover AN CAUTION Ifyou use the syntax f ile name where file name is the name of a file but does not include the file s path Discover will look for that file only in the SDPIPE HOME scripts directory Ifyou use the syntax f ile path where file path is the name and path of a file including either or V Discover will look for that file only in the specified path This file is a regular ASCII text file with one read community string per line Discover will try the community strings in this file until the first successful SNMP request is performed Community Strings Files Discover uses the entries in the Community Strings file in the order in which they appear The sequence is as follows 1 Discover polls the first node using the strings identified in the Community Strings file Discover polls using the first string If there is no response Discover tries the second string then the third etc When the node responds to a poll Discover classifies that device as responding to that community string and goes on to the next node 2 Discover does the same for each node Itis up to you to determine in which order the communit
281. he latest timestamp row in interval Creating a New Report Defining a new report focuses on the data to be reported on and what it should be called in the report The series of steps to follow are 1 2 3 4 5 Select a MIB Select statistics Assign aliases Creating a New Report Create expressions Select a group usually a set of MIB objects i e miblet 10 5 mh i E m z g ley Q 10 TRENDbuild 6 Save the report If there is already a report open in TRENDbuild you can clear it from the current context by selecting the New option under the File menu Selecting MIBs and Groups A category is an organizational level for collections or groups of data In general a category is equivalent to an SNMP MIB file The next level of data organization un der a category is a groups A group includes a collection of related raw data tables and the rollup tables derived from the raw tables To choose a category click on the Select Mib button to bring up the Select Mib window Select Mib Ea Select From cisco100 ifentry dsi ecam framerelay dte rfc1 315 Figure 10 2 Select MIB Window Once a category is chosen its name appears in the Mib box Scroll through the en tries in the window to find the one you want 10 6 Creating a New Report TREND 3 6 User s Guide Select a category by clicking on a category name listed then clicking on the OK button or simply by doubl
282. he Exit command found in the File menu of the Main TREND window is used to quit TREND Similarly the Exit command found in the File menu of all subsequent modules lets the user quit that module and return to the Main window About TREND To determine the version number of TREND running on your system select About from the File menu in the menu bar of any TREND application Commonly Used Commands 1 Introduction to TREND TREND 1 20 TREND 2 TREND Configuration for trendadm N m Sz zU oO J30 as 0 Q Q 3E v z 3 The TREND administrator trendadm has access to configuration and administra tion options not available to other TREND users The trendadm can for example define users and views made up of different groups nodes and types to collect data on This tailoring of the configuration is described in detail in this chapter TREND has advanced features that allow the trendadm to change defaults or rollup and aging parameters of the data in the database In addition only the trendadm can truncate and delete tables in the database The following sections describe these ca pabilities in more detail Configuring the System The main TREND window allows the trendadm to configure the system and to take advantage of some of the advanced features of TREND including distributed poll ing setting user privileges and utilizing multiple databases You access these func 2 1 2 TREND Configuration f
283. he File menu untitled ok J m z o 7 7 d s Figure 11 2 TRENDbuild Main Window To finish defining the new report you must set its display parameters This includes which node and key or keys to display column precision specifying report titles and setting the column order and column widths Finally Save the fully defined report Creating a New Tabular Report 11 5 11 TRENDsheet Note After modifying a newly created TRENDsheet to a file and saving it to a file if you add a where clause to the entry there must be a space before the word where to properly run the report Selecting Statistics 11 6 The first step in creating a new tabular report is to select a category MIB and group of data Use the Select Mib and Select Group buttons in the TRENDbuild window The Select Statistic box contains a list of statistics chosen and expressions created in TRENDbuild when the report was created Select an expression using the Select Statistic box s down arrow button then clicking on the desired statistic Use the scroll bar to see all the available names EXAMPLE Select the Weekly RMON Host Group To create a report based on weekly rmon host data first click Select Mib and select rmon from the Select Category pop up box Click OK Next click Select Group and week rmon host data from the Select Group pop up box Click OK The next step is building expressions to be displayed using the available sta
284. he File menu to bring up the Change Database window Change Database Ea Change to CIMARRON SYBASE Y OK Cancel Figure 12 6 Change Database Window Selecting Different Data TREND 3 6 User s Guide Note A report may not display the same information for both databases The name of your current database is displayed in the Change to box To change to a different server Click the down arrow in the Change to box to see a list of the Sy base SQL Servers known to this machine The values presented here are read from the SYBASE interfaces file Select the name of the server you want to access in the report Click the OK button to change to the selected database Click on Cancel to quit out of the window without changing servers Setting the Time Period to be Displayed TREND6raph allows you to set a Start Time and an End Time for data to be dis played in reports If you select Time Period from the View Menu the Set Time Pe riod window is displayed Set Time Period Figure 12 7 Set Time Period Window E N D m Z o Q 5 ko Selecting Different Data 12 13 12 TRENDgraph Change the Start and End times by selecting the button that corresponds to the por tion of the time you wish to set i e Hour Day or Week Next click on either the or buttons to increase or decrease this value respectively Click the OK button to have the sort order you specified take effect
285. he cache named keycache enter the following commands 1 gt sp_bindcache keycache dpipe db dsi dpipe Kib ii ifentry 2 gt go 1 gt sp_bindcache keycache dpipe db dsi dpipe Ky interface perf 2 gt go 1 gt sp_bindcache keycache dpipe db dsi dpipe K5 frcircuitentry 2 gt go Sybase Configuration Parameters C 17 C Performance Tuning Issues Verify the Bindings Were Successful Issue the following command to verify that the binding of the key tables is successful isql Usa P 1 gt sp_helpcache 2 gt go Status should be V for valid on each binding The output will look something like this Cache Name Config Size Run Size Overhead default data cache 0 00 Mb 35 87 Mb 1 93 Mb keycache 1 95 Mb 1 95 Mb 0 15 Mb Memory Available For Memory Configured Named Caches To Named Caches 37 84 Mb 1 95 Mb There is 35 89 Mb of memory left over that will be allocated to the default cache Cache Name Entity Type Status keycache dpipe_db dsi_dpipe Kmon_ethstats_ table V keycache dpipe db dsi dpipe KDrouter traffic table V keycache dpipe db dsi dpipe Kmon2 alhost table V keycache dpipe db dsi dpipe KDhp filesystem table V keycache dpipe db dsi dpipe Kmon2 almatrix table V keycache dpipe db dsi dpipe Kdbaselineday table V keycache dpipe db dsi dpipe KHhp compsystem table V You can perform these functions as necessary Add or modify the named cache keycache when you suspect that the cache allocation is out of date i e
286. he desired fill type Similarly fill colors can be changed using the Color box Adding Changing and Deleting Plot Lines To add a line definition to the graph once its parameters are set click the Add Line button The new line definition will appear in the Lines to graph box To change the definition of an existing plot line Click the line to be changed in the Lines to graph box Its parameters will be displayed in the appropriate areas in the window Make the desired changes then click the Update Line button The line s definition in the Lines to graph box will display the new values Use the Update All Lines button to update the values of all lines after making changes to multiple lines To delete a line Click the desired line in the Lines to graph box then Click the Delete Line button To delete all the lines in the graph Click the Delete All button When you want to display the defined lines in the TRENDgraph window Click the Graph button The graph will redraw according to the current line definitions Setting X Axis Parameters 12 16 The horizontal or X axis represents time in TREND graphs You can adjust the time scale to affect how spread out or compact the graph will be The larger the time scale increment the more information will appear in a smaller space The trade off is crowded data The default setting is Autoscale Setting Report Display Parameters TREND 3 6 User s Guide To change the X axis
287. he device if you have specified host name translation If your system has a protocol to translate IP addresses to host names and the host name is found Discover associates it with the discovered device If the host name is not found Discover lists the node by its IP address IP Discovery Table Population When Discover gets a response from an SNMP capable device it inserts the host name or IP address of the device into the dsi nodes table Once it exists in dsi nodes the discovered device is accessible to TREND What Discover Does 5 8 5 Discover Note TRENDdiscover only places primary IP addresses or hostnames if using translation into the dsi_nodes table You may use an ISQL session to reference the dsi_node_list table to determine the TRENDdiscover anal ysis of primary versus secondary addresses The dsi_primary and dsi_macindex columns in the dsi_nodes table contain the addresses The dsi_nodes table also has a column that defines whether or not an SNMP device is RMON manageable However IP Discovery cannot determine this RMON capa bility is discovered by SNMP Type Discovery When Discover finds a new SNMP capable node it inserts it into dsi_nodes as an SNMP device by default SNMP Type Discovery 5 4 Note SNMP Type Discovery is scheduled by default to run once a day In addition when you install a TREND Report Pack the Autopilot asks if you want to run SNMP Type Discovery at that time SNMP Typ
288. he first day of the month to midnight on the last day of the month Reports definitions that are processed by Ipr launch read their data from the data base named by the DSQUERY environment variable of the user who invokes Ipr launch When invoked by trendtimer Ipr launch uses the DSQUERY value of the user trendadm lor launch Options Ipr launch has the following options d Set a debug output level Values of 0 1 2 or 3 are valid The higher the number the more detailed the information The default is 0 which means no debug output Debug output is written to standard out i The time interval to indicate whether to run daily weekly or monthly reports specified as follows il Daily i7 Weekly i 31 Monthly This parameter must be specified on the command line There is no default V Displays the version stamp for lpr_launch Examples gt D v o Q o o To have Ipr launch run reports configured to be run weekly lpr launch i 7 mw collect mw collect mw collect is available in two versions standard and cache The standard version is a non cache version that is part of the TREND system The cache version provides the ability to run without the database and is available for additional purchase If you have the cache poller please see Instructions for Loading the Cache Version of the Poller on page A 28 mw collect polls nodes for SNMP data mw collect
289. he for log pages only logonly option Once the cache has been created you need to bind the transaction log to it Note The database must be set to single user mode before you can bind the transaction log to the log cache To bind the transaction log to the newly created log cache 1 Place the database into single user mode as follows be sure there are no active TREND transactions before proceeding isql Usa P sp dboption dpipe db single user true go The output looks something like this Database option single user turned ON for database dpipe_db Run the CHECKPOINT command in the database that was changed return status 0 2 Switch to the dpipe db database use dpipe db and issue a checkpoint command followed by the sp bindcache command as follows use dpipe db go checkpoint go sp bindcache logcache dpipe db syslogs go Sybase Configuration Parameters C 13 9 O 2h O 3 pi S 1 el 3 Ce C Performance Tuning Issues 3 Switch back to the master database to return the dpipe_db database to its normal multi user state use master go sp_dboption dpipe db single user false go The output looks something like Database option single user turned OFF for database dpipe_db Run the CHECKPOINT command in the database that was changed return status 0 4 Checkpoint the dpipe_db database use dpipe db go checkpoint go 3 All
290. he formulas used in 3 5 1 Table A 1 Formulas Used in TRENDsum 3 4 x Function Rate Counter Rate Gauge Summary Item tot same as 3 5 1 val cnt same as 3 5 1 avg Yval YAt same as 3 5 1 same as 3 5 1 wav X val At cnt X val ent X val cnt variance cnt Z val At same as 3 5 1 same as 3 5 1 used in E val At computing cnt cnt 1 std where val is the resolved value of the expression to which the function is applied and cnt is the number of samples being summarized In both TRENDsum 3 4 x and 3 5 1 the val is considered to be counter data if more than one column appears in the expression or if a single column expression is of counter data type The val is considered to be gauge data only if the single column expression is of gauge data type While the formulas in TRENDsum 3 4 x and 3 5 1 are the same except for the differences noted in the table above the results obtained when these formulas are applied to the expressions in a TRENDsum file can be quite different in 3 4 x and 3 5 1 The discrepancy occurs because TRENDsum 3 4 x performs a hidden divide by delta_time operation when all functions except tot and cnt TRENDsum are applied to rate counter expressions This divide operation is not explicitly specified in the expression but is performed by the TRENDsum processing code In 3 5 1 on the other hand the hidden divide by delta_time is not perform
291. he scroll bar to view the list of avail able colors Select the fill pattern for the current statistic by clicking on the Fill Pattern box s down arrow button then clicking on the desired pattern Use the scroll bar to view the list of available patterns Note Keep in mind how the graph will look on paper while you set it up espe cially if you have a color screen and a monochrome printer Besides the shades of gray available only fill patterns will be distinguishable on a monochrome printout Setting X Axis Parameters 13 16 The horizontal or X axis represents time in TREND graphs You can adjust the time scale to affect how spread out or compact the graph will be The larger the time scale increment the more information will appear in a smaller space The trade off is crowded data The default setting is Autoscale Setting Report Display Parameters TREND 3 6 User s Guide To change the X axis time scale select the X Axis option under the View menu to bring up the X Axis window 2 gt o o Q D e h o I S iv om Figure 13 9 X Axis Window The time scale options in the Select Scale box include autoscale Scale the graph to the optimal scale automatically so all available data fits in a single window 15 Minutes Each hash mark on the X Axis represents 15 minutes 1 Hour Each hash mark on the X Axis represents 1 hour 1 Day Each hash mark on the X Axis represents 1 day 1 Week E
292. heet gt TRENDgraph gt The Tools menu contains two options TRENDsheet and TRENDgraph Create a simple tabular or graphical display of data in a database table by invoking TRENDsheet or TRENDgraph from the Tools menu Note For RMON data only Control Tables can be viewed TRENDsheet and TRENDgraph displays of RMON raw and roll up database tables are not available Setting Up New Polling Requests 7 4 Collect Data provides a mechanism for defining and managing the set of polling re quests that comprise an SNMP and RMON collection policy TREND s polling agents collect data from nodes on your network according to the request parameters defined This information includes What data to collect Which nodes to poll How often to poll Which system does the polling 9 Where to store collected data Setting Up New Polling Requests TREND 3 6 User s Guide Specifying What Data to Collect TREND collects data on variables i e MIB objects and groups of individual vari ables called MIBlets To define the polling parameters for the selected group of variables or table click the Collect Data Periodically option under the File menu This will bring up the Collect Data Periodically window Collect Data Periodically ciscol00 ifentry_interface pe individual node J 134 70 18 255 suwanee desktalk com select Views gl Select Types E suwanee desktalk com N e ol o o o D
293. heir DESCRIPTION field often contains the word count or total number as in Total number of octets since the device booted For example the following field should be defined as a Counter sysLinkCellsTx SYNTAX INTEGER ACCESS read only STATUS mandatory DESCRIPTION Count of ATM cells transmitted on the link sysLinkEntry 10 The correction follows sysLinkCellsTx SYNTAX Counter ACCESS read only STATUS mandatory DESCRIPTION Count of ATM cells transmitted on the link sysLinkEntry 10 Fields that function as a Gauge usually indicate how much or how many of something there are at any given point in time Their DESCRIPTION field often contains the words in use used or available For example this field should be defined as a Gauge Preparing MIBs for MIBwalker TREND 3 6 User s Guide sysCurrentBuffers SYNTAX INTEGER ACCESS read only STATUS mandatory DESCRIPTION The number of system buffers currently in use sysLinkEntry 22 The correction follows sysCurrentBuffers SYNTAX Gauge ACCESS read only STATUS mandatory DESCRIPTION The number of system buffers currently in use sysLinkEntry 22 2 ius 2 x D These data types should be corrected before data is collected from the MIB so the database tables use the appropriate field types Note If the changed variable appears in a SEQUENCE OF list
294. hem on or No to turn them off To change the color of the grid lines Click the down arrow of the Grid Color box and select the color you want Of course the color will only matter if you are display ing grid lines Click the OK button to activate the revised selections Click on Cancel to retain the original values Setting Report Display Parameters TREND 3 6 User s Guide Drilling Down for More Detailed Data TREND graphical reports contain a drill down feature that allows the user to click on a region of a graphical report to obtain a second more detailed tabular report dis playing the data of interest This data can point to other reports that will provide a complete picture of the area of concern Ethernet Multi Segment Utilization Comparison m ial 03 00 Mon Apr 07 1997 23 00 Tue May 06 1997 49 64 TI 4255 I 3546 IN 2838 21 27 L_ 03 04 05 06 OF 08 D9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 03 00 Mon Apr 07 1997 Hourly Scale Percent Utilization orinoco avg ALL 1 Percent Lttilization rmonfntr2 avg ALL 3 Percent Utilization rmontntr2 avg ALL 21 Figure 12 12 Graphical Report with Supporting Drill Down Tabular Report mah N J m 2 O Q o o T Drilling Down for More Detailed Data 12 21 12 TRENDgraph Command Line Options TRENDegraph is invocable from a command line via the command agx_graph Enter the command ag
295. ice for most medium and large installations Number of Physical Disk Devices If the system needs to update multiple files scattered in different locations on the disk subsystem a small number of disk drives for example one can reduce system performance considerably In this case the disk actuator must perform numerous C 4 Hardware Considerations TREND 3 6 User s Guide seeks A seek operations means a movement of the disk heads which is the slowest disk operation Limiting the number and distance of seek operations substantially improves throughput of disk I O bound systems 9 O Dm 2h 3 O S 1 D el 3 Ce In general the more disk drives the better throughput assuming that critical files are strategically placed on these drives For example if all data that is used for a normally operating system is placed on a single device and the remaining devices are used for backup no benefit is derived from a multiple disk configuration When considering the use of multiple disks make sure the database transaction log is on a separate device from the database itself if possible Also place the operating system s page swap virtual memory files on a separate device or devices Some vendors offer disk array products that not only have multiple drives but also provide sophisticated hardware that allows for hardware partitioning of database tables imbedded cache disk mirroring RAID and other options that im
296. ide counters by delta_time this is the default behavior and define this column to be of gauge data type define IGNORE_TIMEADJUST This statement disables the ta_period delta_time adjustment mechanism which means that TRENDstep ignores any differences that may occur in the delta_time column This statement is valid only if the input table is a rate table See the description for define TIMEADJUST for the default behavior FINAL ENDFINAL These statements enclose the statements for the final block The statements in the final block are executed only once on the result columns Only output data columns and output key columns can appear inside a final block A 59 gt D v o Q o o TRENDstep A 60 Statement Description assignment There are two types of assignment statements a b which means a equals b a b which means a equals a b To initialize a column to 0 use the fol lowing statement column name 0 P by_variable expression Specifies the dsi_target_name and dsi_table_key values for the output key table The first by_variable identifies the target name all the other by_variables define the table key This statement is required Use the expression KEY column_name to identify a key table column name Examples by_variable KEY dsi_target_name by_variable KEY dsi_data_source by_variable dsi_agg_type case logical_oper
297. ific MIB data from any SNMP or RMON agent Pol lers collect data and deposit the data into the server database The server contains the data repository with polling information data tables and indexed key tables Note TREND servers can be physically distributed across the enterprise It is also possible to configure multiple servers with a server poller combina tion see Satellite Database Servers below TREND employs specialized modules for defining and implementing an SNMP polling policy Collected data is bulk copied into a designated database using a com mon schema The common database schema allows data from multiple sources in multiple formats to be processed and used for reporting by TREND Using the MIB walker tool the user can browse MIB structures and define additional data polling requests on an ad hoc basis The particular group of MIB variables or even the entire MIB group selected will be added to the TREND database for future polling requests MIBwalker uses the Collect Data application and the mw_collect process to enact data polling Once collected raw data must be reduced into a more usable form Data reduction in TREND has been designed to meet several goals including speeding reporting A Brief Tour of TREND 1 9 1 Introduction to TREND producing a schema in which simple queries produce powerful results and maintain ing a manageable database size TREND reduces data by rolling it up over time or a
298. ify definitions made in an existing reports using the Aliases Expressions com mand see Modifying an Existing Report on page 12 5 Opening an Existing Graphical Report Existing reports are stored as files on disk You open them by loading them into TRENDgraph Click the Open option under the File menu This will bring up the File Selection window see Figure 1 2 The Select File Window 12 4 Opening an Existing Graphical Report TREND 3 6 User s Guide Modifying an Existing Report Any of the defined display parameters can be modified at any time by substituting new values and repeating the steps followed to create the original format To modify the definitions setup through TRENDbuild click the Aliases Expressions option under the Edit menu This will put you into TRENDbuild in the context of the cur rent report That is the TRENDbuild category group statistics aliases and expres sions windows will contain the values used by the report being modified After you finish making changes in TRENDbuild reload the report into TRENDgraph using the Open option under the File menu for those changes to be reflected in the report on your screen If you made any changes to the report definition before requesting Aliases Expres sions TRENDgraph will ask if you want to save those changes before proceeding Creating a New Report from an Existing Report There may be times when you want to use one report as the basis for another To
299. ify input utilization for a device first click in the New Expressions box and enter a name for the expressions to be created e g Input Utilization Click Add to place the expression in the Active Expressions box Click the new expressions name in the Active Expressions box to continue make sure it is highlighted Creating a New Grade of Service Report TREND 3 6 User s Guide Select the following combination of statistics and operators to create the expression for Input Utilization at the bottom of the Edit Expressions window ifInOctets 8 ifSpeed 100 0 Kar 2 lo o Q D ie D S 7 m It is important that the left and right parentheses are paired Click Add to View to save the expression Once you finish defining the basic report as described below in TRENDbuild you will need to save the query file you have just created The Save Save As and Exit options on the TRENDbuild File Menu will all prompt you with the File Selection window to store the file under a name of your choice The file will then be automat ically loaded into GOS Selecting Data When you run a new report GOS will bring up the Select Data Set window after you exit TRENDbuild Alternatively the Select Data Set window can be brought up by clicking on the Select Data option under the View menu Select Data Set Select Statistic Fill Pattern Color itadminstatus E Wee Z Black z Select Key Scor
300. ilities of TREND allowing the advanced TREND user to select new groups of MIB variables and add them to the database schema MIBwalker also provides conventional MIB browsing via its MIB tree structure and point and click interface mw collect TRENDS primary polling agent for collection of SNMP data from nodes on your network Report Launcher Report Launcher is used to display reports by selecting a query file and related network element for which data has been collected Report Scheduler Report Scheduler is used to set up groups of pre defined reports to be printed for groups of devices on a daily or weekly basis rmon collect rmon collect is a specialized polling agent tailored to the unique RMON MIB It collects RMON history host matrix and statistics groups and supports RMON net work and application layer functionality via AXON ECAM Netscout Systems Domain View and other extensions 2 of 5 TREND Overview 1 5 1 Introduction to TREND 1 6 Module Description TRENDbuild TRENDbuild lets the user create meaningful graphs and tabular reports without using SQL Using a point and click GUI interface to select variables assign custom aliases and build expressions TRENDbuild generates report definitions which are used by TRENDgraph TRENDsheet or Grade of Service These report defini tions are query files that can be stored anywhere the user chooses allowing creation of rep
301. ime and compute multiple complex statistics about the summarized data Specifically TRENDsum can compute the following statistics Total average weighted average standard deviation median minimum and maximum ta period and delta time when minimum and maximum values occur 90th 95th and 98th percentiles 30 60 and 90 day resource utilization forecasts 9 9 Days until threshold forecasts While TRENDit generates a separate row for each statistic TRENDsum generates only one row for all the statistics you select to compute Each statistic appears in a separate column in the row TRENDsum generates separate rows for each of the following grouping categories Any property that is defined for an element Examples of element properties include the device itself a router for example as well as properties of that Other Data Aggregation Functions 9 15 9 Data Aggregation and Manipulation device for example Ethernet FDDI token ring serial 56 serial T1 interfaces the speed of an interface the location of an interface such as Sales Department or the client to whom the resource is dedicated such as the XYZ Company Time The time categories by which you can group collected data include Hour day week month quarter and year Day of the week for example by all Mondays Tuesdays and so on This grouping results in 7 rows in the output table for each element property being summari
302. in block 2 Again TRENDproc must execute the commands in block 2 in their proper sequence However since block 2 is a nowait block TRENDproc also begins to execute the sequence of commands in block 3 Since block 3 is also a nowait block TRENDproc can also begin to execute the commands in block 4 At this point TRENDproc is executing commands from three different blocks at the same time Block 4 is a wait block so once again TRENDproc must complete the commands in block 4 before beginning the processes in block 5 However it is possible that TRENDproc will execute all of the commands in block 4 before blocks 2 and 3 are completed In this circumstance TRENDproc does not have to wait for blocks 2 and 3 once block 4 has been executed TRENDproc will start with block 5 Step 3 TRENDproc completes the sequence of commands contained in blocks 2 4 and be gins to execute the processes in blocks 5 7 Again since block 7 is a wait block the commands in block 7 must be fully executed before TRENDproc can begin the se quence of commands in block 8 How TRENDproc Works TREND 3 6 User s Guide Step 4 When TRENDproc executes the commands in block 8 the sequence is finished N pas E D m 2 oa TS mg fe 7 TRENDproc can execute commands from several blocks at once In the example above we see TRENDproc execute commands from three different blocks This al lows a large part of the individual trend proc file to be run in pa
303. in dsi protocols In most cases you can safely ignore any invalid protocols returned by the interface However if a protocol you wish to report on is returned as invalid you can rename it in such a way that the Protocol Definition Interface recognizes it as being valid The procedure for this is described below 1 Click View Invalid to display the invalid returns 2 Click one of the displayed protocols to highlight then click Change Invalid Name The Change Invalid Protocol Name window appears Invalid Name Physical Layer Me 45 54 42 0 Address Type Network Layer Unknown Address Transport Layer IP Address IPX Address Application cc DECNET Address XNS Address OID 8 1 0 0 1 0 0 0 240 2 0 0 Add Modified Figure 17 8 Change Invalid Protocol Name Window 3 To rename the protocol you can either Input the correct name using the RMON naming convention Enter any meaningful name you choose to make it recognizable to you However if you choose to enter an unconventional name the following rules apply DataPipe Components TREND 3 6 User s Guide You must enter an alphanumeric value of up to eight characters in each of the four fields Only the Application layer field can include a period Lj N E fe 2 N o o 3 4 Once you have defined the name and address type that you wish to apply click Add Modified The Protocol Definition Interface adds the newly defined
304. in the raw tables and creating rate tables as output Each row in the rate output table represents delta data that is computed from a pair of rows from the raw input table as follows Assume for this example that the first sample is taken at time t on day 1 which is 12 15 a m if 15 minute sampling intervals are used At time t that is at 12 30 a m the second sample is taken at time t 12 45 a m the third sample is taken and so on The value v for a counter data item in period t is then subtracted from the value v for the counter data item in period ty The difference which is called a delta value is stored in the rate table row associated with period t For all other noncounter data types the value v collected at period t is stored in the rate table row associated with period t Value v for noncounter data items is lost The time in seconds between the two sampling periods t tj is stored in a col umn called delta time Delta time is computed by subtracting the SysUpTime of the first sample from the SysUpTime of the second sample dividing the result by 100 and then truncating the remainder For the following diagrams if a 15 minute sam pling interval is used the delta time between all rows is approximately 900 15 minute period 60 seconds per minute However the actual delta time could be 894 901 905 or any number close to 900 depending on the calculation Data Cleaning 9 5 9 Data Aggregation
305. in the table but keep the table itself Truncating a table does not make any entries in the Sybase transaction log so it can not be recovered Only the user trendadm can truncate tables For all other users this option will be grayed out To truncate one or more tables Changing Aging on a Specific Table TREND 3 6 User s Guide 1 Select the tables to be truncated by clicking on their rows in the Table Informa tion area The selected rows will be displayed in reverse video 2 Select Truncate table s in the File menu 3 For each table to be truncated this window will be displayed N m 032 Ag no 3 o as o Q Q ERE P i e D Figure 2 10 The Truncate Table Confirmation Message The window shows the user visible name and the internal database name of the table to be truncated 4 Click on the Truncate button to truncate the table 5 Click on Cancel to keep the data in this table 6 Anew window will be displayed for the next table if more than one was selected Deleting Tables Deleting a table is when you drop in database terminology the table from the da tabase The table and everything in it are discarded Deleting a table makes an entry in the Sybase transaction log so it can be recovered However if the transaction log space required to delete the table is larger than the space available the delete will fail If this occurs truncate the table see Truncating Tables on page
306. indicators of PVC Executive Summary performance utilization and congestion are shown individually and combined into a Grade of Service to reveal PVC health at a glance Hourly Yolume AIIPYCs Tues Jun 30 0 00 Tues Jun 30 23 00 2378112 208 84 8 178 358 4 148 632 zt 5 118 905 6 a 89 1792 59 452 8 28 7264 0 00 00 02 00 04 00 06 00 08 00 10 00 12 00 14 00 16 00 18 00 20 00 22 00 Hourly Scale Figure 9 4 Sample Frame Relay Executive Summary Report Baselines In this report the hourly baseline is drawn as a line over the hourly PVC volume The baseline is computed for each hour of each day for the preceding 42 day window Ranking Ranking TRENDrank applies ranking criteria to summarized data and stores the results in the TREND database Ranking enables you to chart the volatility of a given metric for a given element As an example use utilization and congestion among 100 PVCs to answer the following types of questions 9 23 o 5 2 D ao E D 5 uonebaibby ejeq 6 9 Data Aggregation and Manipulation 9 24 Which ten PVCs rank the highest in terms of utilization and congestion on any given day How often does the same element appear in the top ten in terms of utilization and congestion Are there significant variations in element rank from day to day delta rank A large delta rank indicates a volatile element which you should probably inves tigate A rel
307. ing Weight TPt 2Pts 3 Pts lt 4Pts lt 5 Pts lt 6 Pts lt Stats Add Statistic Update Statistic Update All Statistics Delete Statistic Delete Ai Statistic Device Key Descriptio Fill Patte Bar Colo Weight Graph Cancel Figure 13 4 Select Data Set Window Creating a New Grade of Service Report 13 9 13 Grade of Service 13 10 Note This window allows the user to establish the relative weight that each statistic selected contributes to grade of service for a given device The total weight of all statistics selected must be 100 Total points accumulated from all statistics collected on a device provide a score that reflects that device s level of performance Points are assigned to each statistic as de fined in the table at the bottom of the select data set window Each statistic contrib utes points to the overall grade of service according to thresholds that the user defines for that statistic in the Scoring area of the Select Data Set window The Select Key box invokes the Select Key window Figure 13 5 which contains a list of all keys for the selected device in the database table being reported on by the graph Click the Device box down arrow to view a list of devices and select a device for reporting Select an aggregation type using the Aggregation Type pull down ar row to view and select it Select Key L ol x Device Aggr
308. ing 7 6 directory SDPIPE HOME scripts 5 16 5 19 add 14 10 current 14 3 path 15 3 remove 14 10 Top Level Report 14 4 directory navigation actions 14 8 disassociate an alias from a statistic 10 12 Discover button 5 6 IP discovery 5 2 Notes About Some Discover Com mand Line Options 5 14 discovery community strings 5 10 5 16 for an SNMP GET request 5 15 defining an IP discovery 5 8 hostname translations 5 11 how IP discovery works 5 3 how SNMP type discovery works 5 5 IP discovery 5 2 launching the discovery interface 5 6 running discover from the command line 5 11 setting a range 5 9 SNMP type discovery 5 2 5 4 5 17 Command Line Options 5 18 subnet mask 5 9 type discovery files 5 19 using the graphical user interface 5 6 what discovery does 5 2 Display Options 11 4 display parameters 11 5 12 14 13 16 TREND 3 6 User s Guide distribute polling load 7 8 divide by zero avoiding 9 12 condition 9 14 dl type table 5 5 5 13 5 14 5 19 Do Hostname Translations option 5 11 domain 16 9 domain description 16 4 Domain Name Service 2 6 dotted name format 17 16 notation 17 16 dpipe_db B 4 B 13 DPIPE_HOME 20 5 DPIPE_HOME lib 20 5 dpipe rmon2 17 5 dpipe tlog B 13 drill down 1 11 1 12 12 21 grade of service 13 20 sequence 15 9 drop table 2 17 Dropped Frames 17 2 17 4 dsi agg type 9 9 DSI ELMHOST 2 19 2 22 environment variable 2 22 dsi key table 18 7 dsi node list E 1 table 17 6 dsi nodes 5 19 E 1 table 5 3
309. ing box of the Grade of Service Select Data Set window enter the relative weight the selected statistic contributes to the health of the interface e g 50 for Percent Utilization Enter the upper limit for each scoring level identified next to the number of points that will be assigned to the parameter for a given level of service For example for Percent Utilization one point will be earned for a utilization level under 25 so 25 must be entered in the box next to Pt Similarly 35 will be entered in the 2 Pt box 50 in the 3 Pt box and 75 in the 4 Pt box Keep in mind that a lower number reflects better health in the Grade of Service report It is probably easiest to assign a fill pattern and color to represent the current statistic in the GOS stacked bar graph see Changing GOS Bar Attributes on page 13 16 for further details Creating a New Grade of Service Report 13 11 13 Grade of Service Now that the parameter has been fully defined click Add Statistic to enter Percent Utilization into the Statistics to graph table Repeat these steps for the additional four parameters to complete the table then click Graph at the bottom of the Grade of Service Select Data Set window to display the GOS report Select Data Set Select Statistic Fill Pattern Color Percent Utilization z solid sj Green gt Select Key Scoring owe 1pt lt 25 2 Pis B5 apts lt 50 apts lt 75
310. inter specified for printing Oo File Contains the name of the file to which the report is to be printed p Copies Defines the number of copies of the report to be printed ID The number assigned to the report by Report Scheduler Printing a Report 15 3 15 Report Scheduler Report Instances All report information is presented here use the scroll bar to move through the data if it is not displayed in the table The display area is organized into the following areas Node Contains the name of the node associated with that instance of the report The name default appears if a node is already assigned Key Contains the node key number Description A description of the node key Line Corresponds to a parameter in the query file which can be manipu lated by the user Adding New Reports The Add New Report option invokes the File Selection window See The Open Command on page 1 15 for a complete description of the file selection process Report Chaining Adding Drilldown Reports Report Scheduler s report chaining capability gives users the ability to generate graphical reports for a series of related network elements based on a corresponding listing or exception report There are two general scenarios for use of this feature 1 Inventory Reports For example a report listing all router interfaces is gener ated each week Corresponding graphical reports for each router interface dis play interface health and in
311. ion Parameters on page C 6 TREND Processing Options C 19 9 O Dm 2h 3 O a 1 D el 3 Ce C Performance Tuning Issues TRENDcopy When you use TREND copy to copy database tables there are a variety of processing options For example you can copy several tables at a time thus enabling processor overlap If a specific table is not specified with the t option TRENDcopy copies tables of the appropriate processing option serially This may result in excessive run times By employing the t option you can run multiple TRENDcopy instances simultaneously thus improving throughput and reducing run times TRENDit TRENDsum When you use TRENDit or TRENDsum you can implement a number of processing options with command line switches In environments that are not resource constrained you can reduce elapsed processing time by running multiple instances of TRENDit and TRENDsum in parallel For TRENDit use the t command line option You must know and identify all tables that require TRENDit processing Since you invoke TRENDsum on a per table basis you will know the identity of the tables being processed In a typical environment TRENDit is launched from the trendtimer sched file located in the DPIPE_HOME lib directory The file looks something like this C 20 TREND Processing Options
312. ion opera tions ensure that the tmp directory in UNIX or DPIPE_TMP or c tmp direc tory in Windows NT exists If not create it 7 18 Incomplete Data Collection or Aggregation TREND 8 Data Manager Data Manager is an administrative tool for monitoring the use and growth of a TREND database The TREND user needs a way to monitor the database which is continually changing due to automatic table creation population rollup and aging Data Manager provides an easy to read tabular display of information about every data table in your database s plus a central location for controlling data rollup and aging Any TREND user can view information about the database via Data Manager Only trendadm however can use Data Manager to change defaults modify aging and rollup for individual tables and delete and truncate tables Main Display and Menu System Invoke the Data Manager screen from the main TREND window Oo o D o 3 pi Q C 8 1 8 Data Manager Data Manager ol x File View Tools patuxent_sybase db used 11 Object Name SQL Name Type Rollup Size rows Size KB Keep data for d day cisco100 Isystem pen D100 Isystem perf hour cisco100 Isystem pe H100 Isystem perf rate cisco100 Isystem per R100 Isystem perf week cisco100 lsystem prW100 Isystem perf month cisco100 Isystem r M100 Isystem perf monthly cisco100 Isystem perf 0100 Isystem
313. irectory name in the Look in edit box The Up one level entry Double click on this entry to make the immediate parent of the current directory the new current directory The current directory is always the one identified in the Look in edit box in the Re port Launcher window The entries in the Reports list box are always taken from the current directory Selecting the Report File Directory 14 7 14 Report Launcher A summary of the types of directory navigation actions you can take from the fold ers list box follows refer to Figure 14 2 Report Launcher Window as you review these actions c trendsnmp reports is the current directory c trendsnmp reports is also the Top Level Report Directory but you cannot tell this just by looking at the Report Launcher window The only way you can actually display the current name of the Top Level Report Directory is to dou ble click on the Top Level Report Directory entry in the Folders list box The name of that directory will appear in the Look in edit box Double clicking on the Up one level entry will make c trendsnmp the current directory Cisco Interfaces RMON and ethernet are the names of the first level subdi rectories of c trendsnmp reports Double click on any one of these directory names to make that directory the current directory The directory name will appear in the Look in edit box Designating the Top Level Report Directory When you first invoke th
314. is Note If you want information regarding an understanding of how Discovery interacts with the schema and expanding the list of devices that TREND discovers see Discovery Considerations on page E 1 IP Discovery 5 2 IP Discovery is a manually initiated process It can be initiated either from the Discover user interface or from the command line The purpose of IP Discovery is to find the devices on a network and determine whether or not they are SNMP manageable What Discover Does TREND 3 6 User s Guide Note Any TREND user can initiate Discover but only from a TREND server On TREND Clients the Discover tool is disabled c o ao 7 e lt D mg How IP Discovery Works You initiate an IP Discovery by defining a range of IP addresses a Subnet mask and one or more read community strings to be searched Discover then pings each IP ad dress except beginning subnet addresses i e 0 and broadcast addresses i e 255 in that IP address range If there is no device at the IP address there is no response If there is a device at that IP address that recognizes the ping it responds to Discover From this response Discover knows that there is a device at that IP address Discover then Sends an SNMP GET message to the responding device If the device recognizes and responds to the GET Discover identifies the device as SNMP manageable Attempts to discover the host name of t
315. istration Guide Guide To Building TREND Reports Guide To Viewing TREND Reports Sybase Manuals SOL Server Transact SOL User s Guide Sybase SQL Server System Administration Guide SQL Server Reference Manual Volume 1 Commands Functions and Topics SQL Server Reference Manual Volume 2 System Procedures and Catalog Stored Procedures SQL Server Troubleshooting Guide Standards Documents XXX Structure and Identification of Management Information for TCP IP Based Internets RFC 1155 Simple Network Management Protocol RFC 1157 Concise MIB Definitions RFC 1212 Management Information Base for Network Management of TCP IP Based Internets TREND 3 6 User s Guide Additional Publications Case J M Fedor M Schoffstall and J Davin A Simple Network Management Protocol RFC 1098 University of Tennessee at Knoxville NYSERNET Rensse laer Polytechnic Institute Proteon August 1988 Rose M T The Simple Book An Introduction to Internet Management Second Edition PTR Prentice Hall 1994 Stallings W SNMP SNMPv2 and CMIP The Practical Guide to Network Man agement Standards Addison Wesley October 1993 Waldbusser S Remote Network Monitoring Management Information Base RFC 1271 Carnegie Mellon University November 1991 Bowman J Emerson S Darnovsky M The Practical SQL Handbook Second Edition Addison Wesley 1993 Online User Manual Availability TREND is shipped with online viewable user ma
316. ith TRENDabview TREND 3 6 User s Guide To add qualification text from the Columns in Selected Table Box select the appropriate table in the Select Data Tables Key Tables Box Its defined columns will appear in the Columns in Selected Table Box Click the column you wish to insert as qualification text The text appears in the Qualification Text Box at the point where the cursor is located To add qualification clauses from the Clauses Box simply scroll to the clause you want and click it The clause appears in the Qualification Text Box at the cursor point 3 Once you have defined the qualifications you wish to apply to your view click OK The Define Qualifications window disappears and you return to the TRENDdbview main window Select Columns for a View N N D m z o ex o i 1 Click the Select Column Button The Columns window appears with your selected data and key tables in the Selected Data Key Tables Box In addition you will see thirteen 13 default columns in the Columns in View Box Note Do not delete the default columns They are necessary if the view is to function properly 2 Highlight one of the tables in the Selected Data Key Tables Box Its columns appear in the Columns in Selected Table Box 3 Select one or more columns from the Columns in Selected Table Box and click the Insert Replace Button The highlighted columns appear in the Col umns in View Box You can select
317. key id equals 04 00 00 AM on the previous day It concatenates user name and the first three characters of the ta period value in Rsi dbstats and places the result in the dsi descr column in Ksi_dbstats_ It creates a column named dsi eurodate in Ksi dbstats if the column does not already exist and unconditionally updates that column with the result of concatenating positions 5 7 1 4 and 8 11 in ta period 19 10 Examples TREND 20 TRENDexporter This chapter describes the TRENDexporter utility This application allows you to run a TRENDsheet report against your database and to store or export the output in an ASCII text format The document has four parts Overall Process and Functionality on page 20 1 Controls and Switches on page 20 3 Input Parameter File on page 20 6 9 9 Output Format on page 20 14 Overall Process and Functionality trend export generates ASCII files of data extracted from TREND using the SQL query defined in any TRENDsheet query file The ASCII file generated can be saved in any location In short here is the process involved 20 1 N j D m 2 o D x o P 7 20 TRENDexporter 20 2 1 Generate a source TRENDsheet query file using TRENDbuild and TRENDsheet 2 Invoke the ASCII generator utility pointing to this source query file The utility makes use of the SQL generated by TRENDbuild and TRENDsheet and saved in a query file with a q
318. l This removes all of the defined expressions 5 Once you have defined the expressions you wish to apply to your view click OK The Define Expressions window disappears and you return to the TRENDdbview main window Once you have selected the appropriate tables and define the qualifications expres sions and columns you wish the parameters procedures are complete It only re mains to name save and run the view Name Save and Run a View 1 From the View MIB Combo Box select one of the available MIBs preferably the MIB of a data table that has contributed at least one column to the view 2 Inthe View Name Text Field type the name you wish to use to identify your view By performing these two steps you define the name of your view This name becomes important when you run the view script which creates the view in your database The name of a view takes the form of MIBname View viewname where MIBname is the MIB you selected from the View MIB Combo Box and viewname is the text you typed in the View Name Text Field If for exam ple you choose mib II from the View MIB Combo Box and type routers1 in the View Name Text Field then save and run the view the name of the view appears in mib ii as mib II view routers You can use Save As to give the view any file name you wish 3 From the File pull down menu choose Save The view is saved as an execut able file to the default directory 22 18 Wo
319. l be used to control rollup and aging for tables that do not have specific values set see Changing System wide Rollup and Aging Defaults on page 2 14 The Change Database command provides access to data tables in other databases Tables are read out of the S YBASE interfaces file View Show by Source Show by Type Sort Find Object Unselect All Refresh Select Show by Source to have Data Manager only display values for tables from the selected source Select Show by Type to have Data Manager only display values for tables of the selected type Select Sort to sort the table rows in the specified order Select Find Object to search Data Manager for a particular table by whole word or partial word Select Unselect All to deselect table rows highlighted Select Refresh to update the Data Manager display with current information on the state of the database The information is not updated on the fly you must use the Refresh command to view the latest changes o E o o 3 0 Q D Main Display and Menu System 8 5 8 Data Manager Tools TRENDsheet TRENDgraph View the datain a table by selecting the appropriate row then enter TRENDgraph or TRENDsheet directly from Data Manager via the Tools menu Obtaining Table Profile Information By double clicking on a row in the table you can view information that describes each column in that table full col
320. l before assuming the target node is not going to respond The default is 5 If the target node does not respond by the r poll rmon collect will make a No response from host entry in SDPIPE HOME lib trend log and go on to the next target The time rmon collect waits between retries is set by the o op tion For devices that are slow to respond a retry count of 3 combined with a longer timeout value can improve the success rate of your polling rmon_collect S This is used to round off the collection time if started later than the ex pected time Ifthemw collect parent kicks off a collection at 3 07 and if using the default option of 300sec 5 min the actual ta period value for the collection will be 3 05 t The name of the MIB table which you want to collect The raw data ta ble must already exist in the database The tables you can use with this option are shown in Data Manager with type data Use the value in the Table Manager Object Name column as the argument u Displays the command line formats for rmon_collect V Displays the version stamp for rmon_collect W Stops the collection of data when the database used size reaches the specified percentage The collection routine checks the dbstats tables to estimate the current state of the database full size and determine if it should write to the database The default is 90 for 90 Examples Interval Polling To have rmon_collect execute all collection
321. l the name is unique and that becomes the table name If you are using a MIB file named framerelay dte rfc1315 mib and you create a MIBlet named frCircuitEntry the resulting Object name is framerelay dte rfc1315 frCircuitEntry the resulting SQL name is x5 frcircuitentry Figure 6 11 If you are using a MIB file named test MIB name with CAPS amp 30char mib and you create a MIBlet named test miblet sk mib2 the resulting Object name lstest MIB name with CAPS amp 30char test miblet sk mib2 the resulting SQL nameis st miblet sk mib2 Figure 6 11 Data Manager OP x File View Tools POWDER db used 0 Object Name SQL Name test MIB name with CAPS230char test miblet sk mib2 st miblet sk mib2 framerelay dte r c1315 frCircuitEntry Figure 6 11 MIBlet Name Examples in Data Manager Preparing MIBs for MIBwalker TREND 3 6 User s Guide Note If a vendor has multiple versions of their MIB the version number should be preserved in the MIB file name before the mib extension To accomplish this append the version number to the file name with a character other than a period For example version 1 1 of a MIB called Desktalk mib is identified by naming the file Desktalk 1 1 mib Desktalk1 1 mib or Desktalk_1 1 mib If the file is called Desktalk 1 1 mib or Desktalk 11 mib the 1 1 mib or 11 mib component is dropped from the name of the database table Add the version number after the name so th
322. le table a in the Selected Tables Box of the Select Tables window If you choose to define any parameters you can do so in any order We have chosen to describe the procedure to define qualifications first but this is merely random Once you have selected your tables you can perform any of the remaining procedures in any order you choose Define Qualifications for a View 1 Click the Define Qualifications Button The Define Qualifications window appears with the default qualifications showing in the Qualification Text Box If you do not wish to apply the default qualifications click Clear and the Qual ification Text Box is emptied The Clear Button removes all qualifications If you make modifications to the qualifications Text Box then decide you wish to restore only the default qualifications click Default Qualification The Qualification Text Box will return to its default setting If you add qualification text then decide you don t want it you can select and delete it as you would in a typical text editor 2 To add qualifications click within the Qualification Text Box A cursor will appear Any text you add will appear at the site of the cursor so be sure to position the cursor at the point where you wish to insert your modifications You can add text by typing directly into the Qualification Text Box by adding text from the Columns in Selected Table Box or by adding clauses from the Clauses Box Working w
323. le A 33 Host 16 2 host group RMON 16 3 host name 5 11 5 14 translation 5 3 hostEntry 16 3 How TRENDproc Works 21 1 Identifier 6 4 ifspeed 9 12 9 14 Ignore Case 6 9 import nodes 3 4 inaccurate data 9 8 inoctets 9 5 9 12 9 14 input parameter file 20 2 20 5 20 6 installing a ReportPack 4 1 instance 15 5 16 3 16 4 16 6 20 2 INTEGER 6 14 interface file 2 4 2 5 speed ifspeed 9 12 inter sample delta values 8 3 interval polling A 32 intervals 16 5 invalid protocols 17 13 IP address 5 2 5 3 5 9 5 11 5 14 5 15 6 29 Discovery 5 11 discovery 5 2 E 2 name 17 10 Range 5 10 range 5 12 5 14 ISO Tree 6 11 J Java enabled web browser 1 13 K Keep Data 8 4 Keep Data for 8 3 key 11 10 12 10 13 10 15 4 18 7 20 9 license 2 19 table 10 4 16 4 16 9 18 7 19 1 22 1 22 2 22 3 22 5 22 9 22 12 22 13 Index 9 Index Index 10 key table A 10 A 69 C 16 keystats utility A 10 keyword 14 13 20 6 20 7 20 9 condense 20 9 frequency 20 9 max 20 9 style 20 9 L label 6 4 12 19 13 12 13 18 Large Fonts 3 2 last value collected 9 9 launch report 14 4 license F 3 attributes 2 19 database size 2 20 key 2 19 B 3 D 2 process 2 19 manager 2 22 D 2 D 7 server 2 22 licensing mechanism B 3 limit 22 3 limits of IP Address range 5 9 line definition 12 9 12 16 line width 12 15 Lines to graph box 12 16 link related reports 15 3 Load Mib option 6 6 log scale 12 19 log_backup
324. le graph In the Select Data Set window click on Add Line Continue to select keys and ag gregation types and add lines until all lines have been specified for the graph Note The Select Key window must be open to ADD or UPDATE lines in the Select Data Set window EXAMPLE Selecting Data for an Average Input vs Output Utilization Report Using the pull down arrow in the Select Statistic box recall the Input_Utilization expression created previously created in the Assign Statistic Aliases window Click on Select Key to invoke the Select Key window Creating a New Graph TREND 3 6 User s Guide Use the pull down arrow in the Device box to highlight the node from which you col lected data Select a Key by clicking on the desired row in the Key Description table Select average from the Aggregation Type pull down Click on Add Line in the Se lect Data Set window Repeat the above steps for Output_Utilization First select Output_Utilization using the Select Statistic pull down arrow in the Select Data Set window Displaying the Graph A basic graph can now be displayed Line bar and area formats are available and should be selected according to their ability to convey the information in the report At the bottom of the Select Data Set window point and Click the Line Bar or Area graph box You can choose to define graph attributes before or after displaying a re port for the first time by following the paragraphs
325. le identifies the type of device it is SNMP Type Discover populates this table by interrogating each new node in dsi_nodes and adding an entry to dl_type The dsi_nodes table contains a row for each recognized node on your system In addition one column of dsi_nodes identifies whether a node is SNMP or RMON default is SNMP When SNMP Type Discovery identifies an RMON manageable device it updates this column of dsi_nodes What Discover Does 5 5 5 Discover See Discovery Considerations on page E 1 for additional technical information about the dis files and type discovery Using the Discover Graphical User Interface TREND provides a Graphical User Interface GUI for its Discover tool This sec tion describes how to use this GUI to run IP Discovery Only IP Discovery can be run using the Discover GUI SNMP Type Discovery must be run from the command line Note IP Discovery can also be run from the command line Instructions for how this may be done are provided in Running Discover from the Com mand Line on page 5 11 Launching the Discover Interface The TREND Main window includes a button to launch the Discover interface When you click the Discover button the Discover tool appears as shown below 5 6 Using the Discover Graphical User Interface TREND 3 6 User s Guide c o ao 7 e lt D mg 134 70 65 0 134 70 65 255 Figure 5 1 The TREND Discover Interface Note If y
326. le in the window s header bar To load another MIB do the following 1 Select File gt Load Mib The following screen appears showing the DPIPE_HOME mibs directory by default Select Mib 7 x Look in E mibs zl e C Reference Only E framerelay dte rfc1315 E wf application 780 E cisco100 E hostresources fcl514 E wf datalink 780 E ctron rev22 amp mib ll E wihw sw sys 780 E ctron rev4 E repeater fc1516 E wf internet 780 153 dsi cisco interface E vines El wf line 780 E ethernet ric1 284 E wellfleet 560 E wf wan 780 Fienme Files of type Files mib Cancel Figure 6 10 Select MIB Window 2 Select a MIB The MIB name appears in the File Name field Click Open The newly loaded MIB name appears in the window s header bar and the MIB appears in the MIB map Loading a MIB TREND 3 6 User s Guide Preparing MIBs for MIBwalker MIBwalker uses standard Structure of Management Information SMI compliant SNMP MIBs Raw MIB files must be modified to comply with typographic and syntax rules set by the MIB parser Object types must be validated as well and the MIB must be properly named Once the MIB meets these requirements it can be loaded successfully Typographic Rules Lines in a MIB file must be terminated by a carriage return CR and a line feed LF Lines cannot be terminated by a CR alone o ius 2 x D Syntax Rules The following MIB syntax
327. le is the one for dsi_key_id where the maximum ta_period timestamp is less than or equal to the current time with hours and minutes zeroed minus 1 hour For example if the current time is 10 30 AM on June 12 1998 1 Get the dsi_key_id record in the source data table where the ta_period timestamp is Jun 12 1998 09 00 00 AM 2 If no record exists for 09 00 00 TRENDlabel decrements the hour by 1 until a match is found For example if there is no row for the dsi_key_id where the ta period time is 09 00 00 TRENDlabel looks for a time of 08 00 00 decre menting the hour by in this manner until a match is found Ensuring Key Table Data Table Compatibility You must ensure that the specified data table t option uses the specified key table k option in the TREND database TRENDlabel does not check for this Update Restrictions You cannot update the values in the key table columns dsi key id dsi target name or dsi table key Extracting Substrings from Column Values When you use the e c and r options you can specify that a substring be extracted from the source column and used to populate the destination key table column with the following syntax column offset length 19 6 Usage Notes TREND 3 6 User s Guide where column Is the column name in the source data table offset Is the starting character position of the substring in the column length Is the number of characters to include in the substring beginni
328. leaned source for values that satisfy one or more predefined condi tions TRENDstep can conditionally process groups of rows This capability can prevent divide by zero conditions help process RMON2 data and grade device performance TRENDsum processes the data in a ReportPack s base table and groups data by the element properties and time periods thereby creating the levels of summari zation and consolidation needed for the reports TRENDsum calculates standard statistics such as counts and averages and more complex metrics such as utiliza tion forecasts and rolling baselines used to compare element performance TRENDrank ranks performance according to defined categories such as uti lization congestion or Grade of Service GOS so you can keep track of things like the top ten best or worst performers TRENDrank also tracks the change in an element s ranking so you can identify stable and volatile resources easily o TREND 3 6 User s Guide So 2 o Led z0 2 gt Figure 9 1 summarizes this processing E e c Co D Q zo EE A L os gt 8 Delta Base Raw Data TRENDit Data TRENDstep Table TRENDsum TRENDrank Summary Tables Volumes Exceptions Forecasts Baselines Figure 9 1 Data Aggregation and Manipulation Process The remainder of this chapter provides an overview of how these processing func tions are performed and relates the processing to TREND supplied ReportPacks Man Pages on
329. lected objects can be polled together in a single table This is known as a MIBlet This section provides a description of the steps for creating a MIBlet 6 20 Creating a MIBlet TREND 3 6 User s Guide A CAUTION Note You do not need to create a MIBlet to run any of the existing ReportPacks provided by TREND Necessary MIBlets are created automatically When creating a MIBlet you must be sure to select only objects that are commonly indexed MIB objects in the same table are by definition commonly indexed MIB ob jects from different tables may be commonly indexed For example some enterprise MIBs contain tables indexed by interface number just like MIB II ifEntry is indexed Creating a MIBlet containing objects that are indexed differently results in invalid data in your database Note The maximum width for a table created from a MIBlet is 2 000 characters If you create a table that exceeds this width a diagnostic message appears when you run TRENDit Perform the following steps to create a MIBlet 1 Decide which variables MIB objects you wish to include in your MIBlet and select those objects by clicking each one with the right mouse button Creating a MIBlet 6 21 2 ius 2 x D 6 MIBwalker Figure 6 14 shows that a box appears around each selected object Node powder MIBwalker E TREND mibs mib Il_mib r iflndex ifDescr ifType m ifMtu ifSpeed ifPhys
330. lecting Settings from the Start CAUTION menu and then Control Panel Display Settings Make sure the Font Size is set to Small Fonts 3 2 Display and Menu System TREND 3 6 User s Guide The menu bar and application access buttons provide access to all TREND user vis ible GUI applications and functions The buttons invoke the following applications which are described in detail in their respective chapters in this guide Application Chapter Autopilot 4 Discover 5 Collect Data 7 Report Launcher 14 Report Scheduler 15 Data Manager 8 All TREND application modules are brought up under your user name with your environment Some menu commands e g table aging truncation and deletion from Data Manager can only be executed by the trendadm user Users who are not trendadm will see these commands grayed out indicating that they are unavailable Menu Command Summary The menu bar has File View Configuration and Tools pull down menus The File Menu View Configurati Open Irmgard desee Exit Display and Menu System 3 3 o o E 0T o O 3 3 The TREND Main Window The File menu contains four options Open Import Nodes Exit and About Open Exit and About are all standard commands and so are described in Commonly Used Commands on page 1 15 Import Nodes 3 4 The Import Nodes command allows the user to import nodes views and types via an ASCII file
331. lete Entry Node View Type User Frequency rmon all ay rmon trendadm i mib ll trendadm i cisco100 trendadm i cisco100 ifentry trendadm i framerelay dte rf rg i rmon monar fro i5 mnues rmon2_praber frengaam 20 Minutes rmon2_probet rmon all trendadm 1 Day ecam Figure 7 1 Collect Data Window Menu Command Option Summary The Menu Bar of the Collect Data window has three pull down menus File View and Tools 7 2 Collect Data TREND 3 6 User s Guide The File Menu The File menu contains four options Collect Data Periodically Change Database Exit and About Exit and About are standard commands and are described in Commonly Used Commands on page 1 15 Collect Data Periodically Use the Collect Data Periodically option to define new data collection parameters for a selected table see Specifying How Often to Poll for the Data on page 7 7 Change Database Select the Change Database option to view and collect data from tables in alternate databases see Changing Databases on page 7 9 The View Menu N e o o o D r 5 The View menu currently has only one option Sort Order Sort Order is used to select the sort order of the polling information from the list of items available see Reordering the List of Collection Requests on page 7 11 Collect Data 7 3 7 Collect Data The Tools Menu I n ol s TRENDs
332. lling agent rmon_collect supports polling of RMON data The user can select a polling interval for individual nodes or groups of nodes rmon_collect supports the Ethernet and Token Ring RMON MIBs for History Host Matrix and Statistics groups TREND assumes that the RMON probes or agents have already been configured using a dedicated RMON real time application typically provided by the RMON probe vendor TREND automatically adjusts to RMON probe configuration changes and no integration with other RMON applica tions is required Group Description Statistics Statistics measured by a probe including the number and size of packets broadcasts and collisions MAC layer and promis cuous versions of Token Ring RMON Statistics tables also supported History Periodic statistical samples recorded over time used to analyze trends MAC layer and promiscuous versions of Token Ring RMON History tables also supported Host Statistics of the hosts on the network including MAC address es of active hosts Matrix A matrix of statistics that tracks conversations between pairs of hosts Once TREND creates the host table it can be sorted on any field in ascending or descending order with the desired number of entries selected Also all the statistics 16 2 RMON Group Support TREND 3 6 User s Guide in the selected hostEntry table rows are available to the user These abilities exceed the limitations of and ther
333. lowing scripts from the DeskTalk Systems FTP site and use them as an alternative to the trendit command TrendIt sh TrendItLoop sh To use these scripts 1 Place the downloaded scripts in the SDPIPE HOME bin directory on the satel lite server where TRENDit is to execute 2 Replace the TRENDit commands in the appropriate trendtimer sched file with one of the following commands Trendit sh S r Z error 96 t table name ie D o lt i o m 3 Z E E oO ko 3 This command invokes TRENDit to calculate delta data from the raw table specified by table_name and generates some logging information Include a TrendIt sh command line for each raw table_name from which you want to calculate delta data apjayes e dp bumas a TrenditLoop sh s r Z error_ This one command invokes concurrent instances of TrendIt sh to calculate delta data from all raw tables Note These scripts are available for UNIX environments only For NT envi ronments you must use a form of the TRENDit command described above Data Management on the Satellite Server D 3 D Setting Up a Satellite Server Environment Using TRENDcopy You can use TRENDcopy in one of the following ways To push the data from the satellite server onto the central server That is run TRENDcopy on the satellite server according to the scheduling information given in the trendtimer sched file residing on the satellite serv
334. lowing RMON data to be inserted directly into the database There are however two unique components of the RMON2 DataPipe The RMON collector dpipe rmon2 is the tool that polls an RMON2 manageable device and collects the RMON data the device has stored The Protocol Definition Interface is a graphical user interface that allows TREND users to populate the dsi protocols table These components are described in the sections below DataPipe Components 17 5 Lj N E fe 2 N 7 l Ko 3 17 RMON2 Support The dpipe rmon2 Polling Agent How TREND polls a device node is determined by how that device is defined in the dsi_node_list table If a device is defined as being an RMON capable device TREND launches RMON collect which is a parent collector The parent collector does not perform the actual polling of the device Instead the parent collector deter mines precisely what kind of device the target is and launches the appropriate child collector which in turn polls the target device In the case of RMON2 the child collector is named dpipe rmon2 The dpipe rmon2 collector collects the data stored by the RMON agent and stores that data in the TREND database The Protocol Definition Interface 17 6 When dpipe rmon2 collects RMON data the data is organized by protocol In this context a protocol defines an exact description of how two devices communicate with each other RMON make
335. lter on protocols and applications so a user can create a domain that only collects statistics about IP traffic for example and by ap plying that IP domain to the history table see history statistics generated exclusively by IP traffic on the monitored network TREND understands Netscout RMON probe domains and allow the user to create reports that take advantage of them The key tables for the RMON data tables in clude a filter field By creating a report which specifies the name of the filter a report can isolate and display domain specific information For probes which do not support domains TREND uses the value ALL in the filter field This value implies that the data is not limited to a specific domain Note In Collect Data the Tools button has a TRENDgraph and TRENDsheet option Each of these allow you to do a quick unformatted dump of the data in the selected mib group s raw rate hourly daily or weekly table For RMON the table you would see by doing this is the control table rather than the data table and consequently by default only raw data will be available See About Data and Data Tables in TREND on page 10 2 pon D fe 2 7 Ko E Domains 16 9 16 RMON Support TREND 16 10 Lj N E O 2 N 7 I Ko 3 TREND 17 RMON2 Support The RMON2 MIB is an extension of the RMON MIB It includes all the data objects defined in the RMON MIB as well as
336. lues ciated delta times Maximum Maximum of the counter Maximum of the input data values divided by their values associated delta times Total Total of all the input values Average of all the input val ues It is not sensible to total gauge data Average Total of all the input values Average of all the input divided by the total delta time values Data Cleaning 9 9 v 5 2 D O E D en 5 uonebaibby ejeq 6 9 Data Aggregation and Manipulation 9 10 In the TREND database you can tell what type of data a table contains by the first letter in the SQL name of the table For example ji Contents SQL Table Name Raw mib II ifentry R Rate Rib ii ifentry H or SH Hourly Hib ii ifentry DorSD Daily Starts on a particular day of the week Dib_ii_ifentry W or SW Weekly Wib_ii_ifentry M or SM Monthly for a calendar month Mib_ii_ifentry Tables can come from any source but must be properly typed to ensure that they are processed properly The main Data Manager window displays this table name in the SQL Name column the type of data in the table raw rate hourly daily and so on appears in the Type column See Data Manager on page 8 1 TRENDit provides the default set of roll up and summarization processing functions described above for The ReportPacks in TREND releases earlier than release 3 5 Any data table for which Yes is s
337. m 7 11 Interval 7 7 poll dynamic list of nodes A 23 nodes A 32 user defined groups of devices from a single data collection request 7 13 poll_type 3 4 Polling RMON tables 16 4 polling 7 1 7 4 7 10 agent 16 1 16 2 agents 7 4 7 8 by device type 7 14 Collect Data Periodically option 7 3 cycle 7 6 7 15 deleting request definitions 7 13 edit modify 7 13 frequency 7 15 group 7 13 7 14 interval 7 7 16 2 mode 7 10 all nodes in view 7 7 all nodes of same type 7 6 all nodes of same type in view 7 7 individual node 7 6 organizing 7 13 7 15 parameters 7 5 7 13 period 16 7 policy 1 4 1 9 7 8 common 7 7 SNMP 1 9 poll from line 7 8 process mw collect 7 15 rmon collect 7 15 request 7 15 RMON history group 16 5 Sort Order option 7 3 specifying which system should poll 7 8 strategy 7 13 target 7 6 utilities DataPipes 9 1 polling efficiency 17 15 polling frequency RMON 16 4 polling interval 16 5 16 6 A 33 polling mode individual node by key 7 6 polling policy 7 1 A 26 D 7 populate ranked table A 48 populate raw data tables 1 10 port number 17 10 postscript 11 3 11 21 12 3 12 22 13 21 15 3 postscript file 15 7 pre defined TREND report 11 4 prevent divide by zero conditions 9 2 previous reporting period 14 4 Print option 1 18 to File 1 18 printing reports Scheduling with Report Scheduler 15 1 Index 13 Index Index 14 privileged access 1 14 probe 16 6 16 9 17 9 procedure Create MIBlet 6 24
338. m Administration Guide for additional information regarding the use of the Backup Server On the NT platform a backup server is automatically created during Sybase installation On the UNIX platform however you must use an additional utility to install a backup server The utility depends on the version of Sybase you are running For Sybase 11 9 2 use srvbuild See Installing a Backup Server for Sybase 11 9 2 Systems on page B 15 Installing a Backup Server for Sybase 11 9 2 Systems When you install Sybase a backup server named SYB BACKUP is created by default If you want to create a backup server with a different name use the srvbuild utility on UNIX systems Follow this procedure 1 Login as root 2 Bring up a terminal window 3 Ensure that the SSYBASE variable points to the directory in which Sybase is installed 4 Change to the directory in which Sybase is installed cd SSYBASE bin 5 Invoke the srvbuild utility srvbuild ga o 2 9 0 g TO 9 D O o m 3 Backup Server B 15 B Database Administration The Select Servers to Build window appears 6 Select Backup Server as the server type The default name of the backup server appears in the Server Name list box We recommend that you accept the default name 7 Click OK The Server Attribute Editor window appears B 16 Backup Server TREND 3 6 User s Guide Default values for the required fields are filled in for
339. m name Trendtimer is the master scheduler program and it is associated with normal process initiation and process completion Date and time the process starts Process id number The system assigns a process id number to each process Process name File name and path of the process that was started by trendtimer N E m 032 ag no 3 o as oa fo Q z E P i o 3 o mm de Process options that are associated to the process specified in step 5 See Man Pages on page A lor the appropriate chapter in this document for more information 7 Process completion message This message shows when a process completes General Message Format All messages specify the program name followed by the date and then the message in the format program name date message An example of the complete log entry follows trendtimer Fri Apr 10 01 00 02 1998 Process id 5678 terminated 0 sec where Message Handling 2 25 2 TREND Configuration for trendadm 2 26 program name is trendtimer date is Fri Apr 10 01 00 02 1998 message is Process id 5678 terminated 0 sec Every time trendtimer initiates a process mw collect rmon collect trendit Ipr launch db delete data etc an entry appears in the log Periodic or regular processes appear every time the specific process is called such as at regular intervals of 5 10 15 20 30 or 60 minutes An entry appears when the process initiates and when the process
340. m name date message The following example shows the log entry of an error message trendit Fri Apr 10 01 20 01 1998 Server net xyz com 30 No licenses currently available 2 28 Message Handling TREND The TREND Main Window The Main window provides a point and click graphical user interface for access to TREND applications and reports It is the control panel for TREND invocable from a command line or from any configurable command menu The Main window also allows the TREND system administrator trendadm to con figure the system and to take advantage of the advanced features of TREND includ ing distributing polling setting user privileges defining interface descriptions and utilizing multiple databases o o Ss E o2 o D Note You can get started using the basic features of TREND without custom configuration or trendadm expertise 3 1 3 The TREND Main Window Display and Menu System When you invoke TREND using the TREND executable the Main window appears on your screen F TREND Pls ES File View Configuration Tools Autopilot Discover Collect Data Report Launcher Report Scheduler Data Manager Figure 3 1 The TREND Main Window In a Windows NT or Windows 95 environment you can lose information on the TREND main window if Large Fonts is specified for the windows Font Size display settings option Check the Font Size setting by se
341. m the devices in your enterprise TREND DataPipes can collect raw data from Simple Network Management Protocol SNMP Remote Monitoring RMON RMON2 and other agents depending on the device The names given to the raw data items and their data types are determined by the Management Information Base MIB associated with the devices being polled Most collected data is not readily usable in its raw form The raw data must be processed to make it reportable in terms that provide usable performance information about the enterprise Four TREND utilities work together to process the raw data consolidate it and finally display it in the reports supplied with the TREND ReportPacks TRENDit performs some validity checks on the raw collected data and turns SNMP counter data into delta data This process is called cleaning TRENDit can also summarize the data in standard ways More specifically TRENDit can compute counts averages minimums and maximums and sum marize an element property at the hourly daily weekly and monthly level This process is called rollup 9 1 o 5 2 D O E D uonebaibby ejeq 6 9 Data Aggregation and Manipulation 9 2 TRENDstep creates the table on which all reports in a ReportPack are based This table is called the base table TRENDstep creates the base table for each installed ReportPack To create the base table TRENDstep examines each row of data that is input from a c
342. mat for the messages that appear in the log Example The following example shows a portion of a log where processes initiate and complete The highlighted items show two examples of processes initiating and completing note that each pair has the same process id Pid number o O trendtimer Fri Apr 10 01 05 01 1998 Pid 6630 usr TRENDsnmp bin rmon_collect n i 5 p10 trendtimer Fri Apr 10 01 00 05 1998 Process id 6323 terminated 5 sec O trendtimer Fri Apr 10 01 05 01 1998 Process id 6630 terminated 0 sec trendtimer Fri Apr 10 01 03 04 1998 Process id 6328 terminated 183 sec trendtimer Fri Apr 10 01 01 35 1998 Process id 6330 terminated 93 sec 2 24 Message Handling TREND 3 6 User s Guide trendtimer Fri Apr 10 01 10 00 1998 Pid 6668 usr TRENDsnmp bin mw collect n i 5 p10 trendtimer Fri Apr 10 01 10 01 1998 Pid 6670 usr TRENDsnmp bin mw_collect n i 10 p10 trendtimer Fri Apr 10 01 10 02 1998 Pid 667 1 usr TRENDsnmp bin rmon collect n i 5 p10 trendtimer Fri Apr 10 01 10 02 1998 Pid 6672 usr TRENDsnmp bin rmon collect n i 10 p10 trendtimer Fri Apr 10 01 05 03 1998 Process id 6629 terminated 3 sec trendtimer Fri Apr 10 01 10 02 1998 Process id 6670 terminated 1 sec trendtimer Fri Apr 10 01 10 02 1998 Process id 6671 terminated 0 sec 1 Process initiation message This message shows when a process starts 2 Scheduling progra
343. med that the database on host B looks complete you can turn on trendtimer on host B to see how things work Check the following Policies You must change the value of the Poll From field to the new host name for each policy listed Verify that your text change is correct by using it with a ping command 2 Environment variables for trendadm and all other users Make sure that the files of DPIPE_HOME lib Cshre and DPIPE_HOME lib Profile have been changed either by new installation of by editing to have new settings specifi cally DISPLAY DSQUERY and DSI ELMHOST TREND 3 6 User s Guide Database revisions You must change the dsquery column in the tl_groups table for all entries to be the new Sybase environment variable value for the host B system You can do this with ISQL using the following commands isql Udsi dpipe Pdsi dpipe 1 update tl groups set dsquery sSybase DSQUERY value 2 data db Sybase DSQUERY value 3 gt go Reporting If you have installed packages with AutoPilot or have added reports then copy the following files and directories from host A to host B DPIPE_HOME lib print files DPIPE_HOME lib info files DPIPE_HOME reports directory SDPIPE HOME scripts directory m Make sure that any custom report files qss qgr gos which can reside in any directory on host A are copied to host B 2 Make sure that any custom scripts which can reside in any directory on
344. mib ii ifentry table Similarly use the alias name for the source table column in the column or sub variable statement when that column belongs to a data table When it belongs to a key table use the name as known to the database since there is no group name alias for a key table column TRENDrank Sub Variables Sub variables provide a means to restart the rankings within a given date much the same as a subtotal restarts a summation With no sub variables TRENDrank assigns the rankings according to all the values for the specified metric on a given date With some sub variables TRENDrank assigns the rankings within the same values for the sub variables Upon a change in any sub variable for the date the ranking begins again TRENDrank As an example suppose there are 10 values for the metric in question on a given date With no sub variables TRENDrank assigns the rankings from to 10 How ever if you specify cust_id as a sub variable TRENDrank assigns the rankings be ginning at one to those dsi_key_ids that have the same value for cust_id When cust_id changes the rankings begin again at one for the new cust_id Each sub variable must exist as a column in the key table While similar to the by variables used by TRENDsum sub variables differ in that they do not cause any update of the rank table s key table nor do they affect the grouping of rows for the rank table They affect just the contents of the current rank colu
345. mns in the rank table Example Assume the following statements are stored in a file named test1 rnk source day pvc SDsummary target day pvc rank column in util in utilization d column out_util out_utilization d column errors a Enter the following command to execute the TRENDrank utility using testl rnk as input trendrank f testl rnk gt D v o Q o o This command ranks data from a summary table named day pvc SDsummary and places the ranked columns in an output table named day pvc rank The rankings for the output columns associated with in utilization and out utilization are in descending order The ranking for the output column associated with errors is in ascending order Note that there is a column with the name errors in both the source and target tables so it is not necessary to specify the target alias for the errors column A 51 TRENDstep TRENDstep TRENDstep conditionally processes each row of a TRENDit input table and stores the result in a new output table This processing enables you to combine data from multiple rows of an input table into a single row in the output table and compute new columns of data in the output table trendstep b bcp packet size d debug level filename l logsize high water mark m dbsize high water mark u v Options TRENDstep options are b Specifies the bulk copy packet size which can be from 512 to 8192 bytes d Sets
346. mpt you to run SNMP Type Discovery and will launch the program for you if you choose At this time there is no GUI that allows you to define SNMP Type Discovery Running Discover from the Command Line 5 17 5 Discover 5 18 Launching SNMP Type Discovery SNMP Type Discovery is a variation of the Discover tool which is launched from the command line as trend_discover To launch SNMP Type Discovery use the syntax trend_discover t This starts the Discover program and specifies an SNMP Type Discovery SNMP Type Discovery Command Line Options The only valid command line arguments for SNMP Type Discovery are trend discover d debug level gt H help p lt max entries per pdu default 20 gt P lt snmp port number default 161 gt r lt SNMP retries gt t do type discovery u lt username gt E echo progress F lt type definition file gt o lt SNMP timeout in seconds gt v lt display version gt Running Discover from the Command Line TREND 3 6 User s Guide oa g Option Descriptions o 2 o d Specifies the debug level F E Causes the progress of Discover to be displayed on the screen F Names the type definition file Can only be used in conjunction with the t option H Causes a list of available Discover command line options to be dis played on your screen 0 Determines how long Discover will wait before timing out an SNMP GET request p Resets
347. n lt blockname gt In addition this statement de fines the wait nowait condition of the block The syntax for the beginning of the nowait block pintail is as follows begin lt pintail gt nowait Each block must have a unique name A block can contain one or more statements to be executed sequentially These statements can be The fully qualified path of an executable file The command line execute command of a TREND application A block does not need to contain a command In this case it is an empty block This will not affect the performance of the TRENDproc so long as the populated blocks are formatted correctly A block ends with the statement end blockname Creating and Applying a TRENDproc 21 9 21 TRENDproc Note Putting the block name in the End statement is optional It can however be useful for identification purposes especially if you nest one TRENDproc inside of another The words begin and end must be in lower case Comments within a block should begin with the number character There should not be a space between the begin end and the colon 9 9 The default option for wait nowait is wait If you do not specify wait or nowait on the begin line it will default to wait Scheduling TRENDproc 21 10 After you ve created your TRENDproc file you need to place an entry in the trendtimer sched file Otherwise TRENDproc will not launch at the appropriate time The syntax of
348. n a network segment This is accomplished via an RMON agent which typically runs on a net work device A device supporting an RMON agent is able to store significant infor mation about the network devices that are talking to each other RMON is also the name of the defined data structure MIB by which RMON agents order and store the data they collect Thus RMON agents are those agents that col lect and store data defined in the RMON MIB TREND components provide data collection aggregation and reporting capability for network environments utilizing RMON probes RMON data presents added challenges to the collection and storage of network data due to its high volume and dually indexed table structure The primary difference between TREND s SNMP and RMON pollers are the data they collect and the meth ods they employ to collect the data The RMON polling agent has been designed to collect RMON data and convert it to a TREND format so that it can be processed and used for reporting by generic TREND modules TREND includes support for RMON network and application layer functionality via AXON Networks Enterprise Communications Analysis Module ECAM and mh o D fe 2 7 o 3S 16 1 16 RMON Support NETscout software Domain View extensions TREND supports all RMON probes and operates transparently to the user RMON Group Support RMON data polling is set up via the Collect Data application TREND s RMON po
349. n is in UPPERCASE Valid values for this option are 1 for SNMP V1 2 for SNMP V2C The default uses SNMP V1 packets which is 1 However if the table contains a 64 bit counter the poller automatically switches to SNMP V2C packets which sets the value to 2 The name of the MIB table that you want to collect The raw data table must already exist in the database The tables you can use with this option are shown in Data Manager with type raw Use the value in the Data Manager Object Name column as the argument See Data Man ager on page 8 1 for a discussion of the Data Manager Sets fast time conversion The default is on When this option is set to off the timestamp calculations are done using system time functions which could be very expensive This option is not available when run ning in ASCII mode This option is in UPPERCASE Displays the command line formats for mw_collect Displays the version stamp for mw_collect This option is in UPPERCASE Stops collection of data when the database used size reaches the speci fied percentage The default parameter is 90 for 90 This option to turns off the TRENDpm capability This option is in UPPERCASE Specifies which clock to use to calculate Delta Time The value of 1 directs the procedure to use System Uptime to calculate Delta Time The value of 0 directs the procedure to use the Agent Clock Column for the calculation The default is 0 This option is in UPPERCASE
350. nd tempdb2 dat for dpipe db2 dat in the command Enlarging the Database TREND 3 6 User s Guide 1 gt disk init 2 gt name dpipe_db2 3 physname usr300 Dpipe db dpipe db2 dat 4 vdevno 6 5 size 51200 65 go Note If you are enlarging tempdb skip step 7 7 Create the space for the extra log device The variables are similar to the ones in the previous disk init command 1 disk init 2 name dpipe log2 3 physname usr300 Dpipe tlog dpipe tlog2 1og 4 vdevno 7 5 size 10240 65 go 8 Map the newly created space to the database being enlarged The variables used in the command are database name Name of the database being enlarged Refer to the name in the select name from sysdatabases query in step 5 on name from disk init command Size of additional file in MB for that name If you are enlarging tempdb change the following command to read simply alter database tempdb on tempdb2 xxx 1 alter database dpipe db 2 on dpipe db2 100 log on dpipe 10g2 20 3 gt go Extending database by 51200 pages on disk dpipe db2 Extending database by 10240 pages on disk dpipe_log2 gt om 0 3D n5 uo OX m a0 om El Enlarging the Database B 7 B Database Administration B 8 Note The following step does not apply for tempdb Reset the required options on the database This step is only required if these options were originally set on the database being enl
351. nd other commercial applications see Exporting Data on page 11 18 The Change Database command provides access to a different SQL Server to dis play its version of the information presented in a report see Displaying Data from a Different Database on page 11 11 Quit the TRENDsheet application using the Exit command see Exit on page 1 19 and Selecting Statistics on page 11 6 The About command invokes TREND version information see About TREND on page 1 19 Time Period Sort Order Column Format Select Node Constraints Display Options Refresh ok J m 2 o 7 7 i e Time Period allows setting of the time range of data to be reported see Setting the Time Period on page 11 12 The Sort Order command allows sorting on different fields or in ascending rather than descending order see Sorting the Data in the Report on page 11 14 Main Display and Menu System 11 3 11 TRENDsheet The Column Format option allows you to change the precision of numeric col umns see Formatting Columns in the Report on page 11 15 To focus a report on a specific node and one or all of its keys for the table click Se lect Node see Select the Node and Key Being Reported on page 11 9 The Constraints command is used to specify thresholds to be used in exception re porting Only values that meet certain criteria e g exceed user specified er
352. ndow 0c eee eee eee 22 7 Define Qualifications Window 0c eee eee 22 8 Working with TRENDabview lisse 22 9 Starting TRENDdbview ssssleee nh 22 9 Building a VIeW 6 deo d ege dette wm Aves erate m dice pn 22 9 Create a New View 1 0 cee eee 22 11 Define the Data and Key Tables in a View 22 12 Define Qualifications for a View 0c eee eee 22 14 Select Columns for a View 22 15 Define Expressions for a View ee eee 22 16 Name Save and Run a View 000 a 22 18 Modifying an Existing View aaan aaa aaan 22 19 Open an Existing View 00 0 eee eee eee 22 19 Save Changes to an Existing View 0 22 20 Auditing Data and Generating Reports 22 20 Man Pages cun EE ened et EE NG REESE rede A 1 age files sese mx meme me E Rem Rm doe ee A 2 Description cns d eek de Exp eee ER A 2 Options oue em ea Be eser ia meer ey RR ee S A 2 db delet data sco ir umet at Ne ee aW A 4 Description else e mee RAS KANA e d ee A 4 xiv TREND 3 6 User s Guide Go rs P A 5 Examples 2 5 5 sciet halle nh pele th huit rae diee geste A 8 keystats aa oz mE ARR I Rede perd e RD A 10 Description cist aden a botnet adh Pues et Cp bU eps A 10 OptlOriS c oce tote e p RO ot el Rt ANNA RU RE A 10 log back ps siia strebt ELE RR AUR ERRARE ES DR REPRE A 11 Description zio ERE LYRSP RERTE
353. needs to be smaller or larger Bind new key tables to the named cache C 18 Sybase Configuration Parameters TREND 3 6 User s Guide TREND Processing Options Data You can improve throughput on machines with sufficient CPU and RAM resources by running processes in parallel The notion of parallel processing applies to collection copying of TREND tables TRENDit processing TRENDsum processing and db_delete_data processing In CPU constrained environments running multiple processes concurrently may result in an increase in run time and a decrease in throughput because additional processor overhead is associated with the operating system s task management DeskTalk systems has created scripts to implement parallel processing functionality for example scripts that invoke concurrent instances of a program each with specific arguments Contact DeskTalk Systems Inc via e mail at the following address for information on the availability of scripts for your platform and for download instructions support desktalk com Collection The mw_collect and rmon_collect programs have a command line switch c that allows you to control the number of concurrent collection processes By default this number is five You can reduce SNMP and RMON collection cycles by increasing this number for example to 25 If you change this number make sure enough user connections are configured for Sybase as discussed in Sybase Configurat
354. new view in one of two ways N N J m z o ex o By launching TRENDdbview if it is not already running The TRENDdb view main window appears with TRENDdbview untitled in the Title Bar By default this is a new view and you can begin the process of defining its characteristics Whenever the word untitled appears in the Title Bar the view is a new one By selecting New from the File pull down menu of the TRENDdbview main window If the name of an existing view appears in the Title Bar you need to select New from the File pull down menu in order to create a new view Once you have created the new view you must define its characteristics A view is composed of columns taken from existing data tables You can also in clude columns from previously created views 1 You define the view by Identifying the tables and or views that will make up the view Working with TRENDdbview 22 11 22 TRENDabview 22 12 Specifying which columns to include from the selected tables and views Defining any qualifications or expressions you want to apply You must select your data and key tables before making any other definitions After the tables have been selected you can define qualifications or expressions or select columns as you wish There is no set order for these procedures and you can skip from one to another 2 Once the parameters are defined to your satisfaction you nee
355. ng with the offset column position For example ta_period 13 8 means that the 8 character substring beginning in position 13 of ta_period be extracted from the column value Concatenating Column Values You can populate a key table column with a value that is concatenated from the values of multiple columns and or column substrings in the source data table by repeating the e c and r options using the same alias and different source column names or substrings For example the following TRENDlabel command trend_label k Ksi_dbstats_ t Rsi_dbstats_ c dsi_eurodate ta_period 5 3 c dsi_eurodate ta_period 1 4 c dsi_eurodate ta_period 8 4 e J m 2 EU ley 2 creates if needed and populates a key table column named dsi_eurodate with the concatenated substrings from positions 5 7 1 4 and 8 11 of the ta_period column in the source data table In this case if the value of ta_period in the source data table is Jul 29 1998 04 00 00 000AM the value used to populate dsi_eurodate in the key table is 29 Jul 1998 Usage Notes 19 7 19 TRENDlabel Examples Example 1 The TRENDlabel command trend label k Kib ii ifentry t mib ii ifentry e ifdescr010 populates the dsi descr column in the Kib ii ifentry key table with the data contained in the ifdescr010 column from the mib ii ifentry source data table If the time when the trend label command executes is 11 45 AM the dsi key id record with
356. ng System Sybase configuration parameters 9 9 TREND processing options This document discusses each area Note however that not all identified areas apply to each installation Factors such as platform or workload may also influence the overall effectiveness of the configuration area being discussed 9 O Dm 2h 3 O S 1 D el 3 Ce C Performance Tuning Issues Hardware Considerations TREND is a database application that is designed to run on one to many Sybase databases The hardware factors that affect overall performance are Processor type and clock speed Number of processors Amount of RAM Disk drive interfaces 9 9 Number of physical disk devices Processor Type and Clock Speed C 2 Depending on the selected platform a variety of CPU chips are available for the various system configuration options For example a Windows NT system can run on a variety of processors including the following Intel 386 486 Pentium Pentium Pro Pentium II AMD K5 and K6 Cyrix Note TREND for Windows NT only runs on Intel compatible processors not on Alpha MIPS or Power PC Similar considerations apply to Sun Solaris HP UX and IBM AIX platforms Factors such as the number of managed elements collection frequency and data retention period affect the selection of an appropriate processor for the desired TREND implementation Hardware Considerations TREND 3
357. ng null values during logical operations Expression Result NULL NULL TRUE NULL not null value FALSE NULL NULL FALSE A 65 TRENDstep Expression Result NULL not null value TRUE NULL lt gt NULL FALSE NULL lt gt not null value FALSE Not null value lt gt NULL FALSE NULL lt gt NULL TRUE NULL lt gt not null value FALSE Not null value lt Z 22 NULL FALSE NULL AND OR anything FALSE Final Block The statements in a final block are executed only once on the result columns Only output data columns and output key columns can appear inside a final block A 66 TRENDstep Figure A 4 shows an example of a Final Block FINAL if RKEY speed NULL ROTATE ipoctets NULL RKEY speed 0 iputilization 0 else iputilization ROTATE ipoctets 8 100 RKEY speed endif if ROTATE iputilization lt 5 grade BAD else grade GOOD endif ENDFINAL Figure A 4 Example of a Final Block Switch Blocks Figure A 5 shows the construction of a switch block switch expressions case lt logical operator gt expression column_name expression column_name expression case lt logical operator gt expression column_name expression default column_name expression endswitch Figure A 5 Example of a Switch Block A 67 gt D v o Q o
358. ngs and their membership in polling View and Type lists is stored in the database called the To pology Database in the TREND Define Users and Groups window The polling in structions that mw_collect uses include the name of the user who created the request That user is a member of a polling group and from that group s definition mw collect determines where the Topology Database it needs to execute the current polling instructions is located mw collect gets the list of devices to be polled and their community strings from the Topology Database See TREND Configuration for trendadm on page 2 1 for details The polling instructions include the name of the system that is intended to do the polling entered in the hostname field of the Collect Data Periodically window If mw collect is run with the n option it checks this field to see if the hostname matches the name of the system on which it is running If they do not match mw collect does not execute the poll Without the n option mw collect ignores the hostname field and runs the poll regardless of any system name specified there The group to which the user who defined a polling request belongs also specifies the name of the database to which the collected data is to be written the Data database in the TREND Define Users and Groups window mw collect loads the collected data into that database once the polls are completed Direct Polling You can use mw collect to imme
359. ngs up a secondary dialog entitled Network Transport Editor 7 Select tli tcp as the Transport Type enter the secondary server s host name enter the port number that is configured for Sybase connections on the second ary server and press OK This returns you to the Server Entry Editor dialog 8 Press OK This returns you to the Directory Service Session dialog 9 Press Close Session This returns you to the Select a Directory Service dialog 10 Press Exit 11 Press Yes when prompted about exiting dsedit This returns you to the syb setup dialog 12 Press Exit the icon that has an arrow through the door You can validate the result by viewing the SSYBASE interfaces file D 10 Creating the Sybase Interfaces File on UNIX Systems TREND 3 6 User s Guide For Sybase 11 5 1 Use the dsedit utility to install the Sybase interfaces file for Sybase 11 5 1 on UNIX systems Perform these steps 1 Bring up a terminal window 2 Bring up the dsedit utility by typing the following command SSYBASE bin dsedit The Select a Directory Service dialog appears o D lt iy g m E E e 3 3 2 r apjjajes e dn Bunes q Creating the Sybase Interfaces File on UNIX Systems D 11 D Setting Up a Satellite Server Environment 3 Highlight the Sybase Interfaces File on the Select a Directory Service dialog verify the location of the interfaces file and click OK The Directory Service Session dialog
360. nistrator trendadm See TREND Configuration for trendadm on page 2 1 for a complete description Display and Menu System 3 5 3 The TREND Main Window Define Users and Groups Define Users and Groups allows trendadm to set up users and groups see Defin ing Users and Groups on page 2 2 Define Nodes Define Nodes adds nodes to be polled to the database see Defining Nodes on page 2 5 Define Views Define Views creates a view in TREND A view is a set of nodes that are grouped together so they can be polled by a single collection request see Defining Views on page 2 7 A view is used to group together nodes on a geographic or organiza tional basis Define Types Define Types associates a type with a group of nodes A Node Type is the kind of device e g router RMON etc and allows you to poll nodes collectively that share a common type see Defining Types on page 2 9 3 6 Display and Menu System TREND 3 6 User s Guide Specify Report Directory Specify Report Directory changes the default path for the report definition directo ry see Changing the Default Report Directory on page 2 11 The default path is DPIPE_HOME reports Edit Descriptions Edit Descriptions adds or modifies descriptions for interfaces associated with spe cific device tables see Defining Keys on page 2 12 Set Application Color Set Application Color sets the background color for the
361. node for those fields Once all target devices have been polled the returned data is written to the database and mw collect exits Distributed Polling In the simplest case all polling instructions node information and data tables are on a single system and all polling is done from that system However it is possible to maintain polling instructions on one system store target device names their com munity strings and polling View and Type lists on a second system execute polling from a third system store the returned data on a fourth system and run reports against the returned data from a fifth system This is the most extensive example of distributing the application but any combination from the simplest to the most ex tensive is possible and can be configured according to your requirements You can install mw collect on any system in your network to distribute the polling load to the most efficient locations The trendtimer program invokes mw collect and uses trendadm s DSQUERY environment variable value to determine the database mw_collect to be queried for polling instructions Since any system can access a central database from anywhere on the network to get its list of polling instructions you can define the polling policy for an entire network from one central location Alternatively you can define polling for each region locally within that region Information about nodes to be polled including their community stri
362. nt to copy information from data tables contained in PRIMARY_SYBASE to BACKUP_SYBASE The information you want is from May 29 through August 15 Use the following command trendcopy s PRIMARY SYBASE b 3m 1 2m 18d S BACKUP SYBASE Note When defining baseline and length you cannot specify a zero length In addition you can only define one baseline length combination on any one TRENDcopy command 18 14 Usage Notes TREND 19 TRENDlabel TRENDlabel populates one or more columns in a key table with data from its counterpart data table Note TRENDlabel cannot be used with summarized data tables That is columns from tables generated by TRENDsum will not populate columns in a key table Each data table in the TREND database is associated with a properties key table By default most of the columns in a data table are not present in its associated key table However you may find it useful to include some of these data columns in the key table TRENDlabel allows you to do this e J m 2 o ley 2 TRENDlabel is executed from the command line and can be scheduled to run regularly in trendtimer sched 19 1 19 TRENDlabel Synopsis Options 19 2 trend label c alias column C d debug level e column h k key table o hour r aliasz column t source_data_table V Identifies the column in the source data table that is used
363. nterval in minutes rmon collect will execute entries in the mw collection table which have this value in their interval field Note that how frequently rmon collect is actually run is determined by the configuration of TRENDtimer but the idea is to be consistent so that a collection request with an interval of 5 is run every 5 minutes This option when set will display the key values of the device you are polling It will only display the interface key not actually do a collec tion This is used in the Collect Data Collect Data Periodically when polling individual node by key for the first time for a device and when the Select Key option is accessed this option is used to get a list of keys Enable distributed polling If this option is used rmon collect will only execute the collection request if the hostname field in the mw collection table record for this collection request matches the host name ofthe machine on which rmon collect is running If you omit this option rmon collect will execute all polling requests whose interval matches the value ofthe ioption regardless of the hostname specified to do the polling in the polling instructions A 35 gt D v o Q o o rmon_collect A 36 0 P r The number of seconds rmon_collect will wait for a response after sending a poll The default is to wait 5 second after the first poll 1 sec ond after the second and third attempts and 2 seconds after the four
364. nuals in Portable Document For mat PDF along with the Adobe Acrobat Reader These documents and the Adobe Acrobat Reader are available prior to installation in the Documentation di rectory of the installation CD You need to install the Adobe Acrobat Reader to view these documents online These documents are also installed onto your hard drive in the docs directory in DPIPE_HOME during installation XXXi Preface You can download the latest TREND manual updates from the DeskTalk Systems Customer FTP site Look at the document number on the manual s title page to de termine if you have the latest version of manual IG TREND Installation Guide UG TREND User s Guide RPR TREND ReportPack Guide TW IAG TRENDweb Installation and Administration Guide BTR Guide To Building TREND Reports VTR Guide To Viewing TREND Reports Product version e g 31 means version 3 1 Publication date in yyyymmmdd format TW IAG 31 1999FEB01 Call DeskTalk Systems Customer Support for download instructions xxxii TREND Introduction to TREND Desktalk System s TREND is a network management software product that collects network information from SNMP agents stores it in a commercial relational data base or multiple databases manages data and provides historical reporting based on that information TREND is comprised of a set of generic application modules for data collection ag gregation management and reporting as w
365. o be run sequentially at a predetermined time In addition TRENDproc enables the execution of multiple instances of an application allowing them to run in parallel How TRENDproc Works The concept behind TRENDproc is relatively simple 1 Take a sequence of processes which in the past had to be accomplished indi vidually and string them together in such a way that they can be executed with a single command In TRENDproc terminology this series of sequential commands makes up a block the equivalent of a macro 2 Next make it possible to string blocks together like individual commands 21 1 21 TRENDproc When one block finishes its assigned tasks automatically launch the next block and so on until all of the blocks of processes have been executed 3 Then to make the process really powerful make it possible to launch multiple blocks concurrently so that they run in parallel 4 Finally make it possible to launch the entire process with a single entry in trendtimer sched The result is a system that allows TREND users to automate many of the tasks that have traditionally been accomplished manually This system is made up of two parts The TRENDproc Application which reads the trend proc file and runs the commands it contains trend proc files which contain the commands that are to be run The TRENDproc Application TRENDproc is a program that is initiated by Ipr launch When launched TRENDpr
366. oc reads the appropriate trend proc file and executes the scheduled com mands in the appropriate order The TRENDproc is not intuitive it acts as a robot following the instructions provided by the trend proc file Since a trend proc file can itself call another trend proc file multiple instances of TRENDproc can run concurrently trend proc Files A trend proc file contains one or more blocks of commands A block is a sequential list of commands to execute Assuming there are six items in a block TRENDproc will execute the first item and wait for that command to be completed As soon as 21 2 How TRENDproc Works TREND 3 6 User s Guide the first item is complete TRENDproc initiates the second item When the second item finishes TRENDproc starts the third and so on until all of the items in the block have been executed N pare E D m Z a o 9 O Most trend proc files contain multiple blocks arranged in a linear progression It is necessary for the blocks to be in a particular order because TRENDproc reads the blocks sequentially However TRENDproc is able to execute commands from different blocks concur rently This is so because TRENDproc recognizes two different kinds of blocks in a trend proc file wait and nowait A wait block is so called because it causes TRENDproc to wait until the block has been executed in full before moving on to another block A nowait block permits TRENDproc to go to the nex
367. ocate a buffer pool to the newly created cache logcache isql Usa P sp poolconfig logcache 4M 8K go 6 Restart Sybase to enable the new log cache Transaction Log Size Determine the TREND transaction log I O size by entering the following commands isql Usa P use dpipe db go Sp logiosize go The output of this command looks like this C 14 Sybase Configuration Parameters TREND 3 6 User s Guide The transaction log for database dpipe_db will use I O size of 2 Kbytes return status 0 9 O 2h e 3 O D el 3 Ce By default Sybase is configured to use a 4K transaction log But Sybase does not allocate any 4K buffers in the default cache configuration for the transaction log to use Thus a 2K transaction log is used for systems that have not changed the cache configuration Sybase recommends a 4K logiosize for user databases particularly OLTP activity Some TREND installations fare better with an 8K log I O size To change the log I O size allocate a cache pool for only the transaction log 8K for example by entering the following command isql Usa P use dpipe db go sp logiosize 8 go The output of this command looks like this The transaction log for database dpipe_db will use I O size of 8 Kbytes return status 0 Number of Locks The number of locks has little to do with performance But if the total number of locks specified
368. od BS pest BS Rod Deo BSS Kod BS pe Doo pa Pay bee Med Pil bed Med BSH bes Med poy Pal Red pes Doo Poe Kod Deo bo 3 000 00 bes Poet kasi bes ps kd Best Poot EE Posh psa lesd 1829 padi 94 fst pt Kot bes Dd Kot POL ERR BOS bash PS Rod pas Doe Ee pea FST ped eo Feed Best post E pogi o lel bed bed eod Ps odd Ee BEST pag kd Pod psi od Bes bag Bot fend Bel oc bed KSI NY KOO ROSA IRA es BOS RS Ksi KOY Pes Ig BSA POSH NA KGG eed S29 KA BA NGA REA KI bes 903 bed poo ES Beo ed Pod basa oed BST Pel PST Re Kd be d ko Pe ed bod Perl oed bed Re xd Be PS eo oed Ro Po 1 500 00 PS Fest Kod pes BS ed Pool BSI Doo boc Nag bod Kod Des bod Ki Pe Mod Des PS Kod KI Deo Poe Kod BOY bo 500 00 KS p8 kasi BI poot oa Book ROY MON Bosh beet EE Bed ood 1394 Bed o E esl ES I best E E Ros bagi Es 1394 bed Be Ka kal Kal Bal Ba ka ka Fa IB KI ka ka ki ka sd Ka Ra det el Ka Ka Ka Sy Ka Best BS Rod BSS 0 00 i EH HH 11 12 13 14 15 16 17 18 19 20 21 10 45 Sat Nov 15 1997 Fifteen Minute Scale WE ever _octets 134 70 18 241 1 EEE other_ip_octets 134 70 18 241 1 E total_ip_octets 134 70 18 241 1 BEZZA snmp_octets 134 70 18 241 1 Figure A 3 Example Stacked Bar Graph Based on Example Output Table A 56 TRENDstep Figure A 3 shows a part of the bar chart generated from the example output table The actual report shows the specified statistics over a 24 hour period
369. of available space is left track how quickly that space is getting used up Next watch the size in Kbytes of your data tables Data Manager s sort feature can make this easier If you are using up space quickly see which tables are largest and Oo O D o 3 pi Q C Showing Particular Tables 8 9 8 Data Manager 8 10 how fast they are growing Control their growth by shortening their aging periods Decide what kind of data you really need to keep for an extended period of time ver sus those you can discard after a day or two If you are not reporting on raw data for example you probably don t need to keep it longer than the day it needs to be main tained for aggregation Since this is the most voluminous data it makes sense to man age it more carefully In extreme cases you may need to truncate a large table to keep the database from filling up The longer the data table the longer it will take for the report to run Note Your database is becoming full at 80 and is completely full at 96 Adjustments are recommended if db used exceeds 70 Do your housekeeping periodically and delete tables you no longer need You can obtain a report that predicts database size by using AutoPilot to install the Database Size ReportPack See the TREND 3 6 ReportPack Guide Showing Particular Tables TREND Data Aggregation and Manipulation TREND polling utilities called DataPipes collect performance data fro
370. of minutes default 60 gt NT version only e max number of processes C number of minutes default 30 gt NT version only d debug level h hostname k n t tablename o timeout in seconds p max entries per pdu P snmp port number default 1615 r number of retries s round factor sec default 300 u V w high water mark default 90 gt rmon collect is TRENDS direct SNMP polling application for specific RMON MIB tables rmon collect functions in a manner very similar to mw collect but has been customized to understand the specialized structures of the various RMON data tables A 81 gt D v o Q o o rmon_collect A 32 Its simplest use is to collect a specific table once from a single probe and store the information in a relational database Fully utilized it will follow user defined in structions to poll a dynamic list of probes anywhere on the network for data at reg ular intervals and store the results on a local or remote system depending on who requested the information Polling Controls rmon_collect uses the same mechanisms for one time and interval polling as mw_collect These methods are described in mw_collect 1L However rmon_collect should only be used to poll nodes whose type is set to rmon in the TREND Define Nodes window The main difference between mw_collect and rmon_collect is that rmon_collect will onl
371. of the time you wish to set i e Hour Day or Week Next click either the or buttons to increase or decrease this value Click the Apply button to have the sort order you specified take effect Click the Reset button to return to the default sort order Click the Cancel button to quit out of the window without changing the current sort order Establishing Constraints TRENDsheet provides exception reporting capability By selecting a statistic a comparison operator and a comparison value tabular reports can be created that will show only the exception conditions specified 11 12 Specifying Data to be Included in the Report TREND 3 6 User s Guide First select a statistic using the pull down box at the left hand side of the Create Constraint box The second pull down box contains the following operators equal to lt less than or equal to gt greater than or equal to lt less than gt greater than not equal to lt not less than gt not greater than and like compares the first value to a following character string Next enter any value or character string in the box at the right This entry will be compared against the value of the statistic specified using the operator selected Click the Select Distinct Values box to report on only the first instance where a con straint is met Finally click Add to place the expression in the Current Constraints box and click OK to exit Cons
372. olumns for snmp packets snmp trap packets www packets other ip packets and total ip pkts define TIMEADJUST source_table Rmon2_statsdata_ destination_table rmon2_protocols switch KEY ap_layer case snmp snmpcount oldiststatspkts009 otheripcount 0 0 oldiststatspkts009 case snmptrap snmptrap oldiststatspkts009 otheripcount 0 0 oldiststatspkts009 case http wwwcount oldiststatspkts009 otheripcount 0 0 oldiststatspkts009 case www http wwwcount oldiststatspkts009 otheripcount 0 0 oldiststatspkts009 endswitch if KEY nt_layer ip amp amp KEY tr layer otheripcount oldiststatspkts009 endif FINAL totalip ROTATE otheripcount ROTATE wwwcount ROTATE snmpcount ROTATE snmptrap ENDFINAL by variable target name001 by variable KEY dsi data source TRENDsum TRENDsum Calculates various statistics such as averages percentiles baseline values and fore cast values trend_sum a baseline table name b by variablel by variablen c column specl column specn e destination table f file g h m n null percentage r t source table V w first day of week y baseline days A 75 gt D v o Q o o TRENDsum Options TRENDsum has the following command line options A b Optional name of the destination table for TRENDsum bas
373. ombines them into a view without actually creating a new table in your database In the course of using TRENDdbview to define your view you assign it to a key table Each data table has an associated key table When you use TRENDdbview to select data tables their key tables also appear You choose which of these key tables you wish to associate with your view The view becomes a virtual table cross referenced by the key table To TRENDS report generating applications the view appears to be another data ta ble As a result you can view the data and generate and print reports just like you would for any data table Note Once you create a view in your database it acts like a data table This means that you can also use columns from a previously created view as part of a new view just as if the existing view were a data table 22 2 How TRENDdbview Works TREND 3 6 User s Guide A Note on Limitations TRENDdbview does not limit the number of data tables and views that contribute columns to a view However TRENDdbview can only support up to two key tables You can include columns from every member data table in a key table and you can do so for one or two key tables However you cannot combine columns from data tables that are members of three or more key tables Elements of TRENDdbview TRENDdbview is made up of five screen elements These are N N D m z g ex o TRENDdbview main window S
374. ommand see Print on page 1 17 The Change Database command provides access to a different SQL server to dis play its data see Displaying Data from a Different Database on page 12 12 Use the Exit command to quit out of TRENDgraph see Selecting Statistics to be Graphed on page 12 6 The About command invokes TREND version information see About TREND on page 1 19 View Select Data Time Period K Axis Y Axis Grid Graph Title Refresh Use Select Data to change the data source node for a report see Selecting Data on page 12 8 Set the time range of data to be reported on with the Time Period option see Set ting the Time Period to be Displayed on page 12 13 E N D m Z o e 5 o Main Display and Menu System 12 3 12 TRENDgraph Change the time scale with the X Axis command see Setting X Axis Parameters on page 12 16 Change value ranges of plotted statistics with the Y Axis command see Setting Y Axis Parameters on page 12 17 The Grid command provides a toggle for adding or removing grid lines from a re port see Adding Grid Lines on page 12 20 Graph Title allows the addition of meaningful titles to graphical reports see Spec ifying Report Titles on page 12 19 Use the Refresh command to update the current graph with current data Edit Aliases Expressions Mod
375. ompletely transparent In addition to the database containing the polling policy definition each polling action based on a view and or a type may involve up to two other databases Use of other databases is based on the name of the user who defined the polling action The user s name can be seen in the User column of the Collect Data window Each user is a member of a group Each group has an associated to pology database and a data database Node view and type definitions are stored in the topology database Different logical topology databases associated with different user groups may share a common physical database Once collected data is stored in the data database The SQL Server which stores collected data is set in the data db field of the tl groups table in the database The group of which a user is a member is set in the group name field in the tl users table The user whose group name and associated data db value is used to determine where to store the data is the user who defined the data request If a user is not a member of a group the data db entry for the group default is used Setting Up New Polling Requests TREND 3 6 User s Guide Saving the Collection Request Definition Once the collection request is defined the way you want it save the definition by clicking OK The definition will appear in the Collect Data window Click Cancel to discard the current definition and exit the window Changing Databases
376. on page 14 5 Specify the report file directories to be used see Selecting the Report File Directory on page 14 7 14 4 Invoking the Report Launcher TREND 3 6 User s Guide Select a report type and report see Selecting a Report on page 14 11 Select the devices and elements to include see Selecting Devices and Elements on page 14 14 Specify the report time period see Specifying the Report Time Period on page 14 18 Launch the report see Generating the Report on page 14 20 Refresh the report list see Refreshing the Report List on page 14 20 Selecting the Database Kar B D Dm fe r o i 3 7 IT D The default database from which report information is taken is identified by the val ue of the DSQUERY environment variable Take the following actions to select a different database Selecting the Database 14 5 14 Report Launcher 14 6 1 2 Select Change Database from the File menu Report Launcher f x File itrendsnmp reports Top Level Report Directory Up one level Cisco Interfaces RMON ethernet Daily RMON Ethernet Statistics Figure 14 3 Report Launcher Window s File Menu The Change Database window appears Change Database LITTLE ha LITTLE LITTLE_BS LITTLE_MS RED_SYBASE Figure 14 4 Change Database Window Select the desired database from the entries in the list bo
377. ons h H l 0 P Specifies how many consecutive attempts to ping a node must fail before the node is automatically deleted from the dsi_nodes and dl_type tables Specifies the number of packets that can be in a discovery ping The default is 1 Specifies the single community string to use for all SNMP GETS Specifies the debug level Suppresses domain names Causes the progress of Discover to be displayed on the screen Specifies that Discover will read a community strings file and use the character strings in that file for its SNMP GETs Name of the type definition file Can only be used in conjunction with the t option Defines the maximum value of the discovery range Causes a list of available Discover command line options to be dis played on your screen Defines the minimum value of the discovery range Determines how long Discover will wait before timing out an SNMP GET request Specifies the time in milliseconds that must pass before Discover times out a ping when Discover is running from a Windows NT plat form The default is 750 milliseconds Resets the maximum number of entries that can be in a Protocol Data Unit pdu The default is 20 Defines the port number to be used The default is 161 Running Discover from the Command Line 5 13 a o i o o lt 2 5 Discover r Defines the number of times Discover will retry its SNMP GET if there is no response S Defines the
378. oor Utilization lt 5 6 15 16 30 gt 30 Collisions 596 690 1090 1196 1590 gt 15 Errors lt 1 296 596 6926 1096 gt 10 Overhead lt 5 6 10 11 15 gt 15 The Grade of Service Concept 13 5 13 Grade of Service If at a given time you receive a Utilization value of 6 Good 2 Collisions of 4 Excellent 1 Errors of 3 Good 2 and Overhead of 13 Fair 3 your GOS for that time period is calculated as 0 40 2 0 25 1 0 25 2 0 10 3 0 80 0 25 0 50 0 30 1 85 In this case GOS is between l and 2 which falls in the Good range Opening an Existing GOS Report Existing reports are stored as files on disk They are opened by loading them into the Grade of Service application Click the Open option under the File menu This will bring up the File Selection window see Figure 1 2 The Select File Window on page 1 15 13 6 Opening an Existing GOS Report TREND 3 6 User s Guide Creating a New Grade of Service Report o o o Q o e p e 7 The definition of a new report including selection of the mib is done in TREND build See TRENDsheet on page 11 1 You can invoke TRENDbuild from inside GOS by selecting the New option under the File menu untitled Figure 13 2 TRENDbuild Main Window Selecting Statistics The first step in creating a new graph is to select the category MIB and
379. option to set up the property tables for devices that you want to poll by key Then access the Collect Data Periodically window and set up a polling policy to poll by key for the desired devices See Collect Data on page 7 1 for an explanation of using the Collect Data Periodically window to set up a polling policy Specifies whether to reject samples if there are spikes A spike occurs when the value of any counter suddenly goes too high TRENDpm detects a spike when the value of any counter in the first sample is greater then the value of that counter in the second sample or the differ ence exceeds the spike threshold The value of the spike threshold is 2 Valid values are Rejects samples if a spike occurs 0 Does not reject samples The default is O This option is in UPPERCASE This option is equivalent to the bSuppress_spike parameter in TRENDpm mw_collect L n N Specifies that the collected data be stored locally instead of in the TREND database This option is in UPPERCASE This option is only available in the cache version of the poller Sets the minimum filter value The procedure rejects the sample if the delta value of a counter falls below this value The default value is 1 which means to accept the entire sample This option is in UPPERCASE This option is equivalent to the line suppress value parameter in TRENDpm Enables distributed polling If this option is used mw collect executes
380. or Device Name 14 15 Match Value Window for Element Description 14 17 Report Launcher Window Time Period Considerations 14 18 Set Time Period Window 00022220055 14 19 Report Scheduler Window 2 nanana 15 2 Report Parameters Window 00 e eee eee 15 6 Sort Order Window 2 ce eee ees 15 7 RMON Protocols Window 22220000ee eee 17 7 Sort By Window 0 ne 17 8 Search Window 0 eee eee ees 17 8 Get Probe Protocol Data Window 17 10 Get Probe Protocol Data Window with node identified 17 11 Get Probe Protocols Data Window After a Get 17 12 Invalid Protocols Display a 17 13 Change Invalid Protocol Name Window 17 14 Extended Dotted Name Notation 4 17 16 TRENDdbview Main Window llle sels 22 4 Select Data Tables Window 0000 cee eeu 22 5 Select Columns Window 000020000 eae 22 6 Define Expressions Window llle 22 7 Define Qualifications Window sells 22 8 TREND Preface This manual is intended for anyone using TREND network management software It describes the philosophy of operation and hands on use of the product It also helps you to interpret the data provided and to maintain TREND Document Overview This user s guide is designed to provide both tutorial instructions and reference in formation
381. or display in the Select Element list box are restricted to those that match the specified value Element Description The Match Value window for element description appears if you select this filter Element Description I Partial Word PO RS Clear OK Cancel Figure 14 11 Match Value Window for Element Description The filtering criteria is applied to the Element Descrip tion column in the database the elements that are retrieved for display are restricted to those that match the specified value 8 Even after filtering criteria is specified the list of entries in the Select Element list box can be lengthy In this case you can search the list for entries that match a specified string by selecting the Search Lists option from the View menu see Figure 14 5 The Search Lists window appears if you select this option see Figure 14 8 Selecting Devices and Elements 14 17 EE A D Ld m o c 3 7 IT Dm 14 Report Launcher 9 Inthe Search Lists window select the Elements radio button and specify a search string in the Search text box If the string is a partial word select the Partial Word check box 10 Click Search to locate the entries in the Select Element list box that match the specified search value Click Clear to clear the text in the Search text box Specifying the Report Time Period Start and end times are specified for some reports in the
382. or the current node in this window does not update their values for the node in the database Periodic polling of the node by mw collect usesthe community strings in the database not the ones specified here Utilities TREND 3 6 User s Guide Polling Data to Screen The SNMP Tool allows you to Get and display data for MIB values on the current node To use the SNMP tool 1 Select a MIB object by clicking the right mouse button on its name in the MIB map This highlights the object In the following example mib II mib is loaded with the system variable ifEntry selected Node powder MIBwalker C TREND mibs mib II mib ifPhysAddress ifAdminStatus ifOperStatus ifLastChange iflnOctets iflnUcastPkts iflnNUcastPkts j iflnDiscards F ifinErrors iflnUnknownProtos j ifOutOctets ifOutUcastPkts ifOutNUcastPkts ifOutDiscards ifOutErrors ifOutQLen ifSpecific Figure 6 26 MIBwalker with Variables of the ifEntry Table Selected 2 Select the SNMP Tool option from the Tools menu to bring up the SNMP Tool window SNMP Tool iof x Set Variable Set table Abort FI Group by Instance Figure 6 27 SNMP Tool Window Utilities 6 31 o gt Uu 2 x mg 6 MIBwalker 6 32 Click the Get button in the SNMP Tool window to initiate a GetNextRequest tree walk of the MIB tree for the group or object selected Figure 6 28 below shows a p
383. or trendadm tions from the Configuration menu in the main TREND window The main TREND window is displayed when you start TREND Perform these steps to invoke TREND 1 Log in as the trendadm user 2 From the Start menu select Programs gt TREND The main TREND window is displayed The Configuration menu has the following options ieee Tools Define Users and Groups Define Nodes Define Views Define Types Specify Report Directory Edit Descriptions Set Application Color Figure 2 1 The TREND Configuration Menu Nodes are organized into functional Types or Views This organization is the basis for Collect Data s data collection setup process where the nodes node types and node views defined in TREND determine which devices on your network are polled as a group for network performance statistics If you are not the trendadm user the options in this menu are grayed out Defining Users and Groups Users are TREND users you are registering with the application Groups are collections of users who will share a common network node organization setup see Defining Nodes on page 2 5 and Defining Views on page 2 7 2 2 Defining Users and Groups TREND 3 6 User s Guide To setup users and user groups select the Define Users and Groups option under the Configuration menu This will bring up the Define Users And Groups window Define Users And Groups PATUXENT_SYBASE PATU
384. ort libraries trendcopy TRENDcopy provides incremental periodic filtered database replication to a backup server without the assis tance of an experienced database administrator Replica tion can be ad hoc using a command line interface or scheduled on a periodic basis Database tables can be cre ated and or incrementally copied to another database based on user definable criteria such as thresholds time spans or sets of nodes so that the database contains only the information required for reporting applications TRENDdbview A GUI application though launched from the command line that enables TREND users to create database views at will trendexport A command line application that allows TREND users to export TREND data as ASCII text This makes it possible to use the data generated by TRENDit and TRENDsum in disparate applications such as spreadsheets 3 of 5 TREND Overview pos TREND 3 6 User s Guide El o a Module Description Q TRENDgraph TRENDegraph reports provide time based line graphs of a information selected from the data repository based on queries defined through TRENDbuild The user can dis play multiple values from a single device or multiple devices or interfaces all synchronized by time for easy comparison The user can select line and data point color thickness and pattern X axis time interval Y axis data range and labels and graph headings Cumulativ
385. ote that they may be discontinued in future releases Examples A 84 Example1 Execute the following command to invoke TRENDsun to read input from a rate ta ble named Rnterface_base_ and create a rolling baseline output table named SD42Rnterface_base_if based on a 42 day rolling baseline period DPIPE_HOME bin trend_sum t Rnterface base e if y42 f SDPIPE HOME scripts if baseline hourofday volume interface sum The following file contains the column and by variable specifications to be included in the SD42Rnterface base if output table SDPIPE HOME scripts if baseline hourofday volume interface sum The contents of this script file are trend sum proc to update hour of day baseline table for combined network volume column thru put z ifinoctetstifoutoctets per95 std wav TRENDsum column volume ifinoctets ifoutoctets tot avg by variable keyid by variable hour_of_day The SD42Rnterface_base_if output table contains 24 rows one row for each hour of the day for each keyid Example 2 Execute the following command to invoke TRENDsum to read a daily summary ta ble named SDifexceptions and create a rolling baseline output table named SDifex ceptionsfore with a 42 day rolling baseline period SDPIPE HOME bin trend sum t SDifexceptions e fore y42 f SDPIPE HOME scripts if forecast sum The file SDPIPE HOME scripts if forecast sum contains the column and by variable specifications to be included in th
386. ou have the TREND Autopilot module open you will need to close it before you can launch Discover The Start Stop and Cancel Buttons There are three command buttons in the Discovery Interface Start Stop and Cancel The function of these buttons is described below Start The Start button initiates the discovery process The Start button is only available when Discover is stopped as shown by the legend Discover Stopped at the bottom of the window Using the Discover Graphical User Interface 5 7 5 Discover Stop The Stop button halts the discovery process The Stop button is only available when Discover is running as shown by the legend Discover Running at the bottom of the window Cancel The Cancel button closes the Discover window It does not affect the running pro cess in any way Note If you click the Cancel button while Discover is running the window closes but Discover CONTINUES TO RUN If you want to stop the Dis cover process you must use the Stop button Defining an IP Discovery Discover also allows you to define the parameters of your discovery The parameters that can be defined are Subnet Mask IP Address Range Read Community Strings 9 9 Host name Translations on off These parameters are briefly described in the following sections 5 8 Using the Discover Graphical User Interface TREND 3 6 User s Guide Subnet Mask oa g 7 1 lt D When the
387. p lt number gt is the number of values returned If you do not want to wait for polling to finish you can abort the Get by clicking on the Abort button Polling stops and any values already received are displayed If a table has multiple instances SNMP Tool displays the data in the table grouped by variable That is all the values for the first field are displayed then all the values of the next field and so on If you want to display the data such that all the variables for one instance are grouped together followed by all the variables for the next instance click the Group by Instance button The fields are displayed in the same order in which they are defined in the MIB table Something not always obvious when collecting data from nodes on the network is whether or not each device actually returns a value for every field in the requested table With the SNMP Tool it is easy to see whether this is the case Poll the current node for the table Do not group by instance Scroll through the returned values and see if any fields are missing SNMP Sets Utilities You can set the values of writable MIB variables through the SNMP Tool window either picking out a single variable or updating any writable variables in a table simultaneously The current node is the target of the set 6 33 o ius 2 F D 6 MIBwalker Setting A Single Variable 1 Select the variable name in the MIB map that you want to set or the t
388. page A 1 describes the syntax and detailed usage of each utility Raw Data Collection You can specify the interval at which the various TREND DataPipes collect the performance data that ultimately appears in the TREND ReportPacks For purposes of this discussion assume that a 15 minute collection interval is defined for each DataPipe although in practice the collection interval typically varies by DataPipe Raw Data Collection 9 3 9 Data Aggregation and Manipulation Assuming a 15 minute collection interval TREND DataPipes attempt to collect four samples per hour or 96 samples per day 24 hours 4 samples per hour TREND DataPipes store the collected data in tables called raw tables in the TREND database In practice data for a given sampling period is not all collected at the same instant in time data collection may take several seconds or minutes to complete depending on the number of devices being polled and the volume of data that is collected Each row of collected data has real timestamps that indicate when the request for the data is made and when the data is received In addition TREND DataPipes associate a virtual timestamp which appears in a column named ta_period with each row in the raw table The virtual timestamp standardizes the real timestamps by applying the same begin time value to each row of data in the sample For example all data collected during the 12 15 a m collection period on June 22 1998 is as
389. pdate button The new value will replace the original in the Select Alias box Assign an alias string to a statistic by clicking on a statistic in the Active Statistics box to highlight it then click on an alias string in the Select Alias box to highlight it Now click on the Assign button The names are now associated A statistic with an alias assigned to it is displayed in the Active Statistics box in the format statistic name gt alias name Creating a New Report 10 11 10 TRENDbuild You can disassociate an alias from a statistic by clicking on the pair in the Active Statistics box then clicking on the Remove button The gt alias part of the value will disappear from the statistic in the box To apply an updated alias string to a statistic that already has an alias string remove the alias from the statistic then assign the new value This will not be done automatically When you finish assigning aliases quit the window by clicking on the Cancel button The assigned aliases will appear with their associated statistics in the Statistics box in the main TRENDbuild window Note The expression length allowed by TRENDbuild is 1024K An expression is a calculation that yields a derived value from existing data Expres sions created in TRENDbuild are used in the select clause of an SQL statement The results are plotted or displayed like the values of a statistic Expressions are built in the Edit Expressions window
390. pears 3 Double click the view you want to open or highlight it and click OK The Open File Dialog Box disappears the information contained in the view Working with TRENDdbview 22 19 22 TRENDabview becomes resident in TRENDdbview and the view name of the view appears in the TRENDdbview Title Bar The view file is now open and you can make any changes you want to it However since the virtual table is created by executing the saved file your changes will not be reflected until you save and run them Save Changes to an Existing View 1 In the File pull down menu select Save The file is updated to reflect the changes you have made Or In the File pull down menu select Save As The Save As Dialog Box appears 2 Inthe Save As Dialog Box define in which directory and under what file name you wish to save the updated view Click OK If you wish you can use the updated view to replace an existing view by highlighting the existing one in this dialog box then clicking OK Once you have saved the updated view you must run it again to update the virtual table it generates Auditing Data and Generating Reports 22 20 Once you have created saved and run your view it exists in your database and acts as a data table You can see the data contained in the view by using TRENDsheet and TRENDgraph However before you can use either of these tools to audit the da ta you must first build an appropriate report This is
391. pecified in the Rollup column Any MIBlets you create on your own with MIB walker and for which Yes is specified in the Rollup column for the item in the database The value given in the Rollup column is No for all tables used to generate the TREND 3 5 Report Packs For a detailed discussion of MIBwalker see MIBwalker on page 6 1 TRENDsum rather than TRENDit creates the summary tables and calculates statistics for the ReportPacks in TREND 3 5 and later releases See Other Data Aggregation Functions on page 9 15 Data Cleaning TREND 3 6 User s Guide The Base Table o 5 2 D O E D 5 uonebaibby eleg 6 TRENDstep processes the cleaned delta data from TRENDit or other sources to create a template table called the base table The base table contains all the columns needed to generate data for all reports in the ReportPack TRENDstep uses a language with if then else case and expression constructs that enables it to perform the following types of operations on the input data Include source data columns in the base table without modification Compute new columns from expressions using existing columns Perform row by row conditional processing Create an other category for RMON data 9 9 9 Assign grades to ranges of input data Row by Row Conditional Processing TRENDstep can examine each row of an input table to see if it satisfies a stated condition TRENDstep can
392. pecifies a baseline and the 1 option specifies the length of the time period to copy TRENDcopy establishes the baseline as it relates to the current day at 12 00 00 a m The b parameter defines how many days weeks months or years before today the baseline will be If today is October 1 and you want to establish the baseline as June 1 you must tell TRENDcopy to set the baseline four months in the past b 4m The b parameter defines the end of a time period If today is May Ist and you wish to copy all the data in your selected tables before this year set b at 4 months TRENDcopy copies all the data stored in the specified tables before the baseline time period The l option can only be used in conjunction with the b option The purpose of the l option is to specify a length of time from the baseline forming a finite segment of time to copy If today is February 14 1996 and you want to copy data from the cal endar year 1995 you set the baseline at January 1 1996 b 45d and the length of Usage Notes 18 13 18 TRENDcopy time at one year 1 1y TRENDcopy then copies the data from the specified tables for the year that ended at 12 00 00 a m January 1 1996 Example 1 Assume today is January 11 1996 and you want to copy 1995 data from PRIMARY_SYBASE to BACKUP_SYBASE Use the following command trendcopy s PRIMARY SYBASE b 11d 1 ly S BACKUP_SYBASE Example 2 Assume today is November 15 and you wa
393. per Samples Raw Table Ram Jabig Description of Nodes er Node Row KB Collected Sime per Size per P P perHour Hour MB day MB PVC miblet 50 5 0 4 12 1 2 29 Cisco interface 200 5 0 5 4 2 0 48 stats etherstats 100 1 0 4 6 0 24 6 TOTAL 83 Furthermore assume that Raw Tables are to be retained for 2 days Rate for 7 days Hourly for 30 days Daily data for 30 days and Weekly data for 60 days B 2 Estimating the Size of Your TREND Database TREND 3 6 User s Guide iigsface RawTable Rate Hourly Daily Weekly Monthly Description Size per Table Table Table Table Table TOTAL day MB Size Size Size Size Size PVC miblet 29 29 10 0 4 0 06 0 01 68 5 Cisco interface stats 48 48 16 0 7 0 1 0 02 112 8 etherstats 6 6 2 0 08 0 01 0 00 14 1 SUBTOTAL 83 83 28 1 18 0 17 0 03 195 4 Retention days 2 7 14 60 180 360 TOTAL 166 581 392 71 31 10 8 1252 The recommended database size in this example is double 1252 megabytes or approximately 2 5 gigabytes Enlarging the Database As your database grows and your sophistication in using it increases the demands you place on it may require more space The process of increasing the size of your database involves a series of steps to first allocate the space on your disk then map that space to the appropriate database Note TREND includes a licensing mechanism to track a purchased TREND database size against the database
394. ports in the current Look in report directory Selecting a Report TREND 3 6 User s Guide Report Type Retrieves Title Matching The Match Value window appears if you select this filter Keywords Match Value All reports in the current Look in report directory whose title matches the keywords you specify in the Match Value window are retrieved Separate multiple keywords by commas Multiple key words result in AND type rather than OR type selec tion criteria gt D 1 pap ag o i 3 7 Iz 3 Ifthe Reports list box has a large number of entries you can optionally search the list more quickly by selecting the Search Lists option from the View menu see Figure 14 5 The Search Lists window appears if you select this option Search Lists C Reports Figure 14 8 Search Lists Window 4 Specify a character string in the Search box 5 Select the Partial Word check box if the string you enter is not a complete word A partial word specification has the effect of sending an SQL WHERE clause with wildcard characters in the search argument For example frame Selecting a Report 14 13 14 Report Launcher Click Search The list of entries in the Reports list box in the Report Launcher window is searched and you are positioned at the first entry that matches the search criteria No database access occurs Continue to click S
395. pres sions you choose to define will be applied to your view However applying expressions is likely to improve your results For this reason we recommend that you apply any expression that will help define the view 2 To define an expression for your view you need to create the expression text in the Expression Text Box name the expression then insert it into the Defined Expressions Box To add an expression click within the Expression Text Box A cursor will appear Any text you add will appear at the site of the cursor 22 16 Working with TRENDabview TREND 3 6 User s Guide You can create the expression text by typing it directly into the Expression Text Box by adding text from the Columns in Selected Table Box and by adding operators from the Operators Box To add expression text from the Columns in Selected Table Box highlight a table in the Selected Data Table Key Table Box Its associated or defined columns appear in the Columns in Selected Table Box Scroll to the column you wish to add and click it The text of the column will appear in the Expression Text Box To add an available operator to an expression scroll to the one you want in the Operator Box and click it The operator appears in the Expression Text box If you add expression text then decide you do not want it you can select and delete it as you would in a typical text editor 3 When the text of your expression is correct click in the Curren
396. pro vided you spaced them out over time to ensure that each TRENDproc finished be fore the next was launched How TRENDproc Works 21 5 21 TRENDproc Rather than this you create a sixth trend_proc file Trees to contain the other five Its blocks would be as follows Block A Aspen wait Block B Birch wait Block C Casuarina wait Block D Dogwood wait Block E Elm wait You then put an entry into trendtimer sched to run Trees once a day Possible Applications of TRENDproc TRENDproc was devised to act as a framework for the new TREND ReportPacks However it can be applied in a wide variety of circumstances Example 1 TREND v3 4 uses the TRENDsum application to maximize the efficiency of data aggregation Since the TRENDsum application requires database tables that are cre ated by trendit it is necessary to ensure that the trendit process has finished before initiating TRENDsum By setting trendit and TRENDsum to run as part of a trend proc block you can en sure that trendit has completed its run before TRENDsum kicks off Such a trend proc block might look like the one below begin blockl wait trendit t mib ii ifentry r trend sum e Rib ii ifentry e SDifentry f trendsum sum end blockl Possible Applications of TRENDproc TREND 3 6 User s Guide Example 2 Those TREND users whose TREND networks include Satellite Servers must export data to their central database every day Typically
397. proc lt proc file This release of TREND includes the following modules Module Description Collect Data An application for defining reviewing and updating an enterprise polling policy for periodic collection of man agement data by TREND polling agents Data Manager Data Manager is an administrative tool that displays the structure and use of the TREND database at a glance Tables can be truncated or deleted and default or table specific aggregation and aging choices can be set through the GUI Data Manager also displays the percentage of database space being used db_delete_data TREND data aging process is run on a periodic basis Data older than the retention time specified are deleted on a day boundary each time the db_delete_data process runs 1 4 1 of 5 TREND Overview L TREND 3 6 User s Guide Module Description Grade of Service The Grade of Service GOS application converts histori cal performance statistics into meaningful graphical dis plays using a weighted stacked bar chart format GOS allows the user to select any set of SNMP or RMON met rics and assign a grading scale to each metric based on relative importance GOS can be used for reporting on routers servers LAN segments WAN links or other net work components where system health is critical Si d O 2 o o 3 o MIBwalker MIBwalker extends the polling capab
398. procedure generates rate data in the TREND database after mw collect collects all the data and stores it in the database for a particular table File Locks mw collect allows only one instance of collection for a particular Collection ID to run at the same time Furthermore only one bcp gateway process runs at the same time for a particular Collection ID table name and data database combination A 23 gt s D v o Q o o mw_collect A 24 When the parent poller mw_collect starts it attempts to place an exclusive lock on the running file in the SCOLLECT HOME feeder collection id directory If the lock fails the poller assumes that the previous instance of the poller is still running and exits The parent poller releases the lock on the running file when every collector child finishes Note that the lock releases before the poller finishes loading all the data into database so that the polling continues even though the data loading portion is not finished Similarly when mw collect starts a bcp gateway process it tries to place a lock on COLLECT_HOME feeder collection_id tablename_dbname running If the lock fails the parent poller assumes that previous job is still running and does not attempt to start the bcp gateway process The data is stored locally Local Storage of Data The cache version of mw_collect provides the ability to perform collection without a database The following table summarizes this fea
399. proto col to dsi protocols Note For more information on the RMON2 naming convention see Com bined Protocol Identifier and Parameter String Values on page 17 16 Deleting Protocols from the dsi protocols Table If you regularly get protocols the dsi protocols table can quickly become unwieldy and can have an adverse effect on polling efficiency It is a good idea to periodically clean it up Make a list of those protocols that appear in your reports and use the Protocol Definition Interface to delete any unnecessary protocols 1 Launch the Protocol Definition Interface as described in The Protocol Defini tion Interface on page 17 6 2 Highlight the protocol you wish to delete and click Delete Protocol The pro tocol is deleted Note The Protocol Definition Interface only allows you to highlight one proto col at a time For this reason you can only delete protocols one by one DataPipe Components 17 15 17 RMON Support Combined Protocol Identifier and Parameter String Values 17 16 In general protocols used in conversations about RMON are only identified by the dotted name notation However the dotted name notation is ambiguous when iden tifying entries in table protocolDirTable since entries are identified by both the value of a protocol identifier string and a protocol parameter string Also it is pos sible for a protocol to have multiple encapsulations under a parent protocol To
400. prove throughput and data redundancy high availability These subsystems are the preferred choice for large installations or for small to medium installations where high availability is desired and where hardware budget is not a consideration Operating System Some operating systems have features that need to be enabled to allow Sybase to make the most out of the available operating system resources For AIX and HP UX asynchronous I O processing should be enabled to ensure maximum I O throughput for Sybase An easy way to tell if asynchronous I O is enabled for the operating system is to issue an sp configure command from an isql prompt as follows isql Usa P sp configure go The output will have a section that looks something like this Operating System C 5 C Performance Tuning Issues Group O S Resources Parameter Name Default Memory Used Config Value Run Value max async i os per engine 2147483647 0 2147483647 2147483647 max async i os per server 2147483647 0 2147483647 2147483647 o s asynch i o enabled 0 0 0 0 Notice the line that reads o s asynch i o enabled This is a read only Sybase configuration parameter you cannot change it Note however that if the run value is zero asynchronous I O has not been enabled for the operating system kernel In this case ask the appropriate system administrator to enable it Note that the text may differ slightly depending on the version of Sybase used Sybase Configuration P
401. ption then the samples are rejected The default value is 10 gt s D E v o Q o o A 47 TRENDrank TRENDrank TRENDrank populates a ranked table on selected metrics from a summary table It enables you to rank and compare selected metrics from the current period with the previous period trendrank d debug level f definition file 7u V Options TRENDrank has the following options f Path and name of the definition file d Sets a debug level Values of 1 2 and 3 are valid The higher the num ber the more detailed the information Debug output is written to the standard output destination This option should only be used for testing in coordination with DeskTalk Systems Technical Support u Displays the syntax for the TRENDrank utility V Displays the version stamp for the TRENDrank utility Note that this option is in UPPERCASE Description TRENDrank populates an output table that has the following columns for each metric being ranked TRENDrank Column Description Current value Current value of the metric Column name is prefixed with the characters CV Current rank Current rank of the metric among the elements being ranked Column name is prefixed with the characters CR Previous value Previous value of the metric Column name is prefixed with the characters PV Previous rank Previous rank of the metric among the elements being ranked Column name is prefixed
402. qual to 3 a m and ta_period greater than or equal to October 1 1998 and less than or equal to November 1 1998 trendcopy g 1 3 e 19981001 19981101 s xyz S XYZ Note The hour of day range used with the g option is 1 through 24 and does not wrap around to the following day To specify an hour range that wraps around to the following day two command line options are required e g trendcopy g 20 24 g 1 7 s xyz S XYZ will copy data from 10 00 p m to 7 00 a m 18 12 Usage Notes TREND 3 6 User s Guide Using the v and n Options To Filter by View and by Type TRENDcopy allows you to copy information about the nodes contained in a view or a type If you do not specify either a view or a type or both TRENDcopy uses the nodes contained in the topology database of the default group The T option is only used to specify a different topology database For example assume you have a view named alabama containing all of the nodes on your network that are located in the state of Alabama You want to copy the data concerning this view from OLYMPUS_SYBASE to VESUVIUS_SYBASE Use the following command trendcopy s OLYMPUS SYBASE v alabama S VESUVIUS_SYBASE a JI m 2 o t o lt Using the b and I Options To Copy Data TRENDcopy provides two parameters b and 1 lowercase letter 1 that used in tandem allow you to copy data from a specific baseline period in time The b option s
403. r D WWW or SNMP traffic snmp octets The total number of SNMP octets counted total ip octets The total number of octets counted including SNMP WWW and other traffic www octets The total number of WWW octets counted A 55 TRENDstep The value shown for snmp_octets in Figure A 2 is 3 692 which differs from the value shown in Figure A 1 for SNMP octets which is 5 264 118 These values differ because TRENDstep has converted the value in Figure A 1 which is a simple counter to a rate value in Figure A 2 The Delta Time for that row in Figure A 1 is 1 426 seconds Thus if the RMON agent recorded 5 264 118 octets of SNMP traffic for the polling period SNMP octets were passed at a rate of 3 692 octets per second the value shown in Figure A 2 All statistics in the output table have been converted to octets per second Now look at the column total ip octets for the 1 40 p m polling period the value in this column shows that the RMON2 agent for device 134 70 18 241 recorded an average of 6 851 octets per second of IP traffic We can see that of the total traffic SNMP traffic averaged 3 692 octets per second WWW traffic averaged 2 157 octets per second Combined all other IP traffic averaged 1 003 octets per second Most importantly all of these statistics are now in a single row of a data table which allows you to generate a graphical report that includes all of these statistics as shown in Figure A 3
404. r implementing the server envi ronment described above The discussion assumes a basic knowledge of TRENDit TRENDcopy and TREND architecture Data Management on the Satellite Server D 2 If you do not need to generate reports on each satellite server you do not have to create hourly daily weekly and monthly tables on the satellite server You only need to acquire delta data from the satellite server The hourly daily weekly and monthly tables can then be created at the central server Use the r option in the trendit command that you issue from the trendtimer sched file to specify that only delta data be calculated at the satellite server For example trendit S r Z error Yo However this command calculates raw tables into delta tables serially If your envi ronment is not resource constrained you can save elapsed processing time by run ning multiple instances of TRENDit concurrently using the t option See Performance Tuning Issues on page C 1 for a complete discussion of the perfor mance implications For example trendit s r Z error_ t table_name This command calculates delta data for only the table specified by table_name In clude a trendit command line with the t option in trendtimer sched for each raw table from which you want to calculate delta data Data Management on the Satellite Server TREND 3 6 User s Guide If your servers are set up in a UNIX environment you can download the fol
405. r the port num ber or the server is down In this case verify the host name of the server and port number If you are still having difficulties contact DeskTalk Systems Technical Support 13 Click OK to return to the Ping window In the Ping window click Done to return to the DSEDIT1 InterfacesDriver window Select Exit from the File menu to exit from the Dsedit utility D 20 Creating the Sybase Interfaces File on Windows Systems TREND Discovery Considerations This appendix describes how TREND stores nodes and how you can modify Discov ery files The discussion addresses the advanced user who needs an understanding of the schema and wants to expand the list of devices that TREND auto discovers A Wellfleet device is used as an example in this discussion TREND uses two node lists A list of discovered nodes stored in the dsi_node_list table A list of manually entered nodes stored in the dsi nodes table The user interface allows you to look at node types to see the discovered nodes that are in each type but does not allow you to see just a list of all discovered nodes However you can use the following commands from an isg session to see just the list of all discovered nodes isql Udsi dpipe Pdsi dpipe 1 gt select dsi target name from dsi node list 2 gt go 9 om oo Qn Rie o2 5 lt 7 E Discovery Considerations E 2 The following query lists all devices that have been discovered bu
406. r valid data The value is the percentage of difference between the delta values of two Received Timestamps and two System Uptimes These statistics come from two consecutive raw data samples mw_collect For example if rl and s1 are the Received Timestamp and System Uptime for the first sample and r2 and s2 are the Received Timestamp and System Uptime for the second sample then the calculation for the value is r2 r1 s2 s1 100 r2 r1 During processing if the calculated value for the samples exceeds the value set by this option then the samples are rejected The default value is 10 This option is in UPPERCASE This option is equivalent to the zerror parameter in TRENDpm Specifies the minimum disk space as percent of total disk space that must exist on the disk where the SCOLLECT HOME directory resides If less space exists on the directory mw_collect does not execute The default is 10 which represents 10 percent This option is in UPPERCASE Use 3 5 x compatibility That is use the same method for inserting data into the table as version 3 5 x and earlier Sets the debug output level for the TRENDpm process of mw_collect Values of 0 1 2 or 3 are valid The higher the number the more detailed the information The default is 0 which means no debug output Debug output is written to standard out You should only use this option for testing in coordination with DeskTalk Systems Technical Support d
407. rallel However TRENDproc allows an even more sophisticated way to run commands in parallel A trend proc file can call one or more additional trend proc files In fact multiple instances of TRENDproc can run concurrently each running parallel pro cesses The syntax for this call is trend proc f trend proc filename gt This process of creating a trend proc that contains one or more trend proc files can be especially useful Since it is possible for trend proc files to conflict with each other it is sometimes necessary to ensure that one TRENDproc is complete before another is launched If however you create a trend proc file whose blocks are trend proc files you can launch an entire sequence of trend procs with one entry in trendtimer sched Note While it is often a good idea to have a master TRENDproc controlling your trend proc files you should limit this nesting to two levels Otherwise you increase the chance of creating a conflict between your various processes Example You have created five different trend proc files Aspen Birch Casuarina Dog wood and Elm each of which performs the data collection data aggregation and summarization and report generation for a different set of reports Since some of these trend procs utilize data from the same raw data tables it is necessary for these TRENDprocs to be run one at a time You could create a separate entry in trendtimer sched for each trend proc file
408. rent Node on page 6 29 The Community Strings option allows you to temporarily override community strings and specify new values to be used in MIBwalker MIBwalker reads the community strings for the current node from the database See Changing Community Strings on page 6 30 Navigating the MIB Map 6 8 Most MIBs are too large to display all at once in the main window MIBwalker provides several methods of walking through the structure of the currently loaded MIB The simplest method for seeing parts of the MIB map not currently visible is to use the scroll bar to the right of the map Using Object Path Radio Buttons The row of buttons immediately below the menu bar displays the path to the currently selected object in the MIB map The following example shows the buttons that display to the enterprises level of the MIB tree org dod internet private enterprises Figure 6 7 MIB Object Path Radio Buttons Click any of these buttons to jump to the object named in the button This is a useful way to work your way back up the MIB s tree structure The MIB map will start with the chosen object Main Display and Menu System TREND 3 6 User s Guide Using the Next Level Box The scrolling box located to the left of the MIB map and just below the object path radio buttons displays the names of the objects in the next level of the MIB tree below the currently selected object The following example shows the
409. report definition file If so those times are displayed after the Time Period label in the Report Launcher window Report Launcher BEI File View Look in c trendsnmpjreports Folders Top Level Report Directory Up one level Reports Time Period Daily RMON Ethernet Statistics Report Types fan Types Select Devices View Report Prev Day Set Time Figure 14 12 Report Launcher Window Time Period Considerations 14 18 Specifying the Report Time Period TREND 3 6 User s Guide If no start and end times are specified in the report definition file as in Figure 14 12 you must click Set Time to specify the start and end times If a start and end time are displayed you can also click Set Time to change the times You can specify or change the hour day and week for the Start and End Time To do this 1 Click the Set Time button in the Report Launcher window see Figure 14 12 The Set Time Period window appears Set Time Period gt pu 1 m o c 3 1 IT e Figure 14 13 Set Time Period Window 2 Set the start and end times by clicking the Hour Day and Week radio buttons respectively Click the button to increment the selected hour day or week Click the button to decrement the selected hour day or week Click Clear to clear the displayed Start Time and End Time 3 After you set the Start Time and End Time click OK to close the win
410. requests with a 20 minute interval in voke rmon_collect every 10 minutes with the command line gt D v o Q o o rmon collect i 20 To have rmon collect execute all collection requests with a 60 minute interval which are to be run from the machine on which it is installed putting a maximum of 20 MIB variables in each packet trying to poll the node a maximum of 3 times waiting 2 seconds between polling attempts invoke rmon collect with the command line rmon collect n i 5 p 60 o 2 r 3 A 37 rmon_collect Note that the interval polling entries for rmon_collect are in the trendtimer schedule file SDPIPE HOME lib trendtimer sched Direct Polling To have rmon collect poll a probe called rmon1 Desktalk com for the RMON eth ernet History table use the command rmon collect h rmonl Desktalk com t rmon history A 38 TRENDit TRENDit Cleans raw data samples and aggregates raw data into periodic summaries trendit a b c column name value d debug level f seconds past hour h m high water mark default 90 gt p r s t table v host and matrix zero point threshold V w z delta time error percentage Description TREND it aggregates raw sample data into hourly daily and weekly summaries giv ing the minimum maximum average and total for all counter and gauge fields It adds a normalized timestamp ta period for each roll
411. rking with TRENDabview TREND 3 6 User s Guide If you wish to save the view with a different file name or to a different directory choose Save As rather than Save The Save As Dialog Box appears allowing you to define in which directory and under what name the view will be saved 4 Once you have saved your view you can run it By doing so you create a vir tual table in your database From the File pull down menu choose Run The Select File Dialog Box appears then click Run Note You must save your view before you can run it The Run command does not interpret the current information in TRENDdbview Rather it causes the saved vwb file to execute For this reason if you do not save your view it is not available to run N N D m z g ex o Once you have run your view the creation process is complete The view exists and acts as a virtual data table You can use it to generate TRENDsheets and TRENDegraphs and you can audit it in Data Manager Modifying an Existing View Once you have created a view you can open and modify it at any time Then if you wish you can save it as a different view or replace the existing view with the newer version Replacing an existing view is especially useful if you find that its current parameters are too general Open an Existing View 1 Start TRENDdbview and go to the main window 2 Inthe File pull down menu select Open The Open File Dialog Box ap
412. ror rates will be reported in the spreadsheet see Establishing Constraints on page 11 12 Use Display Options to turn the grid on or off select font point size set the maxi mum number of rows displayed and specify titles and subtitles Selecting the Refresh option clears the report window and repopulates with the most current data available Edit Edit Current View To modify the definitions setup through TRENDbuild click the Edit Current View option see Modifying the Tabular Report on page 11 9 Opening an Existing Tabular Report To open a pre defined TREND report select the Open command from the File menu The query file will be opened via the File Selection window described in The Open Command on page 1 15 11 4 Opening an Existing Tabular Report TREND 3 6 User s Guide Using an Existing Report as a Template There may be times when you want to use one report as a template for another To do this open the existing report and click the Save As option under the File menu This will bring up the File Selection window see The Open Command on page 1 15 allowing you to write a new report file to disk The new file can be modified to create a different report then saved to preserve the changes Creating a New Tabular Report New reports are defined using TRENDbuild see TRENDbuild on page 10 1 You can invoke TRENDbuild from inside TRENDsheet by selecting the New option under t
413. rs Input Server Name o Cone 4 Enter the database server name In this example the server name is BANYAS SYBASE which is on a UNIX platform D 16 Creating the Sybase Interfaces File on Windows Systems TREND 3 6 User s Guide Note The Sybase server name on a UNIX platform is typically hostname_SYBASE The Sybase server name on an NT platform is typi cally the same as the host name 5 Click OK The DSEDIT1 InterfacesDriver window reappears BE dsedit DSEDIT1 InterfacesDriver Server Entry Version Server Name x Server Service SQL Server Server Status 4 Unknown Server Address o Dm Ux lt iy m E E e 3 3 2 apjajes e dn Bunes q Server Address attribute Creating the Sybase Interfaces File on Windows Systems D 17 D Setting Up a Satellite Server Environment 6 T 8 Double click on the Server Address attribute in the Attributes column The Network Address Attribute window appears Network Address Attribute WE Down Click Add The Input Network Address For Protocol window appears Input Network Address For Protocol roe mvaa o Coco In the Protocol list box click the downarrow and choose TCP In the Network Address field type the host name of the server banyas in this example a comma and the port number to which the server listens 2052 in this example Note
414. rt definition file so they will automatically be used the next time you run the report Specifying Data to be Included in the Report TREND 3 6 User s Guide Displaying Data from a Different Database Access a different SQL Server to display its version of the information presented in this report Click the Change Database option under the File button to bring up the Change Database window Change Database CIMARRON_SYBASE COK Cancel Figure 11 5 Change Database Window The name of your current database is displayed in the Change to box To change to a different server click the down arrow in the Change to box to see a list of the Sy base SQL Servers known to this machine The values presented here are read from the SYBASE interfaces file Select the name of the server you want to access in the report Click the OK button to change to the selected database Click Cancel to quit out of the window without changing servers ok J m z 7 2 i rec Specifying Data to be Included in the Report 11 11 11 TRENDsheet Setting the Time Period TRENDsheet allows you to set a Start Time and an End Time for displaying data for reports If you select Time Period from the View Menu of TRENDsheet the Set Time Period window is displayed Set Time Period Figure 11 6 Set Time Period Window Change the Start and End times by selecting the button that corresponds to the por tion
415. rted by the RMON2 DataPipe Category Database Name Alias Name MIB Table Ethernet rmone_ethstats_ rmone_ethernet statistics Control information from Ether Statistics StatsEntry Ethernet rmone_ethstatsdata_ rmone_ethernet statistics data Statistics from EtherStatsEntry Statistics and EtherStats2Entry 1 of 3 17 2 Source Data Available from RMON2 TREND 3 6 User s Guide Category Database Name Alias Name MIB Table Token Ring rmone_trpstas_ rmone_token ring promiscous statistics Control information from Promiscuous tokenRingPStatsEntry statistics Token Ring Promiscuous statistics rmone_trpstasdata_ rmone_token ring promiscous statistics data Statistics from tokenRingPStatsEntry and TokenRingPStats2Entry Token Ring rmone_trmstas_ rmone_token ring maclayer statistics Control information from MAC tokenRingMLStatsEntry statistics Token Ring rmone_trmstasdata_ rmone_token ring maclayer statistics data Statistics from MAC tokenRingMLStatsEntry and statistics TokenRingMLStats2Entry Ethernet rmone_hist_ rmone_history HistoryControlEntry and History historyControl2Entry Ethernet rmone_histdata_ rmone_ethernet history data EtherHistoryEntry History Token Ring rmone_hist_ rmone_history HistoryControlEntry and MAC History historyControl2Entry Token Ring rmone_histtrmdata_ rmone_tokenring maclayer history data TokenRingMLHistoryEntr
416. rules apply Top of MIB Tree You must include the top of the ISO Tree down to start of the MIB in the file Make sure the following lines are included 5d I internet OBJECT IDENTIFIE iso org 3 dod 6 1 internet 1 directory OBJECT IDENTIFIER mgmt OBJECT IDENTIFIER t internet 2 Preparing MIBs for MIBwalker 6 11 6 MIBwalker 6 12 Below this point the appropriate part of the tree that reaches the start of the provided MIB must be added For a vendor MIB you most likely need private OBJECT IDENTIFIER internet 4 enterprises OBJECT IDENTIFIER private 1 Many standard MIBs defined in RFCs are under the MIB II tree For these MIBs add the following after the line defining mgmt mib 2 OBJECT IDENTIFIER mgmt 1 jJ Description Length DESCRIPTION clauses containing more than 3000 characters are truncated to the first 3000 characters Object Name Length The maximum length of a MIB object name is 63 characters Object names of 64 characters and above are truncated to 63 characters Exceeding the standard length however does not affect your ability to collect or report on the field since the field is polled by its object identifier which is unchanged The field name is a just a label STATUS Field Values The legal operands for the STATUS field are mandatory current optional deprecated obsolete Preparing
417. rver reports are still generated or attempt ed depending on the rollup tables that are used in the reports For example if no data is available for hourly or daily based reports weekly reports containing only the rollup data that was previously collected are generated However all data is time stamped so the delay in getting data to the central server does not affect its us age when it finally gets there When TRENDcopy is Delayed TREND 3 6 User s Guide Node Lists and Polling Policy In general the central server does not need to know about nodes node types polling policies etc because it only aggregates and reports data that is maintained by unique property key table records Thus the recommended approach is for the satellite servers to maintain their separate node lists and polling policies for those node lists to perform separate current node type discovery and to perform specific IP address range discovery for new nodes where range on one satellite does not overlap with the range on another satellite Then the TRENDit and TRENDcopy processes as detailed above can be config ured on each satellite sever to place delta data onto the central server ie i o lt D S m 3 Z E 3 iv ko 3 apjajes e dp bumas a If you want to deviate from this approach discuss your plan with your Desktalk Sys tem Engineer or Customer Support person Firewall Considerations Several TREND func
418. rver without the assistance of an experienced database administrator Replication can be ad hoc using a command line interface or scheduled on a period ic basis Database tables can be created and or incrementally copied to another da tabase based on user definable criteria such as thresholds time spans or sets of nodes so that the new database contains only the information required for the re porting applications TREND is available with satellite database servers allowing data to be collected and stored for a period of time at any physical location Incremental periodic data trans fer can be triggered from a satellite location into a regional or central data repository This action provides scalability by fully distributing the data collection and manage ment throughout your enterprise For example you may wish to conserve WAN bandwidth by distributing polling agents and servers to remote locations where data is periodically collected and managed in local repositories Then during off hours aggregated data can be moved to a more central database for reporting and analysis functions The satellite database server package contains the satellite database server and a poller Advanced TREND features TREND uses the concept of a special user A user logged on as trendadm has priv ileged access throughout the TREND modules For example no user other than trendadm is allowed to set table aging criteria or to delete tables from the
419. s From the main window you can create a new view or modify an existing one These commands are located in the File pull down menu which also contains the options for saving and running a view Building a View TRENDodbview is used to form a view combining disparate columns from different data tables key tables and views This creates a virtual data table and enables you to view and manipulate the specified data You can then use TRENDbuild to gener ate TRENDsheet and TRENDgraph reports based on the view Simply decide which columns of data you wish to include then use TRENDdbview to create the desired view Working with TRENDdbview 22 9 22 TRENDabview 22 10 In the course of building a view you will need to determine its characteristics This involves up to seven procedures some of which are optional Specifically you need to Create a new view Once you have created views you can open and modify them Typically though you will build a specific view to gather and display data for a spe cific purpose Thus you will most often use TRENDdbview to build new and unique views Select the tables contributing columns to the view You will build your view based on columns contained in these tables In addition you cannot perform any of the other definitions until you have specified your tables Previously created views can also act as data tables Define the qualifications and expressions to be applied to the
420. s for example bcp isql Cache Pools Determine your current cache pool configuration by entering the following command isql Usa P sp cacheconfig go C 10 Sybase Configuration Parameters TREND 3 6 User s Guide The output looks something like this 9 O 2h e 3 O a O D el 3 Ce Cache Name Status Type Config Value Run Value Soratie data asahau Active Default 0 00Mb 112 11 Mb Total 0 00 Mb 112 11 Mb Godet serai daba caches Praat dives ota 58 Config Size 0 00 Mb Run Size 112 11 Mb IO Size Wash Size Config Size Run Size 2Kb 512Kb 0 00Mb 112 11 Mb By default Sybase ships with only one data cache called default data cache This cache has only 2K buffers in it You should add a 16K buffer pool to the default data cache The size varies depending on the available memory For constrained environments start by allocating about 10 percent of the total available see Run Size value in the above example Start increasing the value until a noticeable performance degradation is evident and then back off some Nonconstrained environments typically can allocate 20 40 percent of the available cache to the 16K buffer pool To change add a 16K buffer pool of 12MB isql Usa P sp poolconfig default data cache 12M 16K go This change is dynamic a Sybase restart is not required Sybase Configuration Parameters C 11 C Performance Tuning Issues Online Engines
421. s are too large If 15 is too large decrease it by 1 until the PDU too large messages stop This option allows the collection of SNMP data from the specified port rather than the default SNMP port of 161 This option is in UPPERCASE Sets the log information level for the parent instance of mw_collect Values of 0 1 2 or 3 are valid The higher the number the more detailed the information The default is 0 which means no log informa tion for the parent instance of mw_collect The log information is writ ten to the log file called SDPIPE HOME tmp parent collect dbg log See Note on page 22 The number of times mw collect resends an SNMP request before assuming the target node is not going to respond The default is 5 SNMP retries Sets the minimum number of rows to collect before starting the loading of data into database The number of rows is an approximation since there is an assumption that each row is 500 bytes The default value is 1000 rows which means that the loading of the file into the database starts when the size is 500 000 bytes 500 1000 This option is in UPPERCASE This option rounds off the collection time ta_period If the mw collect parent kicks off a collection at 3 07 and if you are using the default collection option of 300 seconds 5 minutes the actual ta_period value for the collection will be recorded as 3 05 mw_collect t T u V Specifies the SNMP version This optio
422. s be generated Here are the guidelines for the parameter file The rules for formatting a parameter file are 1 One entry per line 2 All entries of keywords with arguments use the format lt keyword gt lt argument gt 3 Each report to be run must be blocked off with a start and end statement 4 An in the first column indicates a comment 5 Any keywords not listed here will be ignored 20 6 Input Parameter File TREND 3 6 User s Guide Valid keywords for the parameter file are o B ES 5 amp Keyword Q Required Valid Values Default E EE So a lt start Yes N A N A end Yes N A N A report r Yes Any qss file path name None output 0 Yes ASCII file path name None headers a No y yes format f No comma tab tab dsquery q No valid SQL Server name DSQUERY envi ronment variable value max m No n no exception No n no instances No valid target name table key None pairs or file containing tar get name table key pairs See below for format 1 of 2 Input Parameter File 20 7 N E D m 2 o D x o D mg 20 TRENDexporter 20 8 z z B ES 3 amp Keyword Required Valid Values Default Q 52 9 e 2 lt maxlines p No any positive integer 5000 condense C No n no style No individual individual combine begindate No da
423. s chapter describes the TREND Discover tool which can be used to get informa tion about the nodes on your network The TREND Discover tool allows you to conduct a search that can Find the nodes on your system Ascertain whether or not each node is SNMP manageable Determine what kind of device router hub switch etc the node is 9 9 9 Automatically update the tables that control data collection As a result Discover gives you the option of using an automated process to define your network Note The use of Discover is optional You can also populate and update your various node tables manually if you choose 5 1 5 Discover What Discover Does Discover can conduct two different kinds of searches It pings a range of IP address es one at a time and then analyzes and identifies the responses it gets It uses that information to populate or update tables in your TREND database There are three specific pieces of information Discover can determine from this process Whether or not there is a device at a specific IP Address Ifso whether or not that device is SNMP manageable If a device is SNMP manageable what kind of device it is One kind of TREND Discover process IP Discovery verifies the existence of a de vice at a particular IP address and identifies whether or not that device is SNMP manageable The other Discover process SNMP Type Discovery ascertains what kind of device the node
424. s cron has the proper environment to know about SDPIPE HOME and has permission to access this directory Loading a Database The following example illustrates the simplest procedure for restoring a database from a predefined dump device isql Usa P 1 gt load database dpipe_db from tapedump 2 gt go where tapedump is a tape device see sp_helpdevices or gt E vv 0 3o 0 g o 9H o O o E Backing Up the SQL Server Installation B 23 B Database Administration 1 gt load database dpipe_db from diskdump 2 gt go where diskdump is a dump device see sp_helpdevices or a file 1 gt online database dpipe_db 2 gt go Note Transaction Loading is not discussed here since it is not used Installing a TREND Database On a Raw Device Instructions for installing Sybase on a raw device can be found in the TREND Installation Guide This procedure includes installing the TREND database Shutting Down Sybase To shut down Sybase enter the following commands isql Usa P 1 gt shutdown 25 go B 24 Installing a TREND Database On a Raw Device TREND Performance Tuning Issues Note You can download the scripts mentioned in this appendix from the Desktalk Systems FTP server Call DeskTalk Systems Customer Support for instructions Many factors can positively or negatively affect TREND performance These factors can be grouped into four functional configuration areas Hardware Operati
425. s daily usage Initially you must call DeskTalk Systems Customer Support to obtain a license key for this feature If daily database usage exceeds the purchased database size a daily warning message is displayed on the screen and in trend log You continue to receive these warning messages until you purchase an expanded TREND database and a new license key is generated Enlarging the Database B 3 gt 2 3 D a D Cd o 5 UJ oa 5 m py o P o D B Database Administration B 4 There are two different databases that may need to be enlarged One is the TREND database which Sybase recognizes as dpipe_db You can see how much space is being used by dpipe_db by looking at the Data Manager window Note In general terms if dpipe_db shows more than 70 of the database as used it is a good idea to enlarge the database Other options include deleting unnecessary data or shortening the length of time you store your data The other database that may need more space is the Sybase database tempdb This is where Sybase creates temporary tables as necessary If for example you run a report that is based upon a database view Sybase will create a temporary table in tempdb in which to store the data returned by the view Sybase installation allocates a small amount of space to tempdb e g 2 MB In order to use TREND the size of tempdb must be 30 of the data portion of your dpipe_db database or up to 1
426. s for the Device and Key in the Select Key window 3 Click the Update Statistic button in the Select Data Set window When the statistics are set the way you want click the Graph button to display the data from the new source Note that you select the device key pair for each statistic independently You can have different nodes providing data to the same report You do not have to use a single node as the source for all data You may want to save the report again so the next time it runs it uses the current data source Selecting Different Data 13 13 13 Grade of Service Displaying Data from a Different Database You can access a different SQL Server to display its version of the information pre sented in this report Click the Change Database option under the File menu to bring up the Change Database window Change Database xi Change to CIMARRON SYBASE v OK Cancel Figure 13 7 Change Database Window Note A report may not display the same information for both databases The name of your current database is displayed in the Change to box To change to a different server click the down arrow in the Change to box to see a list of the Sy base SQL Servers known to this machine The values presented here are read from the SSYBASE interfaces file Select the name of the server you want to access in the report Click the OK button to change to the selected database Click Cancel to quit out of the
427. s it possible to identify not only which devices on a network spoke to each other but also how the conversation took place However the collected data only identifies these protocols by their OID strings Be cause these OID strings are not readily understandable TREND allows you to assign them meaningful names Then when reports are generated from the collected data the conversation protocols are recognizable TREND assigns names to the OID strings via a translation table dsi_protocols The dsi_protocols table associates the OID strings with their appropriate name identifier Note RMON2 protocols have a defined naming convention See Combined Protocol Identifier and Parameter String Values on page 17 16 DataPipe Components TREND 3 6 User s Guide In most cases it is not necessary for TREND users to know the RMON naming convention or to manually populate the dsi_protocols table The RMON2 DataPipe includes a graphical interface that automates the population process Lj N E fe 2 N 7 Ko 3 The Protocol Definition Interface is launched from the command line At the command line 1 Change directories to TRENDsnmp bin 2 Type the following r2proto This brings up the RMON Protocols window Rmon Protocols BEI File EA ad sach 23 Current DSI PROTOCOL Table Entries IP Address wildcard ether2 ip 1 8 1 0 0 1 0 0 8 0 2 0 1 Unknown Address wildcard ether2 vip 8 1 DECN
428. s set Lj D fe 2 7 o 3S A Polling Frequency 16 5 16 RMON Support 16 6 The polling interval used for History data should be based on the total timespan cov ered by the samples maintained by the RMON agent or probe The typical default short History spans 25 minutes i e 50 samples or buckets of 30 seconds each If these default settings are used polling on a 20 minute basis should ensure collection of 30 second grain data without gaps while polling once a day should ensure collec tion of one hour grain data without gaps Note History group polling should be defined with an interval that is less than the total timespan covered by the probe In this way variances in network and probe response times are transparent due to collection of overlapping data The duplicate data is not inserted into the TREND database Probes and agents can support multiple Historys for a given data source It is usually recommended that there be at least two History entries per monitored interface An entry with a short sample period of 30 seconds enables the detection of sudden changes in traffic patterns A long sample of 30 minutes on the other hand reveals the steady state behavior of the interface Since TREND will collect the short history data and aggregate it over time probe resources can be conserved by only configuring a single History where the interval yields the finest grain data of interest TREND s cap
429. s through the Set Application Color op tion of the TREND Configuration menu The Background Color window contains a palette of 64 standard background colors plus sliders for infinite number of user de fined colors To use this application simply select an application select a color and click on Apply Then click on Cancel to Exit N m 032 AB no 3 o as o fo Q 5 P at e 3 Changing Data Rollup Parameters TREND s Data Manager function provides an easy way to read tabular display of in formation about every data table in your database s plus a central location for con trolling data rollup and aging Any TREND user can view information about the database via Data Manager Only the administrative user trendadm can use Data Manager to change defaults or to modify aging and rollup for individual tables ata Manager ed 11 object Name SQL Name Twme Rallup Size rows Size KB Keep data for aay_cisco10e_leysterpen O100_leystem_pen_ dam fo 32 aeram hour ciscoT08 Jeystem pe HiOO Jeystem per hourly of 2 seraut rate ciscot00 system per RTO0 system ner rate of 32 defaut week ciscotO0_tsystem p Wi00_lsystem_pert_ weeky 3 defaut Figure 2 8 The Data Manager Main Window Setting Application Background Color 2 13 2 TREND Configuration for trendadm Changing System wide Rollup and Aging Defaults 2 14 You have the choic
430. s timestamped every 20 minutes If polling once per hour select long History Each hour 50 buck ets will be collected of which 48 will be thrown out leaving 2 samples to be stored in the database If polling once each 20 minutes select short History adding approx imately 40 samples to the database note that the actual number may not be exactly 40 since the polling interval is not precise in a network environment Aggregating Host and Matrix Tables and Managing Size The Host and Matrix tables have the potential to be extremely large and unwieldy To minimize this possibility the TREND rollup process TRENDit suppresses zero rate rows in the raw tables pertain to pairs of devices that did not exchange any data during the polling period TRENDit can also be configured to delete raw data below a specified threshold as it is rolled up See Man Pages on page A 1 for details on using TRENDit Note If noddy talked to notable at some point an entry exists in the RMON probe s matrix table for noddy notable However they may never or rarely communicate again so the probe is returning a zero value row for noddy notable each time the matrix table is collected By discarding pon o D fe z 7 KS E A Aggregating Host and Matrix Tables and Managing Size 16 7 16 RMON Support such rows from the database before they are aggregated and carried into the rate hourly daily and weekly tables the already large m
431. s to process data The procedure cache should be at least 20 of the configuration memory You should use the following formula to calculate the amount of procedure cache required 1 Multiply the number of user connections C by the largest stored procedure size P which is typically 3 MB 2 Multiply that value by an overhead factor of 1 25 3 Divide that value by the configuration memory M which is in 2K pages Divide the configuration memory M by 500 since there are 2K pages in 1MB to use the configuration memory value allocated to Sybase in MB 4 Multiply by 100 for the percent value Procedure Cache Percent C P 1 25 M 500 100 For example if there are 100 connections the procedure cache value is 100 3 1 25 2 37 5 MB If the configuration memory is 180 MB which is 90000 500 the procedure cache percent is 37 5 180 100 which is approximately 21 Ifthe procedure cache is too small the stored procedures are fetched from disk rather than read from cache C 8 Sybase Configuration Parameters TREND 3 6 User s Guide 1 To change the procedure cache allocation enter the appropriate value for your installation instead of the value 21 in the following command isql Usa P sp configure procedure cache percent 21 go 2 Restart Sybase to enable the procedure cache allocation Number of User Connections Find the line for number of user connections in the Memory Use gro
432. s will come back with a received timestamp of 1 00 01 or so To make sure those samples data are included in the 12 00 to 1 00 hour set this option high enough to include the number of seconds after the hour they actually arrived The default is 60 Display the command line format help The database full high water mark By default the rollup process at tempts a rollup if the database full is less than or equal to 90 The TRENDit program uses the dbstats table to determine the amount of available space in the database TRENDit P V W Causes TRENDit to truncate the RMON host and matrix raw data ta bles i e delete all rows after rolling them up The default is not to truncate the tables Use this option to only calculate the rates i e intersample delta values The process stops before running the hourly daily weekly and month ly rollups This option will handle counter resets that occur on the device agent side If the device counters are manually reset this option will not allow a high spike in the data which can happen due to the comparison to the sysyuptime which remains constant when counters are reset The name of the raw data table which you want to roll up The tables you can use with this option are shown in Data Manager with type data Use the value in the Data Manager SQL Name column as the argument The default is to aggregate all tables Note that table names that are 30 charact
433. scrolling box that displays for mib II with the internet group selected Figure 6 8 Listing of Objects in Current Level Click the top level to move up one level of the MIB tree Click one of the object names below the top level to navigate down the MIB tree o Ww 2 D Finding an Object in the MIB Map If you know all or part of the name of an object you want to locate in the MIB map use the Find window to go to it Bring up the Find window by selecting the Find option under the View menu Find T Ignore Case Whole Word Partial Word Cancel Figure 6 9 Find Window Type in as much of the object s name as you want or know in the Find box Click the Ignore Case check box to ignore case differences between your entry and actual object names when searching for the object This can be useful because SNMP object names often have embedded capital letters Indicate how much of what you typed in has to match an object name for Find to consider them a match by clicking Main Display and Menu System 6 9 6 MIBwalker on either the Whole Word or Partial Word box If you choose Partial Word Find compares your entry to any portion of the object name however it is case sensitive unless you select the Ignore Case option Click the Find button to begin or continue the search Click Cancel to close the window Loading a MIB 6 10 MIBwalker shows you the name of the currently loaded MIB fi
434. se activity 4 Install a backup Sybase server on host B Make sure both the Sybase server and the backup Sybase server are running on host B See Backup Server on page B 15 for detailed instructions 5 Turn off trendtimer on host A 6 Wait for any collections to complete Then dump the database to an available disk file Make sure you have plenty of space available to create the dump file You can get latest backup by starting an isql session For example F 1 4m an To BE O o D 3 So O os o F Relocating the TREND Database F 2 isql Usa P 1 gt sp_addumpdevice disk diskdump lt path to diskspace gt diskdump dat 2 gt go 1 gt dump database dpipe_db to diskdump with noinit 2 gt go 7 Mount the A lt path to diskspace gt file system for export on B or ftp the diskdump dat file to host B depending on whether you have available disk resources on B to handle the temporary storage of the dumped database file 8 Load the backup on host B by starting an isql session For example isql Usa P 1 gt sp_addumpdevice disk diskdump snew path to diskspace gt diskdump dat 2 gt go 1 gt load database dpipe_db from diskdump 2 gt go 15online database dpipe dp 25go 9 After this completes unmount the A lt path to diskspace gt file system on B or delete the diskdump dat file on host B depending on the method you chose to get the file onto host B 10 Once you have confir
435. se data from the last sample taken for day 1 at tog is compared with the first sample taken for day 2 Thus the first row of rate data for day 2 is not lost As TRENDit continues to process the new samples that are collected into raw tables it appends new rows to the rate output tables in the fashion described above Data Cleaning 9 7 v 5 2 D O E D en 5 uonebaibby ejeq 6 9 Data Aggregation and Manipulation Validity Checking In theory the raw data that is collected during any one sampling period is completely accurate In practice however several circumstances can affect the validity of the collected data such as One or more scheduled samplings fails for some reason e g the network is down One or more devices returns inaccurate data the device is malfunctioning or no data the device is offline A counter reaches its predefined maximum value and rolls over starts over at zero between sampling periods You manually reset some or all counters between sampling periods You shut down and reboot the system between sampling periods which resets the counters to zero If TRENDit senses that the sample is invalid for any of these reasons the sample is not included in the rate output table Rollup TRENDit can also perform a roll up function on rate data if the r option is omitted from the trendit command line In the roll up function rate data is summarized at the lowest
436. se this alternative the TRENDsum files must be changed the data table columns must be renamed the col alias entries TRENDsum must be updated and the query files must be updated to reflect the new column names In this example the column avginutilization needs to be changed to wavinutilization and expressions such as the following in 3 4 x TRENDsum files column inutilization ifinoctets018 8 ifspeed013 100 avg max need to be changed as follows in 3 5 1 TRENDsum files column inutilization ifinoctets018 8 ifspeed013 delta time004 100 0 wav max If you require the weighted average alternative a or b produces the desired result A simpler approach however is to use the simple average of value delta time as in 2 above and accept the small discrepancy in the result of calculating the avg function in 3 4 x and 3 5 1 The tot function treats gauge expressions differently in 3 4 x and 3 5 1 In 3 4 x tot of a gauge averages the values In 3 5 1 tot totals whatever value it is given regardless of data type Note therefore that the result of tot on a gauge in 3 5 1 is meaningless In 3 4 x TRENDsum files you can have avg max tot and perhaps other func tions acting on the same value If you divide these expressions by delta time in 3 5 1 TRENDsum files to maintain 3 4 x compatibility you need to remove tot to a new expression in another line because you do not want to divide the expression by delta time
437. sed as the Top Level Report Directory which initially is also designated as the current directory double click on the directory name in the Directories list box The directory name will appear in the Current Report Directory edit box Alternatively you can type the full path of the directory name in the Current Report Directory edit box Click Apply to apply the changes you have made close the window and return to the Report Launcher window When you return to the Report Launcher win dow the directory named in the Current Report Directory edit box of the Selecting the Report File Directory TREND 3 6 User s Guide Report Directories window becomes the new Top Level Report Directory and the new current directory displayed in the Look in box Selecting a Report Hundreds of reports are available in TREND To make the job of selecting the report to view with Report Launcher easier you can specify filters to restrict the number of reports devices and elements from which you choose Follow these steps to tailor the list of reports that appears in the Reports list box in the Report Launcher window A D m o i 3 7 T I 1 Select the desired report type from the Report Types list box Figure 14 7 shows sample list box entries Report Launcher ox c trendsnmp reports Interfaces Top Level Report Directory Up one level Forecast Daily Down Interfaces For Admin and Oper
438. see the selection in the Select Group window as shown in Figure 10 3 See About Data and Data Tables in TREND on page 10 2 for a description of the contents of these roll up tables and groups Scroll through the entries in the Select Group window to find the one you want Select a group by clicking on a group name listed then clicking on the OK but ton or double click the group name Click on the Cancel button to quit out of the window without making a selection The name of the group you choose will appear in the Group box in the main TRENDbuild window Selecting Statistics Statistics are individual data fields in a group Once the category and group are cho sen you can build the list of which statistics in the group will be in the report using the Select Statistics window Bring the window up by clicking on the Select Statis tics button in the main TRENDbuild window 10 8 Creating a New Report TREND 3 6 User s Guide Select Statistics ta p dsi target name Target Name dsi table key Table Key dsi descr Description user name User Name Aggregation Type received ts Received TS request ts Request TS delta time Delta Time mh i E m z g ley Q Figure 10 4 Select Statistics Window Note Before selecting specific statistics it is important to consider the unit of measurement for each statistic Instead of selecting and using statistics in their natural form you may inste
439. shows the result Note that you can stretch the window if the description is truncated F i i Description The type of interface distinguished according to the physical link protocol s immediately below the network layer in the protocol stack or ius B F Figure 6 3 Description Pop Up Window from MIBwalker Enumerated A pull down list that contains the set of valid values for the List selected object The values are usually numeric and the MIB gives the meaning for each value This list appears only when it exists for the selected object The processing method for the following object types Table 6 1 varies even though they have the same database storage type See Data Aggregation and Manipulation on page 9 1 for more information regarding the rollup of different data types Table 6 1 SNMP Object Types Defined SNMP Object Type Database Storage Type INTEGER float OCTET STRING varchar OBJECT ID varchar GAUGE float TIME TICKS float Main Display and Menu System 6 5 6 MIBwalker Table 6 1 SNMP Object Types Defined SNMP Object Type Database Storage Type COUNTER float COUNTER64 float OTHER N A Note Make sure to validate the Object Type of each variable prior to creating your MIBlet The MIB must be modified to reflect the proper Object Type to achieve the desired processing method See Validating Object Types on pag
440. signed a ta_period value of Jun 22 1998 12 15 AM Raw data tables then contain rows of data items for each sampling period All rows with the same virtual timestamp ta_period belong to the same sample Data Cleaning 9 4 TRENDit cleans the data that has been collected in raw tables and stores the results of the cleaning process in rate tables The cleaning process has two parts 1 SNMP counter data is transformed into delta data 2 Raw samples are checked to see if they are valid Data Cleaning TREND 3 6 User s Guide Transforming Counter Data into Delta Data The value of an SNMP counter increases continuously from zero by 1 until the counter reaches some predefined maximum value At that point the counter starts over at zero Thus the value of an SNMP counter data type is an integer that has no meaning unless it can be related to a time interval For example a counter named inoctets counts the number of bits input to an interface on a router If inoctets has the value 1 000 000 in a given sample the value is meaningless unless you can state it as a rate such as 1 million bits per second Only then can you calculate a useful metric such as throughput uonebaibby eleg 6 D 5 2 D O D 5 TRENDit enables you to do this by turning raw SNMP counter data into delta data TRENDit invoked with or without the r option performs this function by examining pairs of rows in adjacent sampling periods
441. specify mytest for this name and the input table contains daily rollups TRENDstep creates an output table with the SQL name of Dytest and an Object name of day_mytest This statement appears as the second exe cutable line of the input file and is required TRENDstep creates this table if it does not exist or TRENDstep appends data to this table if it already exists define IGNORE_NULL Specifies to ignore default rules for NULL values When a null value appears in an expression the default result is NULL String variables are exceptions A NULL string value is assumed to be an empty string If this statement is used a NULL value in and expressions is treated as zero For more information see Handling Null Values in Expressions on page A 65 TRENDstep Statement Description define TIMEADJUST This statement enables the ta_period delta_time adjustment zero rate row assumption mechanism for rate tables That is TRENDstep determines the mini mum polling cycle for the entire table and then adjusts corresponding ta_period and delta_time columns to account for the resulting polling cycle When you use this statement make sure the polling interval for all rows in the input table is the same otherwise you will generate invalid data If this statement is omitted and the input table has rate data the output column will have rate values for counter columns That is we will div
442. ss file name extension The utility uses the source SQL in the file including the constraints as the basis for a database query and stores the rows returned as an ASCII file The query can be limited to a single device in stance pair or run against a set of pairs The output can be saved in a separate file for each pair or all the data can be combined into a single output file trend_export can be invoked on a single TRENDsheet report or for a set of reports by using an input parameter file The ASCII files generated can be in either a tab separated format or a comma sepa rated values CSV format The files can contain or omit column headers A facility for generating data for just the most recent time period or between to specific dates is also available The number of lines to be processed can be limited which prevents the generation of overly large output files Lastly each report to be run either individually from the command line or through a parameter file can query any SQL Server system accessible from the user s machine Overall Process and Functionality TREND 3 6 User s Guide Controls and Switches There are multiple parameters to control the execution of the program The com mand line options are Flag Description Argument a Include column headers y The ASCII file will include Headers include both a column headers DEFAULT line for the column names and a line of hyphens underlining each column
443. ssages in the log file preceding the ones written from the beginning of the last boot sequence B 12 Maintaining the Error Log TREND 3 6 User s Guide Fixing a Suspect Database A suspect database is often caused by incorrect ownership of the directory containing the database file and of the database file itself This can happen if the user installs dpipe_db and or dpipe_tlog in the DPIPE_HOME directory or in a directory owned by another application e g CiscoWorks To fix the problem change the ownerships of the subdirectory containing the database file and the database file itself to sybase Once this is done you need to have Sybase reset the database status from suspect back to normal To do this follow these steps 1 Become the sybase user and login to Sybase as the system administrator In this example there is no password su sybase Password sybase isql Usa P 2 Allow updates to the system databases 1 gt sp_configure allow updates 1 2 gt go 1 gt use master 2 gt go 1 gt begin tran 2 gt go 1 update sysdatabases 2 set status status amp 256 3 where name dpipe db 4 go 1 row affected 3 Ifthe previous command returns anything other than 1 row affected type 1 rollback tran 2 go and start again from the begin tran command If the command does return the message 1 row affected continue with the following command to commit the transaction Fixing a
444. styles and colors control X and Y Axis scaling assign report titles and axis la bels and plot multiple statistics from single or multiple sources simultaneously time synchronized for comparison Although data are stored in a relational data base the user does not have to know SQL to create or run reports All functions can be carried out through TREND s windowed interface Main Display and Menu System You invoke TRENDgraph from the Tools menu of the Trend main window The TRENDgraph window appears initially with a blank display area and a menu bar providing access to all TRENDgraph functions explained in this chapter Figure 12 1 N E m z O Q o o T 12 1 12 TRENDgraph P2 TRENDgraph Figure 12 1 TRENDgraph Main Window Menu Command Summary File Use the New command to create a new graphical report see Creating a New Graph on page 12 5 The Open command is used to retrieve a pre defined graphical report see Opening an Existing Graphical Report on page 12 4 12 2 Main Display and Menu System TREND 3 6 User s Guide The Save option is used to store previously saved reports see Selecting Statistics to be Graphed on page 12 6 The Save As option is used to store new reports or to copy or move existing reports see Selecting Statistics to be Graphed on page 12 6 Graphical reports can be printed to a postscript printer or to a file using the Print c
445. systems that are dedicated to running TREND this parameter should be set to 128 MB or at least half of the total RAM in the system In some cases 60 or 70 per cent or more can be allocated to Sybase 1 To change the total RAM allocated to Sybase isql Usa P sp configure total memory 100000 go Memory is allocated in 2K pages so the above example results in a 200MB allocation 2 Restart Sybase to enable the new memory allocation Take care when you change this parameter Some platforms Sun Solaris for example Ay require changes to system memory configuration parameters as well etc system Furthermore take care to ensure that client tasks and Sybase do not cause the system to begin paging thrashing You can check for this with Performance Monitor under CAUTION Windows NT or by using sar or vmstat on the various Unix ports Sybase Configuration Parameters C 7 9 O 2h e 3 O a O D el 3 Ce C Performance Tuning Issues Procedure Cache Percent Find the line for procedure cache percent in the Cache Manager group in the configuration output listing Group Cache Manager Parameter Name Default Memory Used Config Value Run Value memory alignment boundary 2048 0 2048 2048 number of index trips 0 0 0 0 number of oam trips 0 0 0 0 procedure cache percent 20 32934 20 20 total data cache size 0 126978 0 126978 total memory 7500 180000 90000 90000 The TREND application uses stored procedure
446. t Expression Name Text Field A cursor appears indicating that you can enter text 4 Type the name you wish to give to the expression in the Expression text Box and click Insert The name of the expression appears in the Defined Expres sions Box and both the Current Expression Name text field and the Expression text Box are cleared If you wish to replace or change the name of an expression you have already defined Highlight it in the Defined Expression Box Its name will appear in the Current Expression Name Text Field and its text will appear in the Expression Text Box Make the desired changes to either the name the expression text data type or combination of the three and click Replace Again both the Cur rent Expression Name Text Field and the Expression Text Box clear If you changed the name the new name replaces the highlighted name If you modified the expression text it replaces the text of the previously defined expression The previously defined expression is discarded Working with TRENDdbview 22 17 N N D m z o ex o 22 TRENDabview If you define an expression then decide you don t wish to apply it to your view highlight it in the Defined Expression Box and click Delete The expression is removed If you wish to delete more than one defined expression you must delete them one at a time However if you wish to delete all of your defined expressions simply click Delete Al
447. t Instances _New Update Delete _Parameters Figure 15 1 Report Scheduler Window Menu Command Summary 15 2 One command menu File is available from the Report Scheduler window the op tions available are Add New Report Add Drilldown Report Save Print Now Exit About Printing a Report TREND 3 6 User s Guide The Add New Report option under the File menu allows the user to assign an ex isting report to be printed periodically see Adding New Reports on page 15 4 The Add Drilldown Report option under the File menu allows the user to link related reports for automatic printing with the main report see Report Chaining Adding Drilldown Reports on page 15 4 The Print Now option under the File menu prints the highlighted report for the pre vious day s data The Save option should be used before printing Note You must have data e g from the previous day in order for a report to print Scheduled Reports All report information is presented in the Table Display area If all the information cannot be displayed in a single window a scroll bar makes it easy to move through EN the data in the table The display area is organized into the following areas e gt o Report Contains a list of reports selected by the user El n Directory Contains the directory path where the report is located m Printer Contains the name of the postscript pr
448. t a node from the list of Current Hosts by clicking on the down arrow next to the Current Hosts box Click the OK button to make this the new current node Click the Cancel button to quit out of the window without changing the current node 6 29 o gt w 2 x 6 MIBwalker The name of the current node is displayed in the MIBwalker Main window header bar Changing Community Strings 6 30 MIBwalker uses the current community strings in the SNMP Tool If the current node is known to TREND MIBwalker uses the community strings that are config ured for the node in TREND New nodes use the default The default community strings are public for read and private for write You can specify new values to be used in MIBwalker Select the Community Strings option in the Tools menu to bring up the Community Strings window Read MIE Write private OK Cancel Figure 6 25 Community Strings Window The current read and write community strings are displayed in their respective boxes in the window Type in new values for either or both strings Click the OK button to have the new values take effect Click the Cancel button to quit out of the window without changing the community string values MIBwalker reads the community strings for the current node from the database each time you change nodes so the values you enter here are only in effect until you change nodes Changing the community strings f
449. t a target enter a search string and click Search 17 8 DataPipe Components TREND 3 6 User s Guide Using the Protocol Definition Interface to Populate dsi_protocols Lj N E fe 2 N o l Ko 2 Once the Protocol Definition Interface is running you can use it to populate dsi protocols However before you start the definition process it is a good idea to make a note of the different kinds of RMON2 manageable devices you have on your network There are two reasons for this Though there are approximately 5 000 defined RMONG protocols you may not need to use more than a few of them depending on the reports you wish to generate Since the size of the dsi protocols table can affect TREND perfor mance you probably will not want to enter every single protocol available on your network In fact it is recommended that dsi protocols contain only those protocols that appear in your reports Though some networks have hundreds or even thousands of RMON manage able devices most of these devices do not support RMON In addition even if your network has many RMON capable devices it is likely that you have only a few kinds models of devices with RMON agents Since the device manufacturer defines which protocols are collected by the RMON2 agent you only need to define get the protocols once for each device model For example a really large network might include hundreds of Netscout probes all of the
450. t block in the sequence and be gin executing the commands in that block Example A trend proc file contains eight blocks of commands three of which are wait blocks The sequence of the blocks is described below Block 1 six commands wait Block 2 seven commands nowait Block 3 four commands nowait Block 4 eleven commands wait Block 5 eight commands nowait Block 6 five commands nowait Block 7 three commands wait Block 8 twenty seven commands nowait Note Each block must have a unique name Since multiple instances of TRENDproc can run in parallel block names should not be duplicated The names used above are representative of a naming pattern If you How TRENDproc Works 21 3 21 TRENDproc 21 4 intend to create your own trend_proc files it is a good idea to establish a naming convention for your trend_proc blocks This trend_proc file would require four steps to accomplish Step 1 TRENDproc is launched by Ipr_launch based on the appropriate TRENDproc en try in trendtimer sched and executes the first command listed in block 1 TRENDproc cannot execute the second command in block 1 until the first command has been completed Since this block is a wait block TRENDproc must complete the entire sequence of commands in block 1 before beginning the sequence contained in block 2 Step 2 TRENDproc executes the final command in block 1 and begins immediately to ex ecute the commands
451. t may seem that each row from the table created more than one line in the file This just means the lines are wider than your window and have wrapped to fit in the available space Exporting Data TREND 3 6 User s Guide Command Line Options TRENDsheet is invocable from a command line Enter the command aga_sheet h for a description of the available command line options aga_sheet b lt database gt c lt printer copies gt e lt db end time in Sybase format gt f lt spreadsheet gt h i lt print file gt k lt key id from key table gt p lt printer gt s lt db start time in Sybase format gt The Sybase date and time format is yyyy mm dd hh mm ss For example 1997 06 10 00 00 00 J m z 7 2 i rec If you do not specify an extension for the print file file name the output is saved as a postscript ps file You can specify an extension of bmp or gif to save the out put in one of these formats instead Command Line Options 11 21 11 TRENDsheet TREND 11 22 TREND 12 TRENDgraph TREND utilizes time related data that can be viewed in sequence to identify usage patterns TRENDgraph provides easy to use powerful graphing functions that allow you to define display and print reports In creating line cumulative area or stacked bar graphs the user can select data to be displayed create expressions define plot line
452. t not added to a type isql Udsi dpipe Pdsi dpipe 1 gt select dsi target name from dsi node list 2 gt where dsi target name not in select name from dl type 35 go This list identifies The routers or other devices were IP discovered but not type discovered The routers or other devices that were not IP discovered or type discovered In actuality there are two discovery modes IP discovery The dsi_status value on the dsi_nodes_list table for a device is incremented each time the device fails to respond If the de vice responds its dsi_status value is reset to 0 If the dsi_status for a device exceeds either the default of 10 or the value specified with the a option the device is deleted from dsi_nodes dsi_snmp_nodes dsi_nodes_list and di type Type discovery Type discovery is configured to be a scheduled activity each morning if you install a report package via AutoPilot Type discovery uses only dsi_node entries and tries to identify the type that applies to the node If the node is already member of a type being tested it is skipped and no determination of whether it should still belong is made For example consider the situation where the Router_Interfaces type is not dis played in the Define Types dialog because no nodes have been type cast yet Having no nodes matched against the type definition causes the following messages to be written to trend log mw collect Thu Jan 22 16 45 00 1998 Skip type Ro
453. ta Time Period A Axis Y Axis Grid Graph Title Retresh To change the data source node for a report choose the Select Data option This will bring up the Select Data Set window see Figure 13 4 Main Display and Menu System 13 3 13 Grade of Service 13 4 To set the time range of data to be reported click the Time Period option This will bring up the Set Time Period window see Figure 13 8 Set Time Period Window on page 13 15 To change the time scale on the X axis select the X Axis option see Setting X Axis Parameters on page 13 16 To assign labels e g excellent good fair poor to the threshold values assigned in the Select Data window select the Y Axis option see Setting Y Axis Parameters on page 13 18 The Grid command allows you to toggle the grid on or off and alter grid characteristics used for easy viewing of TREND graphs see Adding Grid Lines on page 13 19 The Graph Title allows you to specify meaningful graph titles and subtitles see Specifying Report Titles on page 13 19 The Refresh option allows the user to rerun the query that generated the graph for up to the minute data The graph will clear and redraw itself with the most current data available Edit Aliases Expressions The Aliases Expressions option in the Edit menu allows you to modify definitions setup through TRENDbuild see Assigning Aliases an
454. table but does contain a data source field This group s structure is dealt with in the same manner as the History Host and Matrix tables Polling is performed by node view or type by individual node or by groups of nodes Four tables are created from information from each group s control table and corresponding key table This includes History Host Matrix and Statistics In addi tion a data table is created with values from all probe data tables polled A Matrix group control table for example will be named rmon_matrix and the data table will be rmon_matrix_data Database tables for the supported RMON tables are pre loaded into the database as part of the installation process Polling Frequency The polling frequencies you set for the various tables should take into consideration the size of the table to be returned and the impact this will have on your probe your 16 4 Polling Frequency TREND 3 6 User s Guide network and your database You should also consider how often you are going to look at data and what reporting interval makes sense For example if you are not go ing to look at data in intervals of less than an hour there is no real reason to collect it more frequently Also see the recommended polling intervals in the descriptions of the reports included with TREND Note It is highly recommended that Host and Matrix data be collected ona selective basis with a polling interval of one hour The vast quan
455. te a text file using Notepad or Wordpad and enter the following line SIPACCEPT IP Address of the License Host b Save file in the directory DPIPE_HOME 1lic with name such as dsi_resource N m 032 ag no Ee as oa fo Q z 5 P i o 3 2 Activate licensing with the resource file UNIX Systems a Edit the file etc rc TREND_license_up b After the end of line that starts usr trend license trend elmd m 0 25m add the following phrase with a leading space r usr trend_license dsi_resource c Save the changes d Stop Elan License Manager ps ef grep trend elmd kill xxx where xxxx is the process number shown in the ps listing e Start Elan License Manager sh etc rc TREND license up Windows NT a Open the Control Panel b Double click on Elan License Manager License Key Administration 2 23 2 TREND Configuration for trendadm e f Stop Elan License Manager Add pathname of file created earlier into Resource File prompt Click OK Start Elan License Manager If you have any questions about this implementation whether it should be applied in your enterprise etc please contact DeskTalk Systems Technical Support Message Handling TREND uses logs to keep track of messages regarding the operation of the system Messages appear in the log when processes begin and complete Messages also ap pear when an error condition occurs This section describes the basic for
456. te in US format None begintime No timestamp hh mm 00 00 finishdate No date in US format None finishtime No timestamp hh mm 23 59 frequency No daily weekly monthly None No Any text This is a comment line 2 of 2 The format for an instances line is instances target name tabl key target name tabl key Multiple instances lines can be added for any report see examples instances tmp exception list Input Parameter File TREND 3 6 User s Guide This usage of the keyword points to a file containing a list of target name table key pairs in the form target namel table key 1 target namel table key 2 target name2 table key 1 SEC as Note that the instances file can be generated by specifying a report as an exception report When a report specification block contains the exception y keyword the ASCII output file generated is simply a list in the above described format which is the matching target name table key pairs from the exception criteria This allows for the results from an exception report to be used to drive drilldown reports See example 4 below The style keyword determines whether the rows returned for multiple instances are combined into a single output file written to separate files one for each target name table key If no instances are listed the style keyword is ignored If style individual is specified each output file name will have lt target_name gt lt table
457. te the Object name that appears in Data Manager 4 Click Nexts Creating a MIBlet 6 23 6 MIBwalker The Create MIBlet Assign Property Table window appears Create Miblet Assign Property Table K80_csmacd_errors_ KDempire KDether Figure 6 17 Create MIBlet Assign Property Table Window The elements polled for data in the MIBlet may be defined either by their own property table or an existing property table The Create MIBlet procedure gives you the choice of either creating a new property key table or identifying an existing one for the MIBlet The latter should be used whenever possible since it saves space in the database and allows the properties of elements to be provisioned by the existing table MIBlets that are collected for a common element type must use the same property table If you choose to associate your MIBlet with an existing property table it is very important that you choose the correct table Remember that if you choose any other property table you risk invalidating the data in your new table and any other derived from it Note If you create a MIBlet and are unsure which existing property table is appropriate you should choose to create a new property table This guarantees that your MIBlet does not interfere with other data tables 6 24 Creating a MIBlet TREND 3 6 User s Guide Note The property table you select must be indexed in the same way as the objects in your MIBlet
458. terface utilization for the same week 15 4 Adding New Reports TREND 3 6 User s Guide 2 A Set of Exception Reports for Devices For example a daily report lists Prob lematic WAN Links on the intranet Detailed drilldown reports reveal overuti lized links excessive errors and other components that make the identified links problematic Defining Scheduled Reports Reports selected from the File Selection window are added to the table display A report is scheduled to print daily weekly or monthly depending on whether you se lect Daily Weekly or Monthly in the Scheduled Reports box when you add the new report You can change the Scheduled Reports field by clicking on the down arrow and then clicking on the desired print frequency To delete a report click on the report in the table display click the Delete button and the Save option to implement the change Defining Report Instances Additional instances of a report having different parameters can be created and print ed using Report Scheduler If you click the New button another instance will be add ed From here you can click the Parameters button and change the Report Parameters for that instance see Setting Report Parameters on page 15 6 The original report serves as the template for the new instance E QI D o o 2 wn 7 7 a c mg You can also update change parameters for an instance that already exists Select the instance you
459. ters over the 64 bit field and provide the full value in two pieces In the following example a 64 bit counter called sit Counter is broken into a high order 32 bit field called sit Ct rSplit1High and a low order 32 bit field called sitCtrSplitiLow By collecting the two 32 bit fields and reassembling them TREND is able to store and process the full 64 bit value sitCtrSplitlHigh OBJECT TYPE SYNTAX Counter32 MAX ACCESS read only STATUS current DESCRIPTION The high 32 bits of the 64 bit counter sitCounter sitEntry 7 MIBlets 6 19 6 MIBwalker sitCtrSplitlLow OBJECT TYPE SYNTAX Counter32 MAX ACCESS read only STATUS current DESCRIPTION The high 32 bits of the 64 bit counter sitCounter sitEntry 8 sitCounter OBJECT TYPE SYNTAX Counter64 MAX ACCESS read only STATUS current DESCRIPTION The SectionInTransit units received since the devic booted sitEntry 9 It is critical that the two halves be reassembled in the correct order High Part refers to the high order most significant 32 bit portion of the 64 bit counter Low Part refers to the low order least significant part of the 64 bit counter 64 Bit Counter High Order 32 bit counter Low Order 32 bit counter Creating a MIBlet MIBwalker allows you to select a specific set of MIB managed objects from one or more groups so the se
460. test checks that the device supports where miblet name is the alias name defined in dsi tab alias the miblet specified to belong to the type Example mib II_ifentry Simple oid Oid To pass this test the device must support the oid The value returned by the device is don t care Example 1 3 6 1 2 1 1 1 Simple oid NOT Oid To pass this test the device must NOT support the oid Example 1 3 6 1 2 1 1 1 Running Discover from the Command Line 5 21 a o i o o lt o 2 5 Discover Test Syntax Description Value oid string_value To pass this test the device must support or the given oid and the return value must satisfy the defined expression oid string value y P Example 1 3 6 1 2 1 1 1 cisco router this is regular string compare case oid numerical_value sensitive Or Example 1 3 6 1 2 1 1 1 cisco oid numerical_value this is wild character compare when wild characters are used trend_discover will perform case insensitive compare Example 1 3 6 1 2 1 1 1 bay router this is not equal test Example 1 3 6 1 2 1 1 1 453 or Input File Examples Example 1 this file defines 3com_device type the test is done on sysDescr oid typename 3com_devices collection no 1 3 6 1 2 1 1 1 3com Example 2 this file defines 3com_routers
461. th before sending a fifth attempt If this option is used rmon_collect will wait o seconds after each poll before retrying and will try a total of r times For nodes that are slow to respond increasing the timeout value can increase throughput For example if a node takes 1 5 seconds to re spond with the default timeout scheme no response will be seen until after the fourth poll If the timeout is set to 2 seconds the first poll will succeed 2 seconds is a reasonable timeout to set if you are getting a lot of No response from host messages from devices that can be polled successfully from the MIBwalker SNMP Tool window The number of SNMP variables to include in the varbind list in the GET pdu It is possible to generate a GET request that yields a response too long to transmit The number of variables needed to exceed the maxi mum response packet size depends mainly on whether the request is be ing sent over a WAN link which shortens the allowable packet size compared to a LAN The default p value is 25 For wide area serial lines you may need to use a smaller value If so begin with a value of 15 If that is successful increase it by 1 until you begin to receive error messages that pdu s are too large If 15 is too large decrease it by 1 until the pdu too large messages stop This allows the collection of SNMP data from another port rather then the default SNMP port of 161 The number of times rmon_collect will send a pol
462. the Statistics box in the main TRENDbuild window in the form xpression nam gt expression If you select the expression the word EXPRESSION will appear in the Data Type box in the window Modifying an Existing Report 10 14 An existing report can be modified by loading it into TRENDbuild changing its def inition and saving the changed version to disk This is useful if you want to add or delete statistics in a report change an alias or modify an expression It is also useful as a way to create a series of reports based on common information To load an existing report into TRENDbuild select the Open option under the File menu This will bring up the File Selection window which allows you to specify which report to open Modifying an Existing Report TREND 3 6 User s Guide When the report is open the various TRENDbuild windows will display its defini tion information Modify these as necessary according to the procedures described in Creating a New Report on page 10 5 To save the new version of the report select the Save option under the File menu You can create a new report using the modified definition by selecting the Save As option under the File menu and saving the report with a new name in the File Selec tion window La i E m 2 g ley Qa Modifying an Existing Report 10 15 10 TRENDbuild TREND 10 16 TREND 11 TRENDsheet Tabular data displays are useful for revi
463. the data table and the control table RMON2 Data TREND defines RMON data as data tables that exist in the RMON2 MIB but do not exist in the RMON MIB Supported RMON2 Tables There are ten defined RMON2 tables supported by the RMON2 DataPipe Category Database Name Alias Name Mib Table Protocol statistics rmon2 stats rmon2 stats protocolDistControl Entry Protocol statistics rmon2 statsdata rmon2_stats data protocolDistStatsEntry Network Layer Host rmon2 nlhost rmon2 nlhost hlHostControlEntry 17 4 Source Data Available from RMON2 TREND 3 6 User s Guide Category Database Name Alias Name Mib Table Network Layer Host rmon2 nlhostdata rmon2 nlhost data nlHostEntry Application Layer Host rmon2_alhost_ rmon2 alhost HlHostControlEntry Application Layer Host rmon2_alhostdata_ rmon2 alhost data alHostEntry Network Layer Matrix rmon2 nlmatrix rmon2 nlmatrix hIMatrixControlEntry Network Layer Matrix rmon2 nlmatrixdata rmon2 nlmatrix data nlMatrixSDEntry Application Layer Matrix rmon2_almatrix_ rmon2_almatrix hIMatrixControlEntry Application Layer Matrix rmon2_almatrixdata_ rmon2 almatrix data AlMatrixSDEntry DataPipe Components The RMON2 DataPipe uses SNMP as its transport protocol and the data it collects does not require translation to be inserted into the TREND database The data tables required for RMON2 are installed as part of the DataPipe al
464. the export process is scheduled separately from the data roll up bs h m z o o 27 If you wanted to export the data as soon as it was aggregated you could put both the roll up and export processes in a single TRENDproc file This way as soon as the data had been aggregated TRENDproc would automatically export the data to the proper database Creating and Applying a TRENDproc Creating your own TRENDproc files is a three step process You must 1 Define the commands you want to execute and in what order they should be run 2 Create the TRENDproc file You need to make sure you define your blocks in the correct order using the wait and nowait options Otherwise things may not run as you need them to 3 Schedule TRENDproc to launch The remaining sections in this chapter describe this process Defining a TRENDproc You can design a TRENDproc to accomplish nearly any sequence of TREND pro cesses Simply decide which executable files you wish to call and in which order they should be executed If you are creating a complex TRENDproc file you might want to approach it in a hierarchical fashion Creating and Applying a TRENDproc 21 7 21 TRENDproc Example First Level trendit runs against a specified set of data tables If you are aggregating data from multiple databases you can set one instance of trendit to run against each database with each instance of trendit executed from a differ
465. the maximum number of entries that can be in a Protocol Data Unit pdu The default is 20 P Defines the port number to be used The default is 161 r Defines the number of times Discover will retry its SNMP GET if there is no response t Causes Discover to run an SNMP Type discovery u Defines the TREND user name of the user conducting the discovery V Displays the version number of the product Type Discovery Files When you execute an SNMP Type Discovery trend_discover tries to assign differ ent types to the nodes specified in dsi_nodes Once a node is determined to be of a particular type it is inserted into the dl_type table SNMP Type Discovery is driven by information defined in Type Discovery Files The path of a Type Discovery File can be specified via the F option see notes above By default trend_discover searches for dis files in the SDPIPE HOME scripts directory If dis files are miss ing an error message appears Either install a ReportPack or create a custom report with a corresponding dis file before running Type Discovery again Running Discover from the Command Line 5 19 5 Discover 5 20 Note Once a node is determined to be of particular type trend_discover will not try to rediscover that node for the same type again The valid keywords for Type Discovery Files are the following Keyword Syntax Description typename typename name_of_type Assigns a name to a type
466. the report would query for data from Monday May 19th 1997 00 00 to Sunday May 25th 1997 23 59 Again no matter when the report is actually run the previous week s data is queried for for monthly gt the previous month s data is queried for For example if today was 5 7 97 then the report would query for data from April 1st 1997 00 00 to April 30th 1997 23 59 Example 1 this is a comment line start report apps trendsnmp reports Mib II Daily LAN Interface Inventory qss output tmp daily lan interface inventory headers n format comma dsquery HARDWARE_SYBASE condense n maxlines 2000 style individual end Example 1 would run the TRENDsheet query file Daily_LAN_Interface_Inventory qss and place the output in tmp daily_lan_interface_inventory lt timestamp gt The ASCII file generated would have no column headers and would be in a condensed comma delimited format Lastly the query would be run against the SQL server HARDWARE_SYBASE Input Parameter File TREND 3 6 User s Guide Example 2 this is a comment line start report apps trendsnmp reports Mib I1I Daily LAN Interface Inventory qss output tmp daily lan interface inventory headers n format comma dsquery HARDWARE_SYBASE end start report apps trendsnmp reports Mib II Serial Interface Usage qss output tmp serial interface usageendstart
467. ther long credit card number Each key encodes the number of licenses concurrent users allowed for each group of TREND appli cations also referred to as features or functions as well as the start date and expi ration date of the application a server code and the server s IP address bs m Sz zU oO 3 0 23 0 Q Q 3E O A E Licenses are generally available to anyone on a network who can reach the license server running the license manager daemon The license mechanism does not need to reside on the same server as the TREND application These licenses therefore are known as floating licenses Example If 10 licenses are encoded in the license key any 10 users on the network can use these licenses concurrently When concurrent usage reaches 10 a subsequent re quest for a license will be denied because all licenses are in use Note Desktalk recommends that the number of licenses be one greater than the number of simultaneous users so that someone can always access the system as trendadm Permanent keys are generated based on your sys tem s IP address and hostid a unique identifier for the CPU If either of these changes contact Desktalk for a new keys License Key Administration 2 19 2 TREND Configuration for trendadm License keys in TREND are assigned to application modules based on their func tionality as polling agents rollup aging processes or client applications These groups of appli
468. this entry is as follows trend proc f filename where filename the name of your TRENDproc file Creating and Applying a TRENDproc TREND 22 TRENDdbview TRENDdbview provides an additional mechanism for viewing and comparing data It allows you to combine columns from multiple tables and views into a new virtu al table Using TRENDdbview you can compare and contrast specific data from separate tables either within a single MIB or from completely different MIBs Note The views created by TRENDdbview are database views Database views are unrelated to the topology views described in Chapter 2 TREND Configuration for trendadm This chapter discusses the various aspects of TRENDdbview what it does how it works its components and how to use it Purpose of TRENDdbview TRENDodbview is used to combine individual columns from one or more data tables views and key tables to form a database view Views are virtual data tables and 22 1 N N D m z o ex o 22 TRENDabview TREND treats them as data tables The result is that you can select individual col umns from several different tables and bundle them together to create a new table Once you have created a view using TRENDdbview you can use it to generate the same reports as you would for any data table How TRENDdbview Works TRENDdbview collects the columns of data you request from the data tables you specify and c
469. time scale select the X Axis option under the View button to bring up the X Axis window Select Scale l autoscale 15 Minutes Figure 12 8 X Axis Window The time scale options in the Select Scale box include autoscale Scale graph to the optimal scale automatically so all available data fits in a single window 15 Minutes Each hash mark on the X Axis represents 15 minutes 1 Hour Each hash mark on the X Axis represents 1 hour 1 Day Each hash mark on the X Axis represents 1 day 1 Week Each hash mark on the X Axis represents 1 week Click the OK button to activate the new scale Click on Cancel to retain the original scale Setting Y Axis Parameters The vertical or Y Axis represents the values of the statistics being plotted You can control whether the axis automatically scales itself to exactly fit the actual range of Setting Report Display Parameters 12 17 bs E m 2 O Q o o 7T 12 TRENDgraph 12 18 values displayed or locks in a particular maximum and minimum value and excludes anything that exceeds this range TREND selects auto scaling by default No default axis label is provided To change the Y Axis parameters select the Y Axis option under the View button to bring up the Y Axis window Autoselect maximum value No E Maximum Value fi 00 00 Autoselect minimum value no EE Minimum Value 0 00 Label Pe rcent Use Log Scale
470. tions use port numbers for client server communication Therefore consider the following 1 If you are using a license manager at each server you do not have to be con cerned with the satellite servers and pollers requesting tokens through the firewall However if you are using a central license manager which is inside a firewall you need to know the following information The license manager uses UDP on port 5841 for requests from the remote poller satellite the licensing response to this UDP packet from the remote poller satellite uses a different destination port when it responds to the source port s request Therefore your firewall rules have to allow UDP port 5841 from the remote satellite poller and depending on your firewall Node Lists and Polling Policy D 7 D Setting Up a Satellite Server Environment software allow UDP ports from the license grantor your central TREND server back to the satellite poller 2 The Sybase interface uses TCP IP and is configurable If the database is on UNIX the port number is defined in the interfaces file S YBASE interfaces The default port number is 2052 If the database is on NT you can check the port number via the Dsedit utility the default is 5000 3 The TRENDweb product line uses TCP IP Use the trendweb cfg file to con figure TRENDweb 1 1 Use the RunServer script file to configure TRENDweb 2 0 Use the Configure script to configure TREND web 3 0 and above 4
471. tistics Click Assign Expressions box in the TRENDbuild window to invoke the Edit Ex pressions window The Edit Expressions window provides Selection of statistics that will be included in expressions to be displayed upper left Mathematical operators to be used in expressions beneath Select Statistics box Creation of expressions from statistics and operators bottom Naming and adding deleting expressions and Creating a New Tabular Report TREND 3 6 User s Guide Listing of Active Expressions assigned upper right Edit Expressions delta_time gt Delta Time Figure 11 3 Edit Expressions Window EXAMPLE Creating an RMON Host Table Error Report To define a tabular report displaying packets transmitted transmission errors and error transmission rates open the Edit Expressions window and click in the New Expressions box and enter a name for the first expression to be created e g Pack ets Click Add to place the expression in the Active Expressions box Click the new expressions name in the Active Expressions box to continue make sure it is highlighted ok J m z 7 7 s Select hostOutPkts from the Statistics box It will appear in the Edit Expressions window Click Add to View to save the Packets statistic Repeat the above steps for the expression Errors using the hostOutErrors statistics In the same manner define Error Rate expression and assign its expression
472. tity of data returned due to the size of these tables has an impact on network utilization and database performance The amount of database space used will also increase dramatically when collecting Host and Matrix table data History Group Polling The setting of polling intervals for the History group requires special consideration The RMON history structure provides a major benefit to managers who want to re port on relatively fine grain network activity data Normally obtaining one minute data samples would require polling once every minute In a large network where many devices are to be polled such frequent polling not only utilizes measurable management station and network bandwidth but the skew between samples i e imprecision of the one minute interval will be large Fortunately RMON History group data consists of a set of samples maintained by the RMON agent Hence poll ing on a leisurely basis can still yield fine grain data Historys differ in the number of samples buckets and intervals bucket width If we assume a short History fifty 30 second buckets and poll every 20 minutes the result is 50 data samples each corresponding to a different 30 second period and each containing delta TREND rate table values for each statistic Raw data for the History group is not inserted into the database rather it is already rolled up in the form of rate data Data is therefore directly inserted into the rate table with timestamp
473. to Group button to register the user The Sybase Interfaces File 2 4 Sybase uses a file called interfaces to know where to find the SQL Servers on the network On UNIX systems this file is stored in the Sybase home directory named by your SYBASE environment variable e g usr Sybase Here is an example of an interfaces file PATUXENT SYBASE query tcp null string patuxent 2025 master tcp null string patuxent 2025 console tcp null string patuxent 2026 AMUR SYBASE query tcp null string amur 2025 master tcp null string amur 2025 console tcp null string amur 2026 JORDAN SYBASE query tcp null string jordan 2025 master tcp null string jordan 2025 console tcp null string jordan 2026 The SQL servers named in this file are Defining Users and Groups TREND 3 6 User s Guide PATUXENT SYBASE AMUR_SYBASE JORDAN_SYBASE The Sybase interfaces file on Windows NT systems is named Sybase ini sq ini Here is an excerpt from this file SUWANEE master NLMSNMP pipe sybase query query NLMSNMP pipe sybase query master NLWNSCK suwanee 5000 query NLWNSCK suwanee 5000 banyas master NLWNSCK banyas 2052 query NLWNSCK banyas 2052 N E m Sz Ag e eie o Nio it Q Q P 9 3 The SQL servers in this file are SUWANEE and banyas The servers listed in the Sybase interfaces file can be entered on the Topology Data base and Data Database lines in the Define Users and Groups
474. to collect a specific table once from a single node and store the informa tion in the TREND database Fully utilized mw_collect follows user defined A 22 mw collect instructions to poll a dynamic list of nodes anywhere on the network for data at reg ular intervals and store the results on a local or remote system depending on who re quests the information The mw collect command invokes the mw collect parent poller The mw collect parent poller in turn invokes multiple children pollers Each child poller collects a single node table combination The c option specifies the number of child pollers that run concurrently When the child poller finishes the data that is collected is appended to the latest ex isting holding file for loading into the TREND database The latest holding file is loaded into the database when the file is at least 500 000 bytes or every child poller has finished See Directory Structure on page A 28 for a discussion of the direc tories in which these holding files are stored before they are loaded into the TREND database Note Do not use mw collect to poll for tables in the RMON and token ring RMON MIBs Do not use mw collect to poll for interval tables such as the dsx1IntervalTable defined in RFC 1406 mw collect calls the following procedures for each table and destination database pair Thebocp gateway procedure updates property and data tables in the TREND database The TRENDpm
475. to see as columns in the table The power of TRENDstep is its ability to rotate data from multiple rows of an existing data table into a single row in a new data table while simultaneously aggregating the data into new and useful forms Among other things this allows you to compare one segment of data with another in the same report for example the utilization of Ethernet router interfaces with the utilization of FDDI interfaces on the same device or across devices A 53 TRENDstep Example A good way to illustrate TRENDstep is to see how it converts an RMON data table Figure A 1 shows part of an RMON data table with many rows of data related to network traffic The columns in the table are Time Period Target Name device identifier Table Key Delta Time and Octets Each row of the table represents the octet count for a specific communications protocol as recorded by an RMON agent on device 134 70 18 241 which is identified in the Target Name column Note that the Time Period is the same 1 40 p m for each row The figure shows all of the tracked RMON protocols for a single polling period ta period which is NOV 15 1997 1 40 00 PM For the purposes of this example we focus on the data for this polling period Time Period Target Name Table Key Delta Time protocolDistStatsOctets Nov 15 1997 1 40 00 PM 134 70 18 241 _wildcard ether2 arp 1 426 61 064 Nov 15 1997
476. traints dsi key id I b sg 0 and ifType in 16 17 18 19 20 21 22 23 3 _Add Update _Delete Delete An dsi key id gt 0 and ifType in 16 17 18 19 20 21 22 23 30 31 32 ok J m z 7 7 d O zd Figure 11 7 Constraints Window Specifying Data to be Included in the Report 11 13 11 TRENDsheet Formatting the Report The look of tabular reports can be modified by sorting data and reformatting column parameters Sorting the Data in the Report 11 14 TREND data is time related therefore TRENDsheet s default sort order is by target name table keys for a particular target name and received timestamp i e newest to oldest for a particular table key in that order Since most reports will be run for a particular device and a specific key for that device the timestamp becomes the sig nificant sort field There may be reports that you would rather sort on a different field or in ascending rather than descending order This is accomplished through the Order window which is displayed by clicking on the Sort Order option under the View button Description Speed Admin Status Oper Status Avg In 6Util order by Date desc Router Port Clear LOK Reset Cancel Figure 11 8 Order Window Formatting the Report TREND 3 6 User s Guide To change the sort order first delete the existing sort order by clicking on the Clear button
477. tton is highlighted Note that the Set Table button is highlighted any time a table containing writable variables is polled in SNMP Tool 4 Click the Set Table button This brings up the Set Tool window Set Tool Jol x SelectKey gt Add NewKey Set Table Cancel o DJ S D ifAdminStatus 1 Figure 6 31 Set Tool Window The SNMP Tool window will be frozen while the Set Table window is up The window lists all the writable variables in the current table their data types and current values 5 Select the instance ID for the instance whose variable s you want to set by clicking on the down arrow of the Select Key box then selecting the desired value The variables and current values for the selected instance are displayed By default the first instance and its associated field values are shown Utilities 6 35 6 MIBwalker 6 Enter new values in the value column for the fields you want to set Click in another field or press the Return key to enter the value You can change keys and set values on fields with different keys as part of this same step 7 When all fields are entered the way you want click the Set Table button to apply the new values to the target node If the set is successful you will see a box with the following message on your screen set successful 8 Click the OK button in the box to continue 9 Get the table again to verify the values are correct 10 Click the C
478. ture Result Result Process Failure Point Standard Cache Version Version Parent Failed to connect at Use cached information for Exit Poller startup collection Store data locally Parent During initializa Use cached information in Exit Poller tion lost connection most cases Store data locally Parent Determined during Store data locally Exit Poller initialization that database is full or transaction log is full mw collect Process Failure Point Result Cache Version Result Standard Version BCP gate way Failed to connect to a given database Returns code to calling par ent thread Parent will not attempt to load data into this database during this cycle Data is stored locally Same as Cache version BCP_gate way Lost connection to database database is full or transaction log is full Removes successfully loaded data from input file by default deletes the entire file when operating in ASCII mode a Returns code to calling par ent thread see above Depending on when the database connection was lost or the database became full during the bcp gateway process the removal of suc cessfully loaded data may not be accurate This may cause insertion of duplicate data into raw data which will fail if there are indexes and generate error messages in trend log bcp gateway ignores this error message and continues loading new
479. u System 8 3 8 Data Manager by the value set for Keep lt type gt Data in the Default Storage Time window where lt type gt matches the Type column value for this table See Changing System wide Rollup and Aging Defaults on page 2 14 This field is blank for tables which are not aged e g internal tables Only the user trendadm can change this value Command Menu Option Summary 8 4 The menus available from the menu bar of the main Data Manager window are de scribed below File I View Tools Truncate tablefs Delete table s Set Default Storage Time Change Database Exit About The Truncate table s command removes the contents of a table without deleting the table Truncation does not write the removed rows to the transaction log so it cannot be recovered Only the user trendadm can use this function see Truncating Tables on page 2 16 Delete table s drops the table and its contents from the database The deleted table is written to the transaction log If the table is too large to be written to the transac tion log it will need space twice its size in transaction log for the delete to succeed use the Truncate option to empty the table before deleting it Main Display and Menu System TREND 3 6 User s Guide Only the user trendadm can use this function see Deleting Tables on page 2 17 Set Default Storage Time brings up a window that shows the values that wil
480. u ieres rka kria ERTEKE ESERE Inh 7 3 The Tools Menu ihe aaa ad GA bet peek EA 7 4 Setting Up New Polling Requests llle 7 4 Specifying What Data to Collect 222 00 005 7 5 Specifying Which Nodes to Poll 00 000 cee eee eee 7 6 Specifying How Often to Poll for the Data 7 7 Specifying Which System Should Poll for the Data 7 8 Collections Involving Multiple Database TREND Installations 7 8 Saving the Collection Request Definition 7 9 Changing Databases cece ee 7 9 Managing Polling Requests llle 7 10 Which Request Definitions Are Displayed 7 11 Reordering the List of Collection Requests 7 11 Modifying Request Definitions aaua nauan 7 12 Deleting Request Definitions llle 7 13 Exiting the Collect Data Window l l 7 13 vii Contents viii Some Considerations When Organizing Polling 7 13 PollingXGroups i4 2 xvi ndelseu thet Mees RbURS PD REESE 7 13 Polling Fredgueticy sess e be ER eons Pewee IS 7 15 Viewing SNMP Data ills 7 16 Command Line Options ra Ea cee eee 7 17 Incomplete Data Collection or Aggregation 7 17 Data Manager coui corus a cede celer le IA Rea eee es 8 1 Main Display and Menu System 0 20 es 8 1 Database Information Sik oaa ER eee
481. ude with the new type definition As you select a node it will become highlighted hold down the shift key to select additional nodes When you have selected all the desired nodes click on the Add Type button The name you entered will be included in the Current Types list Note Types can belong to different groups The list of existing types is displayed in alphabetical order in the Current Types scrolling list To select an existing type click on the downarrow of the Current Types box to bring up the list then select the name of the type you want To delete a type select its name in the Current Types list and click on the Delete Type button To edit a type you must delete the old type and create a new one using the methods described above Click on the Cancel button to quit out of the window Note When adding another node you must first delete the nodes in the Nodes of this Type box before adding new ones 2 10 Defining Types TREND 3 6 User s Guide Changing the Default Report Directory The default report directory is where TREND automatically looks for report defini tion files when the user selects the Open option under the File menu The default re port directory is SDPIPE HOME reports Use the Specify Report Directory window to change the default path Specify Report Directory EN xi c TRENDS 1 reports LOK Cance Figure 2 6 The Specify Report Directory Window N E m 03
482. ue to the additional overhead it places on mw_collect This option is in UPPERCASE This option is equivalent to the debug_level parameter in TRENDpm Names the host to be polled which must be an SNMP manageable device This name will be saved in the database along with the data returned from the node so in order to be consistent with other data use the same node name format used when entering nodes in the database A 17 gt D v o Q o o mw_collect i I k K Is the Collection ID which corresponds to the collection interval in minutes mw collect executes the entries in the mw collection table that have this value in their interval field How frequently mw collect is actually run depends on the configuration of TRENDtimer but the idea is to be consistent so that a collection request with a collection ID of 5 is run every 5 minutes See File Locks on page A 23 for addi tional information This option is required Specifies whether to use existing indices on the upload table or to drop existing indices and then recreate them The value 1 means that the existing indices on the upload table are used The value 0 means that the existing indices are dropped and then recreated The default is 0 This option is in UPPERCASE This option is equivalent to the 9 bCheck index parameter in TRENDpm Populates the property tables but not the data tables for the devices you are polling Use this
483. umber of minutes The default is 30 minutes This option is in UPPERCASE Sets the debug output level for the parent instance of mw_collect Val ues of 0 1 2 or 3 are valid The higher the number the more detailed the information The default is 0 which means no debug output Debug output is written to standard out You should only use this option for testing in coordination with DeskTalk Systems Technical Support due to the additional overhead it places on mw_collect The z option sets the debug level for the child instances of mw_collect The d and z options can be used together to control the debug level of parent and child mw_collect processes independently This option specifies how many seconds the parent thread should wait for signals from the collector and bcp gateway threads It is normal to get thread wait timeout messages in the log file when waiting for signals from the bcp gateway thread due to potential long running jobs when writing to the database The program exits when a thread timeout occurs unless the thread timeout occurred when all col lector jobs are finished and only bcp gateway threads remained In this case the timeout only appears in the trend log file The default is the same value as the C timeout value which is 1 800 seconds For example if the C value is 2 which is 2 minutes then the D value defaults to 120 which 120 seconds This option is in UPPERCASE Sets the percentage level fo
484. umn name SQL column name object ID and ob ject type Column Info 0100 Isystem perf rf x Fur Couma Name SoLCeumnName onjeto Object Type farget name Jametname STRING d receved_ts Jetewedis UNXTME equest requestiso03 UNXTME eta me e menos J NTEGER username usernames 5TRING table key table_key006 STRING report_type report type007 3 STRING eceied usec received usecO08 GAUGE sD Figure 8 2 An Example of the Column Info Window 8 6 Obtaining Table Profile Information TREND 3 6 User s Guide Showing Particular Tables By default Data Manager shows information about all data tables and most internal tables If you only want to see a particular group of tables you can specify a group either by source or by type To select a class of table do the following 1 Click the View button in the Menu Bar 2 Select the category by which you want to subset the tables Your choices are Show by Source Show by Type When you select a category a popup window will show you the choices in that category If you selected Show by Source the popup looks like this Show by Source Figure 8 3 Show by Source Window Oo O D o 3 pi Q D C Showing Particular Tables 8 7 8 Data Manager If you selected Show by Type the pop up looks like this Show by Type Ea Weekly Tables
485. unt of samples between the min and max values i e count of samples gt min and lt max tet value Count of samples above value i e count of samples gt value Display the syntax for this utility Use 3 4 x compatibility That is apply 3 4 x TRENDsum summarization algorithms See Changes in TRENDsum 3 5 1 from 3 4 x on page A 86 Percentage of samples that can have a null value Default is 50 For example assume n 10 is specified for this parameter If there are 100 samples and 11 are null due to collection agent problems faulty collections or the device being unavailable at the time the sample was taken the value for the statistic is null Replace an existing destination table with the same name Name of the table that is providing the input to TRENDsum Display the version of this utility First day of the week specified as Mon Tue Wed Thu Fri Sat or Sun Default is Mon This parameter is not case sensitive Number of days in the baseline period A 79 gt D v o Q o o TRENDsum Description A 80 TRENDsum Formulas Formulas for computing the statistics are tot avg wav std med p90 p95 p98 xbar X val where val is the resolved value of the expression to which the tot function is being applied tot cnt where cnt is the number of samples being summarized X val At At sqrt cnt X val val tot tot cnt cnt 1
486. up in the configuration output listing Group Memory Use Parameter Name Default Memory Used Config Value Run Value additional network memory 0 0 0 0 audit queue size 100 42 100 100 default network packet size 512 290 512 512 number of user connections 25 10223 128 128 1 To change the number of user connections isql Usa P sp_configure number of user connections 100 go 2 Restart Sybase to enable the new user connection allocation Choose the number of user connections carefully Under allocation results in tasks being unable to connect to Sybase Too many user connections waste memory that could be allocated to the data cache Sybase Configuration Parameters C 9 9 O 2h e 3 1 D el 3 Ce C Performance Tuning Issues We recommend 100 user connections This is sufficient in most cases In cases where RAM is constrained a more exact calculation may result in additional memory savings To calculate the number of user connections add up the following Total number of baby SNMP collectors mw collect c option If there is no c the default is 5 Total number of baby RMON collectors rmon collect c option If there is no c the default is 5 Total number of concurrent TRENDit processes desired Total number of concurrent TRENDcopy processes desired Total number of concurrent client processes desired 9 9 9 Add 5 to cover miscellaneous Sybase processe
487. up period set to the start of the time period rounded to the nearest hour and includes a count of the number of sam ples ta samples collected during the period A 39 gt D v o Q o o TRENDit Options TRENDit has the following options a b C d f h M To only rollup hourly data that is specified Values 0 23 are valid If specifying a 1 13 the rollup process will only generate hourly data from one o clock in the morning to one o clock in the afternoon This only affects the hourly table Overrides the automatic rejection of data when the delta time is greater than 25 hours Causes TRENDit to filter out data that is below the specified value for the specified column TRENDit applies this test after it has computed the delta values for the column Rows that pass the test are deleted from the rate output table no rows are deleted from the raw input table Set a debug output level Values of 0 1 2 or 3 are valid The higher the number the more detailed the information The default is 0 which means no debug output Debug output is written to standard out The number of seconds after the hour a data collection record can be timestamped and still be considered part of the previous hour For ex ample if a group of nodes is to be polled every 20 minutes data reflect ing what happened from 12 00 to 1 00 will be collected at 12 20 12 40 and 1 00 However some of the 1 00 sample
488. user name is joebloggs the command is trendcopy s PRIMARY SYBASE U joebloggs S BACKUP_SYBASE Using the r and R Options To Copy Rolled Up Data m 2 g t So lt TRENDcopy copies all non system tables present in dsi tab alias when you omit the t option Use the R and or r option to select only certain types of these tables for copying Use the R option to select only raw r rate R hourly H daily D weekly W monthly M quarterly Q and or yearly Y tables from the TRENDsum family of rollup tables Except for raw and rate tables the table names begin with S and the indicated letter You can combine these values to select more than one type for ex ample use R DW to select all SD and SW tables The value A indicates all these types rtRHDWMQY Use the r option to select only raw r rate R hourly H daily D weekly W and or monthly M tables from the TRENDit family of rollup tables Except for raw tables the table names begin with the indicated letter You can combine these values to select more than one type for example use r DM to select all D and M tables The value A indicates all these types rRHDWM You may use either the r or R option to select raw and or rate tables In addition you may use both options at the same time to increase the selection For example r HD R HD selects all hourly and daily tables Usage Notes 18 9 18 TRENDcopy When
489. using the Column Format window Bring up the window by se lecting the Column Format option under the View button Formatting the Report TREND 3 6 User s Guide Column Format Figure 11 9 Column Format Window Select the column whose precision you want to change by clicking on its name in the Column Title box Click the number of Decimal Places you want the column to dis play in the report Values are rounded to the precision you choose If you select a non numeric column the values in the Decimal Places box will be grayed out Click the Apply button to the set the new precisions on the columns modified Click the Reset button to set all values back to their defaults ok J m z 7 2 i rec Click the Cancel button to quit out of the window Column Width To make a column wider or narrower e g after changing the precision of a numeric column click the right edge column divider of the cell header and hold the mouse key down The column divider will become a broken line While still holding the mouse button down drag the broken line to the left or right to make the column nar Formatting the Report 11 17 11 TRENDsheet rower or wider respectively When the column is the desired width let go of the mouse button and the column will resize itself to the new width Changing the width of one column will not affect the width of the other columns Exporting Data 11 18 You c
490. uter_Interfaces not found in DL_TYPES amp DSI_NODES TREND 3 6 User s Guide The discover did not work because the value returned via the sysObjectID did not match the value in the dis files in the DPIPE_HOME scripts directory which are designed to find major brand vendor routers The sysObjectID is probably different for the devices being referenced DeskTalk Systems has placed additional files on its FTP server in ftp pub TREND Discover_Files Currently the files are for Router_Interfaces and Baseline As cus tomers and Desktalk Systems engineers find other devices that are not being discov ered we will place the needed files first on the ftp server and then in TREND For example the Router_Interfaces additions will match the OID string for Well fleet 1 3 6 1 4 1 18 and cause the devices matching this string to be placed in the Router_Interfaces type If this additional discovery file does not solve the issue you can use the MIBwalker identify to help identify the setting necessary for recognizing the devices 1 Bring up MIBwalker Click on the bold system branch to highlight it Select Change Node from the Tools menu Choose an address of the device and click OK Select SNMP Tool from the Tools menu and click Get In the data window look for the line for sysObjectID mi ON MG GB A NG Copy one of the existing discovery files for example the 3com dis file in the SDPIPE HOME1scripts directory to a ne
491. utes rmon token orinoco rmon tr rg 15 Minutes rmon i rmon2 probe1 trendadm 20 Minutes ecam statis rmon2 probe1 rmon all trendadm 1 Day ecam hlHos 4 gt Figure 7 4 Collect Data Window The table in the window lists all pertinent information about each request definition as setup in the Collect Data Periodically window The following columns are in the table Node View Type The combination of values in the first three columns is determined by the polling mode If an individual node is being polled the Node and View columns contain values the Type value is not used here If the request is polling by Type only the Type column contains a value and so on User The name of the user who defined the request 7 10 Managing Polling Requests TREND 3 6 User s Guide Frequency How often TREND will poll for the requested data Mib The Category or SNMP MIB that contains the table being collected Group The Group or MIB table being collected Poll From The name of the system that actually polls for the data Which Request Definitions Are Displayed The list of requests displayed in the window is controlled by the User box Only re quests with a User column value matching the value in the User box are displayed By default the User value is the name of the user running TREND To change the value in the box click on the down arrow next to the User box and select the desired value from the
492. ven in the section on the TRENDsum utility in Man Pages on page A 1 Baselines 9 20 A baseline provides a basis for comparing recent performance with past perfor mance for example for comparing last week s performance with performance for the last two months You use TRENDsum to create a baseline TRENDsum enables you to compute a baseline for any defined element property You can also baseline a computed metric such as utilization which is computed as an expression that refers to one or more element properties TRENDsum enables you to choose how long you want the baseline period to be the TRENDsum y option The default is 42 days six weeks TRENDsum creates a rolling baseline table To perform rolling baseline aggregation use the y option on the command line to specify the length of the baseline period in days In a rolling baseline the baseline table contains n rows each row represents the same statistic or statistics computed for the immediately preceding number of days in the baseline period Here is how a rolling daily baseline is computed if you use a 42 day baseline period 1 At the end of day 1 the baseline statistic is computed from all the sample data collected for day 1 and stored in a summary row for day 1 For example assume you are computing an average value to be used as a baseline If 96 sam Baselines TREND 3 6 User s Guide Baselines ples are collected for day 1 one sample ever
493. view A view does not require either qualifications or expressions However if you want your view to run properly without them you must define their absence Trying to save or run a view without defining its qualifications will generate a warning message Select the columns that will make up the view Since running a view generates a virtual data table the columns that you choose will be the columns that appear in your view Choose your col umns just as you would when creating a data table Choose a name for the view This is done in the TRENDdbview main window and the name you choose is important especially when you run the view This is because of the naming protocol for saving and running views Opening closing and running the view as well as using it as a data table in another view all depend on the name you give your view Working with TRENDabview TREND 3 6 User s Guide Save the view as a file Your view is saved as an executable file When you run this file you gen erate a virtual table in your database For this reason the name and path of the file are significant Each of the above tasks can be accomplished by any TREND user However before the view can be used to generate reports it must be created in the TREND database This is done by running the view Only the trendadm can run the view For this reason views are typically defined by the trendadm Create a New View You can start a
494. w The existing line definitions can be modified or deleted Note The etherHistoryEntry field etherHistory Utilization is recorded in 1 100ths of a percent If utilization is 37 25 it is displayed as 3725 in TRENDgraph Selecting Statistics The Select Statistic box contains a list of statistics chosen and expressions created in Report Scheduler when the report was created Select an expression using the Se lect Statistic box s down arrow button then clicking on the desired statistic Use the scroll bar to see all the available names The Select Key box contains a list of all keys for the selected device in the database table being reported on by the graph Click the Device box down arrow to view a list mah N J m 2 O Q o o T Creating a New Graph 12 9 12 TRENDgraph 12 10 of devices and select a device for reporting Select an aggregation type using the Ag gregation Type pull down arrow to view and select it P2 Select Key olx Device Aggregation Type Cancel os Sort Order Figure 12 5 Select Key Window Select a key by scrolling down and clicking on the appropriately numbered row Use the scroll bar to see all the available keys Note that this mechanism allows you to look at the same data from two different devices or for two keys on the same device simultaneously TRENDgraph aligns all data by time so it is easy to compare con current values from two sources in a sing
495. w name such as new device dis into the SDPIPE HOME1scripts directory 8 Edit the new file to remove the snippet in quotes on the right side of the equal sign and replace it with the sysObjectID result from the MIBwalker Value column shown after the enterprises snippet For example new device dis was 143 6 104 1443 1 4 MIBwalker shows enterprises 141 1 1 6050 new_device dis becomes 1 3 6 1 4 1 141 1 1 6050 E 3 9 3 en Q i mg o E 7 m o 7 1 lt i 3 lt E Discovery Considerations This affects only the Router Interfaces package but it provides a starting point to un derstanding how type discovery is processing so that you can make necessary revi sions as you identify new devices for TREND to monitor E 4 TREND F Relocating the TREND Database If you need to move the TREND database from one host to another DeskTalk Sys tems recommends creating and using backup Sybase servers on both host A the source host and host B the target host Follow these steps 1 Install a backup Sybase server on host A Make sure both the Sybase server and the backup Sybase server are running on host A See Backup Server on page B 15 for detailed instructions 2 Install TREND on host B The database size on host B must be equal to or greater than the one on host A for these steps to be completed CAUTION 3 On host B turn off trendtimer so there is no automatic databa
496. w points at your system s default report directory If a report directory has been established through the Configuration menu then this will be the default directory This path is set as shown in Changing the Default Report Directory on page 2 11 If the report is in a different directory you can get there by one of two ways 1 Type the path name of the directory in the Filter box then hit the Filter button 2 Use the Directories box as a navigation aid You can change directories by double clicking on an entry in this box or clicking on an entry then clicking on the Filter button The box has three types of entries The current directory which is displayed in the Fil ter box If you select this you stay in the current directory The directory above the current directory If you select this you will move up one level in the path name hierarchy For example if you are in usr TREND reports Routers and select you will move to usr TREND reports subdirectory name These are the names of directories below the cur rent directory If you select one of these you will move down to the directory selected For example if you are in a directory called usr TREND re ports and you select a value Routers from the Directories box you will move to usr TREND reports Routers The Select File window shows you the names of all files in the current directory that match the value in the window s Filter box You
497. will make up the view However we recommend that you choose which columns to include Unless otherwise specified TRENDdbview creates a view that contains only the de fault columns headers and footers of the first data table you select If you want to see data contained in other columns you need use this dialog box Define Expressions Window The Define Expressions window allows you to determine which if any expressions to apply to your view Define Expressions N N D m 2 O ex o cisco100 View Misha0797 1 a cisco100 Isystem perf b Kcam hlhostdata A K100 Isystem perf B Figure 22 4 Define Expressions Window It is not necessary for you to define any expressions and there are no preset default expressions Note Since TRENDdbview is typically used to view and report on specific data objects rather than on general information we recommend that Elements of TRENDdbview 22 7 22 TRENDabview you apply whatever expressions you consider useful By doing so you can achieve better results The procedure for defining expressions in a view is described later in this chapter Define Qualifications Window You access the Define Qualifications window through the TRENDdbview main window In it you define what qualifications you wish to apply to the view Define Qualifications Selected Data Tables Key Tables Qualification Text cisco100 View Misha07971 a Cl cisco100 Isystem perf b Where a dsi ke
498. window see Figure 2 2 The Define Users and Groups Window on page 2 3 Entries are case sensitive so make sure the value you enter exactly matches the entry in the Sybase interfaces file Defining Nodes You can add nodes to the database through various methods One method allows you to automatically search your subnets for the nodes using the Discover option from the TREND Main window See Discover on page 5 1 Another method allows you to import a list of nodes from a text file See Import Nodes on page 3 4 A third method allows you to add nodes one at a time through the Define Nodes window To add a node bring up the window by selecting the Define Nodes option under the Configuration menu Defining Nodes 2 5 2 TREND Configuration for trendadm 2 6 Define Nodes xi Current Nodes 134 70 18 255 New Node Read Community String public Write Community String private Node Type snmp E Add Node Delete Node Cancel Figure 2 3 The Define Nodes Window Click on the downarrow of the Current Nodes box to see a list of the nodes already entered To see how one of these nodes is configured select its name from the scrolling list of nodes in the Current Nodes box Its values will be dis played in the various fields in the window which are explained below To add a new node to the database enter its name in the New Node box The name must be able to be resolved to an IP address throu
499. with the characters PR Delta rank Delta rank of the metric absolute value of the change in rank between the current rank position and the previous rank position for the element Column name is prefixed with the characters DR Previous delta rank Previous delta rank Column name is prefixed with the characters PDR The ranking is typically based on daily level summarized data although weekly and monthly summaries can also be used as input Definition File gt D v o Q o o TRENDrank needs a definition file to identify the source summary table the target rank output table and the columns from the source table that are to be ranked This definition file has four types of statements the first three are required A 49 TRENDrank A 50 source_table source_summary_table_name destination table target rank table name i column target_table_column source_table_column s sub_variable source_table_column where s has one of the following values a Rank values in the column in ascending sequence d Rank values in the column in descending sequence default The source and target statements from the 3 5 x and earlier versions are still available Please note that they may be discontinued in future releases Note The table name in the source table or destination table statement is the alias group or generic name not the name of the item as known to the database for example mib II ifEntry for the
500. x Selecting the Database TREND 3 6 User s Guide Note Selecting a database from the Change Database window only identifies the database to be used for reporting with Report Launcher It does not change the value of the DSQUERY environment variable Selecting the Report File Directory TREND installed and user report definitions are stored in one or more report directories The default directory for TREND installed report definitions is SDPIPE HOME reports You can identify the directories you want to work with and make it easy to browse their contents by designating a Top Level Report Directory When you do this that directory becomes the initial current directory If you select another directory to be the current directory you can easily make the Top Level Directory the current direc tory again by double clicking on the Top Level Report Directory entry in the Folders list box B D pap ag o i 3 7 gt Dm The name of the Top Level Report Directory stays the same and survives the termi nation of the Report Launcher session until you change it as described below Remaining entries in the Folders list box include Names of the first level subdirectories of the directory that you have desig nated as the Top Level Report Directory You can double click on a subdirec tory name to make it the current directory The name will appear in the Look in edit box You can also type the d
501. x_graph h for a description of TRENDgraph command options agx_graph b database e printer copies gt e lt db end time in Sybase format gt f graph file h i lt print file gt k lt keyid1 keyid2 keyid3 gt p lt printer gt s lt db start time in Sybase format gt The Sybase date and time format is yyyy mm dd hh mm ss For example 1997 06 10 00 00 00 If you do not specify an extension for the lt print file gt file name the output is saved as a postscript ps file You can specify an extension of bmp or gif to save the out put in one of these formats instead 12 22 Command Line Options 2 o o Q D e h wn D 7 o TREND 13 Grade of Service Grade of Service GOS is a tool for system health reporting and Service Level Agreements SLAs GOS converts sets of individual historical performance statis tics into meaningful graphical displays using a weighted stacked bar format The user can select any set of SNMP or RMON metrics and assign a grading scale to each based on the relative importance of that metric A grading scale e g excel lent good fair and poor allows the user to concentrate troubleshooting efforts im mediately on areas where performance is substandard Grade of Service reports alleviate data overload that can result from massive amounts of network data available from
502. y MAC History Token Ring rmone_hist_ rmone_history HistoryControlEntry and Promiscuous historyControl2Entry History Token Ring rmone_histtrpdata_ rmone_tokenring promiscuous history TokenRingPHistoryEntry Promiscuous data History Host rmone_host_ rmone_host HostControlEntry and hostControl2Entry Host rmone_hostdata_ rmone_host data HostEntry Matrix rmone matrix rmone matrix MatrixControlEtnry and matrixControl2Entry Matrix rmone_matrixdata_ rmone_ matrix data MatrixSDEntry Token Ring rmone_trstation_ rmone_token ring station entry RingStationControlEntry and Stations ringStationControl2Entry 2 of 3 Source Data Available from RMON2 17 3 Lj N E O 2 N o I Ko o 3 17 RMON Support Category Database Name Alias Name MIB Table Token Ring rmone trstationdata rmone token ring station entry data RingStationEntry Stations Token Ring rmone_trroutestats_ rmone_token ring routing stats entry Control information from Source SourceRoutingStatsEntry Routing Token Ring rmone_trroutestatsdata_ rmone_token ring routing stats entry data Statistics from Source sourceRoutingStatsEntry and Routing sourceRoutingStats2Entry 3 of 3 Note that in the above table information from some of the table extensions is stored in control tables If for example you want to include Dropped Frames as part of reg ular RMON data report you need to create a view between
503. y 15 minutes the values for all samples are totaled and divided by 96 to compute the average for the day 1 Day Baseline Statistic 1 aj At the end of day 2 the baseline statistic is recomputed from all the sample data collected for days 1 and 2 and stored in the summary row for day 1 thus replacing the existing day 1 value For example if 96 samples are collected for day 2 those 96 values and the 96 values collected on the previous day day 1 are totaled and divided by the sample count now 192 96 for day 1 plus 96 for day 2 to arrive at value ap Day Baseline Statistic 1 a In the same fashion the baseline statistic is recomputed at the end of day 3 from all the sample data collected for days 1 2 and 3 and stored in the sum mary row for day 1 again replacing the existing day 1 value At this point the value a3 for the baseline summary row represents a 3 day aggregate Day Baseline Statistic 1 a3 The same processing is repeated at the end of each day until a baseline value has been recalculated for each day of the rolling baseline period 42 days in this example Thus at the end of 42 days a single baseline value has been computed based on data collected on the previous 42 days 9 21 v 5 2 D O E D en 5 uonebaibby eleg 6 9 Data Aggregation and Manipulation Day Baseline Statistic 1 a42 This day 1 bas
504. y Group Polling llis 16 5 Aggregating Host and Matrix Tables and Managing Size 16 7 TREND RMON Reports 0 00 eee eh 16 8 Viewing RMON Tables in Data Manager llle 16 8 DOMAINS c TT mAh maa Nah naa aka ah Seana Obed Adena eta 16 9 RMON2 Support ac ee Ha ee ie SEO Se ie ee ERE 17 1 Source Data Available from RMON2 a 17 2 RMON Data uc ee ed eek ee ev bee ee eho a ERE 17 2 Supported RMON Table Extensions 0 17 2 RMON 2 Data i oeaiei EE en 17 4 Supported RMON2 Tables anaana 000 eee eee 17 4 DataPipe Components 0 000 e eee eae 17 5 The dpipe rmon2 Polling Agent 0200 e eee eee 17 6 The Protocol Definition Interface 02 17 6 Using the Protocol Definition Interface to Populate dsi protocols 17 9 Identifying and Renaming Invalid Protocols 17 13 Deleting Protocols from the dsi protocols Table 17 15 Combined Protocol Identifier and Parameter String Values 17 16 TRENDCOPY lk a PLAY ARA ADB he bas chee 18 1 TRENDcopy Syntax 200 eee eee 18 3 Usage NOotes osand naaa prets RE RIED LA ERE 18 7 Using the s S and t Options To Copy Database Tables 18 7 Using the T Option To Specify a Topology Database 18 8 Using the U Option To Specify a User Name 18 9 Using the r and R Options To Copy Rolled Up Data 18 9
505. y Parameters TREND 3 6 User s Guide Specifying Report Titles Report titles are centered at the top of the graph area Of the three title lines two can be set by the user while the third is the date range for data presented in the graph 2 o o Q o e o D S 7 om To change the report titles select the Graph Title option under the View menu to bring up the Graph Title window Graph Title 4 8nterface Health 5 Figure 13 11 Graph Title Window The window displays the current report title values By default Titlel is the name of the report file without the qgr extension Change the titles by typing in new values Click the Apply button to activate the new titles Click Cancel to quit out of the window without changing the original values Adding Grid Lines Grid lines extend the Y Axis hash marks across the graph area to make it easier to determine the actual values of data points in the graph The user can turn grid lines on and off and change their color Grid lines will appear in a printout of the report Keep in mind how they will look on paper while you set them up especially if you have a color screen and a monochrome printer Setting Report Display Parameters 13 19 13 Grade of Service To change the grid line settings select the Grid option under the View menu This will bring up the Grid window Show Grid Grid Color IN feac sj ok Cancel
506. y id A dsi key id Kcam hlhostdata A and b dsi key id B dsi key id and a dsi agg type b dsi agg type and a ta period b ta period Columns in Selected Table and A dsi target name B dsi target name and A dsi table key B dsi table key K100 Isystem perf B Clauses Additional Information Dere erein Car General Estimate of the View OK Cancel Figure 22 5 Define Qualifications Window It is not necessary for you to qualify your view Qualifications are at your discretion If you wish to use only the default qualifications you can access the dialog box then exit without making any changes The default qualifications will be applied 22 8 Elements of TRENDdbview TREND 3 6 User s Guide If you do not wish the view to use the default qualifications you can delete some or all of the default settings and click OK Since TRENDdbview is typically used to view and report on specific data objects rather than on general information we rec ommend that you apply whatever qualifications you consider useful Working with TRENDdbview Starting TRENDdbview Currently TRENDdbview is not directly linked or associated with any other Graphical User Interface For this reason TRENDdbview must be launched from a command line N N D m 2 o ex o 1 Make sure you are logged in as a TREND user 2 At the command line type trenddbview RETURN The TRENDdbview main window appear
507. y poll for the RMON tables which are preloaded into the database by Desktalk Systems Polling for these tables is setup in the Collect Data Periodically window by selecting them from the Poll For list Note Do not poll for RMON tables by loading the RMON MIBs into MIB walker Do not use rmon_collect to poll for interval tables such as the dsx1IntervalTable defined in RFC 1406 Supported RMON Groups The RMON groups supported by TREND and collected by rmon_collect are etherStatsEntry etherHistory tokenRingMLStatsEntry tokenRingPStatsEntry tokenRingMLHistoryEntry tokenRingPHistoryEntry hosts matrix rmon_collect The topN group is not supported because it is not necessary under TREND Using the capabilities of the relational database a superset of the topN function is already available The alarm event filter and capture groups are not supported because they are not ap plicable to the long term performance monitoring and baselining functions of TREND Control Tables The history host and matrix groups all use control tables to store information defin ing the parameters for the various instances of their associated data tables rmon_collect collects the information in the control table entry each time it collects the data in the data table The History Table rmon_collect looks at each instance of the history table and figures out whether the polling interval i will cause an overlap in collected data because the
508. y strings appear in the Com munity Strings file In some cases it is possible for devices on a network to respond to multiple commu nity strings Example Some RMON vendors allow the use of public as the read community string for MIB IItables but require different community strings for RMON tables This community string can also be used to GET MIB II tables In this case the community strings file should list the RMON community string above the public community string Running Discover from the Command Line TREND 3 6 User s Guide Note If a node is discovered with a different community string than the one it had when it was originally discovered its community string will only be updated if the original community string was public oa g 7 1 lt D When you create a Community Strings file we recommend that you list the more common community strings first as this can speed up the dis covery process However you should keep in mind the possibility that your system may include devices that respond to multiple community strings as described above Running SNMP Type Discovery You can initiate SNMP Type Discovery in two ways By the TREND Autopilot when you use it to install a TREND ReportPack From the command line This section describes running this kind of discovery from the command line Note When you use the TREND Autopilot to install a TREND Report Pack the Autopilot will pro
509. y the Sort Order window by clicking on the Sort Order button Sort Order g i E order by as tabe key ase 00 iwy Canen Figure 15 3 Sort Order Window To change the sort order first delete the existing sort order by clicking on the Clear button Next select whether the sort key i e table column should be sorted in as cending or descending order by clicking on the appropriate button in the Order box Select the first sort field from the Column box by clicking on the field s name in the box The key and order you choose will appear on the Order by line You can choose multiple sort keys with a different order for each key When the sort order is the way you want it click the Apply button to save and apply the changes To quit this window without saving the changes click the Cancel button ol D o 2 n 7 TI a c Output to a File Reports can also be printed automatically to a postscript file using the File Print option in the Report Scheduler window see Figure 15 1 When the file name is Output to a File 15 7 15 Report Scheduler created extra characters will be attached to it automatically to ensure that each name is unique Example If you want the output to be written to a file named xyz gif the actual name of the file will be xyz identifier page_number gif where identifier Is a unique string that is added to the file name to ensure uniqueness pag
510. you We recommend that you accept the defaults 8 Click the Build Server button The Status Output window appears When the build is complete the word Done is displayed in the window 9 Click OK The srvbuild question window appears 10 Click ves to close the window and terminate the srvbuild utility After you create the new backup server you need to drop the entry for the old backup server from the sysservers table and add an entry for the new backup server Follow this procedure gt go 0 3 9 a uo OX m a0 o o 3 Backup Server B 17 B Database Administration Invoke the isql utility with the following command isql Usa P Drop the entry for the old backup server named SYB_BACKUP in this example by entering the following directive 15 sp_dropserver SYB_BACKUP 2 gt go Server dropped return status 0 Add an entry for the new backup server bluestone_back in this example by entering the following directive 15 sp_addserver SYB BACKUP null bluestone back 25 go Adding server SYB BACKUP physical name bluestone back Server added return status 0 The remainder of the appendix applies to UNIX and NT platforms Backing Up the SQL Server Installation After you have successfully configured a new backup server use the dump database command to back up your new upgraded SQL Server installation To shut down a backup server specify the backup server
511. you can generate a report to show this count for any single protocol TRENDstep However if you want to generate a single report that shows total IP octets total WWW octets total SNMP octets and total IP octets that are not either WWW or SNMP you need to use TRENDstep to create a different table with each of these statistics as a column in the table That is as shown in Figure A 1 the TRENDstep input table has this information in separate rows The TRENDstep output table shown in Figure A 2 has this information in different columns of the same row Time Period Target Name Table Key other_ip_octets snmp_octets total_ip_octets www_octets Nov 15 1997 1 00 00 PM 134 70 18 241 1 775 5 005 5 781 0 Nov 15 1997 1 20 00 PM 134 70 18 241 1 649 4 007 5 421 765 Nov 15 1997 1 40 00 PM 134 70 18 241 1 1 003 3 692 6 851 2 157 Figure A 2 Example TRENDstep Output Table The Output Table shows three Time Periods 1 00 1 20 and 1 40 We are only interested in the third row for the 1 40 Time Period The output table has seven columns Time Period The polling period of the data Target Name The device RMON2 agent that reports the data gt Table Key The Table Key for the row In this example this value rep resents the interface number of an RMON probe It is a new Table Key created by TRENDstep 3 D other_ip_octets The total number of octets counted that were not eithe
512. ype 12 10 13 10 Index 1 Index Index 2 aging 2 13 8 3 criteria 1 13 1 14 option 2 16 agx_graph command 12 22 alias 13 4 alias string 10 11 Aliases Expressions 12 5 Aliases Expressions option 13 4 allgos 9 15 allocate space B 10 application Collect Data 1 9 3 5 environment 2 22 modules 2 20 toolkit 1 1 Application layer 17 15 Apply button 2 15 ASCII file 11 1 11 18 20 2 20 3 20 5 20 6 20 16 files 20 14 generator utility 20 2 text format 18 1 20 1 ascii_parm dat 20 6 assign aliases 10 10 create 10 11 Assign Expressions box 12 6 13 8 assign labels 13 4 13 18 Assign Statistic Aliases 12 7 Assign Statistic Aliases Window 10 10 automatic deletion of files A 2 automatic printing 15 3 automatic scale 12 17 Autopilot 4 1 5 4 5 17 Autoscale 12 16 13 16 autoscale 12 17 auto scaling 12 18 available space 8 2 AXON Networks Enterprise Communications Analysis Module ECAM 16 1 Background Color window 2 13 backup Sybase server F 1 base protocol layer identifier 17 16 base table 9 11 baseline 7 16 9 2 9 20 9 23 definition 9 20 period 18 13 default 9 20 rolling 9 20 statistic 9 21 summary row 9 21 value 9 21 batch mode 10 3 block 21 2 nowait 21 3 wait 21 3 broadcast address 5 3 bucket 16 5 width 16 5 button Abort 6 33 Add Node 2 7 Add to View 10 14 Add Type 2 10 Apply 2 15 Cancel 5 8 Collect Data 7 2 Delete 2 18 7 13 Delete Node 2 7 Discover 5 6 Get 6 35 Group by Instance 6 33
513. zed Day of the week by hour for example 1 00 a m on Mondays 2 00 a m on Mondays up to 24 00 p m on Mondays 1 00 a m on Tuesdays 2 00 a m on Tuesdays and so on This grouping results in 7 24 rows in the output table for each element property being summarized Hour of day for example 1 00 a m 2 00 a m 3 00 a m and so on This grouping results in 24 rows in the output table one for each hour of the day for each element property being summarized The statistics that TRENDsum can compute and the grouping options available are much richer than the ones available with TRENDit Therefore the TREND ReportPacks in TREND 3 5 and later releases use TRENDsum rather than TRENDit to create the needed summary tables TRENDsum requires cleaned data as input Cleaned in this context means that raw SNMP counter data has been turned into delta data Thus the input source can be a rate table produced by TRENDit or some other facility a TRENDstep file or another TRENDsum file Other Data Aggregation Functions o TREND 3 6 User s Guide So es z0 2 gt In addition to the requested statistics other columns in the TRENDsum output table E include Co 22 Column Description 5 Name P ta_period Identifies the beginning of the grouping period For example if you are grouping by hour all samples collected between 8 00 a m and 9 00 a m are assigned the ta_period value of 8 00 a m ta_samples Contains th
Download Pdf Manuals
Related Search
Related Contents
EXSYS EX-1333VIS AUTRES COLLES SOLS Philips Tornado Spiral energy saving bulb 872790087630700 Xerox 96 NPS Printer User Manual Menu Management User Guide Menu Location Menu Design Menu Español Sony NW-A3000LEATHER Case for NW-A serie ADvERTENCIA ADvERTENCIA - King Kooker by Metal Fusion manual de instrucciones Manual - Northern Tool + Equipment Copyright © All rights reserved.
Failed to retrieve file