Home

Analysis of ATM Call Detail Records and Recommendations for

1. cece eee cene cence eeaeeeneeenaes 31 4 1 1 Generation of Successful CDR Files and Concatenation CollCate 000 31 4 1 1 1 Generation of the Interim Successful CDRs collect 0 ccce cece eee cence ene eeene ees 31 4 1 1 2 Concatenation of CDR Files concatenate ccc cee e cece eee e eee e eee eeneeeaeeeeaeeeas 36 4 1 1 3 B Shell Version of CollCate CollCateBshell 0 cece cce eee eeee cence eens teen e een ees 37 4 1 2 Commas Substitution SubsComma 0 cece eee cence eect eee ene eeeeeeeaeeeeeenaeenes 38 4 1 3 Generation of the Final CDR Pile ses svsts ceccscivecesus deus ov veasnsosaveende can oevsisseesarecavees 38 4 1 3 1 Generation of Interim File before Call Duration Calculation ecee ence eee 39 4 1 3 2 Generation of Call Durations 3 4 i ieseop ei Qebsedin cei aaeees se eee ade eee 40 4 1 3 3 Generation of the CDR File with Call Duration cece ceese eee ence eee eeene ees 41 4 1 3 4 Final Generation of CDR File with Header 00 c cece eee e nee e eee e eae eeneeenaeeeas 42 4 1 3 5 Standardization of Final Piles c0 c21e ncweeereuyexeuvasweersiveuteaeectauannesds EE PA 4 2 CDR Collection from IBM 8265 Switches eccee cece eee eee eeeeeeeaeeeaeeenaeeeaeaeeas 42 4 2 Introduction to CDR Parainietets oso 3 222s sexed ecoeteavexgiiebiaatlamea EEEE EERE E eii 43 4 2 2 The Ode Introd cton she steseveesen i
2. Sustainable Cell Rate SCR SCR expressed in cells per second characterizes the bursty source and specifies the maximum average rate at which cells can be sent over a given virtual circuit UPC is responsible for policing this parameter for rt VBR and nrt VBR virtual circuits 10 Maximum Burst Size MBS MBS specifies the maximum number of cells that can be transmitted at the PCR assuming the receiving buffers are empty at the beginning of the burst Minimum Cell Rate MCR MCR expressed in cells per second is an ABR traffic parameter that defines the minimum transmission rate to which the network will constrain a source in the event of network congestion RM cells contains this field 2 6 ATM Adaptation Layer AAL The ATM Forum 4 and ITU T Recommendation 22 define ATM in terms of three layers the physical layer the ATM layer and the AAL layer The AAL layer provides support for higher layer services such as signaling and data voice and video applications It divides messages or bit streams from higher layers into ATM cells at the transmitter and reconstructs them at the receiver The AAL is sub divided into the following classes AAL 1 AAL 1 is used for CBR and connection oriented traffic e g 64 Kbps voice and fixed rate video or TDM based circuit traffic It does not support any multiplexing functions Timing information is required to be exchanged between the source and the destination It supports QoS Class 1
3. Appendix C Classification of Call Detail Records The following tables list the categories of parameters that are collected by the toolkits introduced in Chapter 3 and Chapter 4 They are available for use by high level analysis packages There are five categories of parameters e Call identification parameters listed in Table C 1 that identify the connection type the switch s logical port number the virtual path identifier VPI and virtual channel identifier VCD e Traffic parameters listed in Table C 2 that are used for statistical analysis e Class of service parameters listed in Table C 3 that distinguish various classes of service e Statistical parameters listed in Table C 4 which may be useful for statistical analysis e Miscellaneous parameters listed in Table C 5 including time parameters and the call status parameter The toolkit modifies the time format of time parameters from ISO 8601 format to the Unix epoch time format as discussed in Chapter 4 to facilitate the calculation of call duration The call status is used as a filter as mentioned in Chapter 4 87 Table C 1 Call Identification Parameters Explanation Call Reference The Q 2931 call reference value for this SVC Calling and Called Number NSAP Network Service Access Point address of calling and called party Call Origination Method Point to point point to multipoint multipoint to point or multipoint to multipoint Call Type PVC P
4. ICR is the rate at which a source should send initially and after an idle period It is expressed as cells per second Rate Increase Factor RIF RIF controls the amount by which the cell transmission rate may increase upon receipt of an RM cell RIF is given as an inverse of a power of 2 ranging from 1 32 678 to 1 Nrm Nrm is the maximum number of cells a source may send for each forward RM cell Nrm is expressed as a power of 2 ranging from 2 to 256 Mrm Mrm controls the allocation of bandwidth between forward RM cells backward RM cells and data cells Mrm is a constant and is fixed at 2 Rate Decrease Factor RDF RDF controls the decrease in the cell transmission rate It is a power of 2 ranging from 1 32 678 to 1 Allowed Cell Rate ACR ACR is the current rate at which a source is allowed to send It is expressed as cells per second Missing RM cell Count CRM CRM defines an upper bound on the number of forward RM cells to be sent in the absence of received backward RM cells The value is an implementation specific integer ACR Decrease Time Factor ADTF ADTF is the time permitted between sending RM cells before the rate is decreased to ICR It is expressed in seconds ranging from 01 to 10 23 with a granularity of 10 ms Trm Trm is the limit of the maximum time allowed between RM cells forwarded by a source It is expressed in milliseconds with a range from 100x27 to 100x2 13 Fixed Round Trip Time FRTT FRTT
5. 0 MBS CLP 0 Tagging 1 requested PCR CLP 0 1 PCR CLP 0 1 SCR CLP 0 1 MBS CLP 0 1 5 2 1 3 QoS Related Objects There are ten objects related to quality of service that are used to describe QoS classes and QoS parameters They are shown below F_QoS Forward QoS Class and B_QoS Backward QoS Class These two objects are specified in the QoS Parameter IE This IE may appear in a SETUP message The possible values can be QoS Class 0 1 2 3 or 4 The meaning of these values has been discussed in Chapter 2 C_CTD Cumulative End to End CTD and M_CTD Maximum End to End CTD These two objects are specified in the End to End Transit Delay IE This IE can be included in SETUP CONNECT ADD PARTY and ADD PARTY ACK messages AF_CDV Acceptable Forward Peak to Peak CDV AB_CDV Acceptable Backward Peak to Peak CDV CF_CDV Cumulative Forward Peak to Peak CDV CB_CDV Cumulative Backward Peak to Peak CDV AF_CLR Acceptable Forward CLR AB_CLR Acceptable Backward CLR The last six objects are specified in the Extended QoS Parameter IE This IE can be included in SETUP and CONNECT messages The Extended QoS Parameter IE is used to indicate the individual QoS parameter values acceptable on a per call basis and to indicate the cumulative QoS parameter values 62 5 2 1 4 ATM Adaptation Layer AAL Object There is only one object to identify the AAL category It i
6. The data Vaiting P The data File 8265 Please ru switch STEP 5 Run server tcl 8265Svcl W lease file SIIl is gen lease file SSV2 is gen lease file SSV3 is gen lease file SCR4 is gen lease file SEPR5 is generated lease lease lease fia erated erated erated erated file SCN6 is generated file SCN7 is generated x You m e You m You m lease file SCT9 is generated lease file SV10 is generated lease file SV11 is generated SvcEntryInterim is generated n Subs SubsBrace ELCOME Entry gt Subs Brace You m You m ay ay You m ay You m ay ay ay ay You are going to process a special file in this file with space input fil e name 8265SvcE nterim ntry 83 urther process urther process urther process urther process further process further process further process To substitute a Interim file generated in step 4 it IE it it You may further process it You may further process it You may further process it ile SC8 is generated You may further process it it it it Brace to generate the final CDR file for this string If you are using 8265CxVclEntry or 8265SvcEntry or 826
7. Toolkit Database Server Figure 5 2 A component model for CDRs generation and collection 5 1 1 CDR Output On Off This command should turn the switch s output of CDRs either on or off It may be desirable to disable generation of CDRs from some special switches for security or other reasons The default for CDR output should be off since some customers may not be interested in the analysis of CDRs 5 1 2 CDR Filter This command should allow users to record or ignore certain types of calls such as PVCs SVCs Smart or Soft Permanent Virtual Path Connections SPVPs Smart or Soft Permanent Virtual Circuit Connections SPVCs or failed calls In our own project we 56 are presently only interested in successful SVC calls If the filter command was available we could use it to select the interesting calls thus simplifying programming and reducing the amount of data that must be processed 5 1 3 CDR Output Destination Configuration This command should allow the user to choose the destination data server IP address port ID output file name and destination path in the server This command should also allow the user to configure the owner of the files where the owner should be an existing account name at the server 5 1 4 CDR Generation Interval Setup This command should allow the user to setup the interval for output of CDRs For backbone switches the CDRs can result in a huge amount of data so the
8. related algorithms and technical tips The thesis attempts to provide necessary descriptions to help users exercise the toolkits effectively However it is the user s duty to read prompt messages carefully while running the program since the prompt messages provide important information 4 1 CDR Collection from FORE Systems ASX Switches The first toolkit is used to collect CDRs from FORE Systems ASX switches such as those used in NET WORK VIRGINIA and the Virginia Tech campus backbone The toolkit includes three programs CollCate the B shell version is CollCateBshell SubsComma and FinalGenerate CollCate collects successful CDRs and generates one integrated CDR file for further processing SubsComma replaces comma delimiters with the TAB character FinalGenerate calls the executable TimeCalculate function and generates the final CDR file for processing by high level analysis applications 4 1 1 Generation of successful CDR files and concatenation CollCate CollCate performs the functions of data collection and file concatenation We introduce data collection code and file concatenation code separately 4 1 1 1 Generation of the Interim Successful CDRs collect The data collection function is written in Unix C shell script It can collect successful CDRs from any FORE Systems ASX switch assuming the program can access a file that is written by the switch and has code required name format Now this program is 31 configu
9. 2 1 Data Collection from NET WORK VIRGINIA Backbone Switches eeee 69 6 2 2 Calculation of Standard Deviation of Received Cells e cece cece eee e eee cence sere 69 References acoso oo EE AAR AEE en cents sence tay ain AT E EAA avaes gine NE AATRES 72 Appendix A ACrony MS eoue i a es LS 74 Appendix B User s Guide for the Toolkits 0 ccc ccc cence cena e eee eeeen tena nena cess 77 vi Ts Fore Systems ASX Toolkit User s Guides g 5252 eieceitii sie cece iki neudedstaee shee eras T11 Iis IBM 8263 Toolkit User s Guide swicicecstepsdunevebsvereban nosy eepe e e a E eens 82 Appendix C Classification of Call Detail Records 0 cece cece cece neene cena essre 87 Vita ae piace ET AEE AE A E op a tad a aaa eet eS nad Metis tata tees caine 89 Vii List of Figures 2 1 UNI and NNI cell header structure a0 cs ses oe cove as ceane desde heeeus Seve av ueeraddcrenebeueved 6 2 2 Object identifiers in the MIB j2 decas secu vests Geeetid ass oy elucan sue ee seamatewameuse ei etu 16 2 3 Signaling Procedure for a Point to Point Connection 00 ccceeeeeee scene eee e eee eeeaseetees 20 2 4 General Me ssag FORMAL iibeieovesoluul ass auctubine cere uated deed ira EENE wut ub ane adden 21 Sr lu Classification Of CDR ws sii ai oes Lae cases a ait Se eh ea ae A Se es 23 5 2 Flowehart of Toolkit Voices susesssvnessawser sieves tev a E wees an ees eae ey eee Nee 26 3
10. ITU T Class A AAL 2 AAL 2 is used for rt VBR and connection oriented traffic to be delivered with fixed delay e g packetized voice or video It supports the multiplexing of multiple traffic streams into one virtual connection Timing information is required to be exchanged between the source and the destination It supports QoS Class 2 ITU T Class B AAL 3 4 AAL 3 4 is a combination of AAL3 and AAL4 Originally AAL3 and AAL4 supported encapsulation of delay tolerant bursty connectionless and connection oriented traffic AAL3 supported QoS Class 3 and AAL4 supported QoS Class 4 AAL 3 4 also performs re sequencing and cell identification operations in support of multiplexing traffic streams into one virtual connection Generally AAL 3 4 supports QoS Classes 3 and 4 ITU T Class C D and QoS Class 0 AAL 5 or Simple and Efficient Adaptation Layer SEAL AAL 5 is similar to class 3 4 but without multiplexing or error detection It has been adapted by the computer 11 networking industry in response to the complexity and implementation difficulties in AAL 3 4 No timing relationship is required between source and destination AAL 5 supports QoS Class 3 ITU T Class C and QoS Class 0 In the data communication area it has become the predominant AAL class 2 7 Service Categories Traffic Parameters and QoS Parameters In the previous sections we described the ATM service categories traffic parameters and QoS parameters indepe
11. Priority CLP CLP is a single bit It indicates two levels of priority for ATM cells Cells with CLP 0 have higher priority than cells with CLP 1 Cells with CLP 1 may be discarded during periods of congestion to preserve the Cell Loss Rate CLR of cells with CLP 0 Header Error Control HEC The one byte HEC field is calculated using a Cyclic Redundancy Check CRC algorithm to correct single bit errors and to detect multiple bit errors It may be used for cell delineation 2 3 Classification of ATM Services The ATM Forum 5 has specified five service categories in relation to traffic management in an ATM network They are listed below Constant Bit Rate CBR CBR service provides a constant or guaranteed rate to transport services such as video voice or circuit emulation which require rigorous timing control and performance parameters Real time Variable Bit Rate rt VBR rt VBR service is like CBR in the sense that it achieves low transit delay but the traffic can vary in its data rate The data here might be compressed video compressed voice with silence suppression or HDLC link emulation with idle removal Non real time Variable Bit Rate nrt VBR nrt VBR is also a guaranteed delivery service where transit delay and jitter are less important than for rt VBR An example traffic type is an MPEG 2 encoded video stream In this case the information may be being retrieved from a disk for one way video distributi
12. and multimedia applications i e it provides rt VBR service Class 3 ITU T Service Class C Class 3 service is VBR service with no end to end timing required It is intended for interoperation of connection oriented protocols such as Frame Relay It is equivalent to nrt VBR and ABR service Class 4 ITU T Service Class D Class 4 service is VBR service with no timing required It is intended for interoperation of connectionless protocols such as SMDS and IP 2 4 2 Quality of Service Parameters ATM is presently the best technology to guarantee QoS and reserve bandwidth Below are the most commonly used ATM QoS parameters 5 Cell Transfer Delay CTD CTD specifies the maximum time for a cell to be transferred from its source to its destination over a virtual connection VC It is the sum of buffering delays propagation delay PD processing and queuing delays 7 Cell Delay Variation CDV CDV specifies the peak to peak cell delay through the network i e the difference between the best case and the worst case achievable on the VC It is an important parameter for time sensitive traffic such as CBR and rt VBR since it is used to specify how closely cells are spaced in a virtual channel connection VCC Due to multiplexing buffering or the insertion of operation and maintenance OAM cells ATM switches are the main cause of introducing cell delay variation 7 Cell Loss Ratio CLR CLR specifies the ratio of the lost cells to
13. brace characters in the interim file are deleted to generate the final CDR file The following sections introduce programs coll_SLI gen_SLI1 8265SvcLogGenerate and SubsBrace All programs are under the directory tcl 8265SvcLogEntry 4 2 2 1 Object Collection coll _SLI1 As the file s name implies file coll _SLII collects object atmSvcLogIndex and outputs the result to the standard output The code is simple but important so we introduce most key statements in the code mib load ibm 65MIBV4 MIB Specific MIB data file must be loaded first to collect the object data Here the IBM private MIB file is loaded set ipAddress lindex Sargv 0 ipAddress will be the first parameter of the command line set s snmp session address ipAddress community 46 private Set s as an SNMP session The address option defines the network address of the SNMP peer The value of address may be an IP address in dotted notation e g 198 169 34 1 or a hostname that can be resolved to an IP address The community option is specific for SNMPv1 and SNMPv2C sessions It defines the community string that is used to identify the sender of SNMP messages The default community string is public The user needs to check the specific 8265 switch if a private community has been defined puts stdout SvcLogIndex gt This command simply adds a title for the output Ss walk SvcLogIndex atmSvcLogInd
14. can run but the user must re name this file with the application specified name The user needs to read the prompts carefully when he or she is not familiar with the file naming method in the toolkit Step 3 Calculate the call duration and generate the final CDR file The executable Unix C shell script FinalGenerate finishes the final step as its name implies This script first calls an executable file TimeCalculate to calculate the call 25 duration of every call Then it adds the Call Duration field to the CDR file Finally it adds a standard header line to the file and removes the interim files to save disk space The calculation of call duration is important for high level analysis The original CDRs only provide the call start time and call collect terminate time using the ISO 8601 time format 16 TimeCalculate written in C is used to calculate the call duration of every call The C language was chosen because a C library provides a header file time h which includes a useful data structure tm to record various time formats and the C library includes a special function mktime that can convert the local time to a calendar value by using tm as its input variable By calling these library functions and using the data structure the code TimeCalculate can easily modify values in the ISO 8601 time format to use the Unix epoch format After format conversion the call duration time difference is easy to calculate Other
15. de seevacsneacacawacsevedenoawe aovaeencawesee radeucawadseoecuensaauganes 4 2 1 ATM Integration of LAN WAN vs iceeeciieenie used sera enkc eed eae ceeeeeioee nw ees 4 22 ATM Cell Header Str t e sce sisses seeren esonte iieo EE e EE Eees 5 2 3 Classification ot ATM Services nease dni nae a E E Red Bae O EE e EE 6 2 4 Quality of Service Classes and Related Parameters ccceee eee eee eee ene eneeeeenaennes 7 2 4 1 QoS Classes Defined by the ATM Forum ssssssssrsrruerereesesssssssssssessserrerrren 7 24 2 Quality of Service P ra ters anseres se eE EAEE EAEE EEE T ESA 8 2 5 ATM Traffic Control Approaches and Related Parameters ossssccccccrreessrrrrerereeesn 9 2 5 1 ATM Traffic Control Approaches sssssssssssseeeerrrrrerrerrresrsssseseererreressesssen 9 2 3 2 ATM Traffic Parameters a cadscre scone sueuleniseutevevusna ers E ERT T EEIE EE EEE EREE 10 2 6 ATM Adaptation Layer AAD eos oc ccceecwedvveeesnes cidhidccentavesureatsadacebnkes vavepeanven vases 11 2 7 Service Categories Traffic Parameters and QoS Parameters ecceceeee cence eeeees 12 2 8 ABR Service Parameters and Resource Management Cells cccceeeeeeneeeee eens 12 2 8 1 ABR Service Patametersvavs 5uccesavnest aces esha ees E e ae ees 13 2 8 2 RM Cell Str Ct re s nenni oneres n e ninini EEE EEEE ed EERE E eae re 14 2 9 Introduction to Management Information Base MIB ssssssssssecrrssrrrssssrrr
16. extract successful CDRs This is implemented by following pipe and filter command usr fore foreview bin fvcrb2a kibitz cdr VT SSubDirectory S YEARSDATESPADHOURSPADMINUTESINTERVAL cc grep _TERM grep v _SETUP_ grep normal grep 470000 gt test TEMPSSubDirectory DATESPADHOURSPADMINUTE The long string after the fvcrb2a utility is invoked is the raw binary CDR file with its path The CDR parameter call status is used as the primary filter to extract successful CDRs There are five possible values of call status classified by FORE Systems 12 For successful CDRs the call status can only be CRM_CALL_INPROG_TERM which represents call type C in Figure 3 1 or CRM_CALL_NEW_N_TERM which is type D in Figure 3 1 So grep _TERM is used in the command line However the call status value of an abnormal case can be CRM_CALL_SETUP_TERM that also includes string _TERM So we need to pick out CDRs which include character string SETUP_ That is done by grep v _SETUP_ The parameter call termination cause is used as the secondary filter to extract successful calls During the test process we found that some CDRs might meet the call status requirement but the call 34 termination cause can be 41 temporary failure 4 or 35 requested VPI VCI not available 4 We still want to exclude these calls to inclu
17. in Chapter 3 the reason is that we need to infer the call duration of successful calls from the in progress calls In every directory there is a readme file an interim file generation program written as a 45 Unix shell script a SubsBrace script written in Perl and multiple pairs of coll_parameter and gen_parameter programs written using Tcl Scotty Compared with amp 265SvcEntry 8265SvcLogEntry category contains more interesting objects such as QoS value bandwidth call start call collect time and service category In the following discussions we consider scripts in the 8265SvcLogEntry directory as an example There are thirteen related objects parameters within this category Each collection of objects is associated with two files coll_parameter and gen_parameter where parameter is just the abbreviation of the full object that uniquely identifies the object So there are 13 pairs of coll parameter and gen_parameter files For example the first parameter is atmSvcLogIndex The pair of files coll_SLI and gen_SLI is used to collect this object Both files are executable The file gen_parameter calls coll_parameter but both of them are transparent to the final user A user only needs to run 8265SvcLogGenerate the interim file generation script written as a Unix C Shell script and SubsBrace The 8265SvcLogGenerate script calls various gen_parameter functions to generate an interim file with proper format By running SubsBrace all
18. indebted to Phil Benchoff Carl Harris John Pollard and the bright and nice people in CNS I also thank CNS for providing me the office workstation and many up to date documents Thanks for Creighton Hager my research mate I really appreciate the way in which we were able to work together He helped testing some codes in my toolkits and he also gave valuable suggestions to my thesis For years my wife Yanhua has been supporting and helping me silently It is not easy to simply say thanks to her I owe her a lot Thanks to my Christian brothers Chuck Schumann Johnny Yu and Xinhua Shi for helping me with my final defense Give all the glory to God Almighty iii Table of Contents VANORE re o Sy ssee si neg cad aotidce head Sovdav heen dick oencdeculs dee oy poeta decd E ii Acknowledgements ne ennen E E R EE N ame ONO EdE E E EE E EAER S iii Table of Contents icici sehicines reife eves KEET EREE Reta ae ee eee iv Listof Figure S ederet eee EEE EEE EE nes are lee weet ENEE E ESEE viii List of TableS s reien ain toes AAE EAEE A EEEE EAEE EEE AEE EAA EEEE bee EEN ix Chapter 1 mtroductionn j 02 55 32 sae oss sg sss 5 dss eee eae gh Ne aaa Gea eA HERRON RST Oe 1 Whe Back round sh swags caterng sty teehee SAE NS aos ht ee AA ae SI et ER 1 Is Rescate a Goals enira toronga cine dea eed EAEE ER E NAE wees E NEEE EE TRE 2 iS Thesis Ore anil zauOn sarrere csp eSEE EEEE EEN PE E EE E EEEE E ENEE EE 3 Chapter 2 Back rout 20 0260 3
19. output interval could be set to be shorter than the output interval of non backbone switches thus reducing the storage required at the switch 5 1 5 CDR Configuration Display The configuration display command should allow the user to view all current configurations This command is useful if the user wants to check whether the current configuration meets the requirements 5 1 6 Additional Commands The above commands are needed to configure the generation and conveyance of CDRs Vendors may also choose to provide more commands For example a vendor might allow specification of a second output server that could be used in the event of a failure at the first server A vendor could also provide commands to turn CDR output on 57 or off at some specific logical ports or on the basis of a VPI VCI rather than the whole switch These schemes are left for the vendor to select and are not considered to be mandatory 5 2 Recommendations for CDR Object and Output Format Standard CDR objects can be subdivided into two groups The first group includes objects such as traffic parameters calling and called party addresses etc These objects can be extracted from ATM information elements IEs and can be stored in a management information base The second group includes objects such as call start time and cells transmitted etc These objects are not included in any IEs They must be observed and measured by the switch at the point where t
20. p k 1 S 5 i a Je 3 71 References 1 J Crowder et al NET WORK VIRGINIA http www networkvirginia net 2 D McDysan and D Spohn ATM Theory and Applications McGraw Hill Companies Inc Hightstown NJ 1998 3 ATM Forum UNI Signaling 4 0 October 1995 ftp ftp atmforum com pub 4 ATM Forum UNI 3 1 Signaling 3 1 September 1994 ftp ftp atmforum com pub 5 ATM Forum Traffic Management Specification Version 4 0 April 1996 6 ITU T B ISDN Application Protocols for Access Signaling Q 2931 February 1995 7 IBM International Technical Support Organization Raleigh Center Asynchronous Transfer Mode ATM Technical Overview October 1995 8 Christos Kolias The ATM Glossary Imark Communications http www imark com com March 1998 9 Geoff Bennett ATM Quality of Service Part 2 QoS in Network Components FORE Systems Inc http academy fore com 1997 10 Randal L Schwards Learning Perl O Reilly amp Associates Inc 1993 11 Daniel Gilly Unix in a Nutshell O Reilly amp Associates Inc 1998 12 Fore Systems Inc ForeView Call and Performance Records http www fore com products 4vw crdinfo htm 13 Fore Systems Inc AMI Configuration Commands Reference Manual March 1999 14 Haissam Alaiwan Rene Castel 8265 BoilerPlate release 4 0 IBM France December 1998 15 J rgen Sch nw lder TU Braunschweig Scotty Tcl Extensions for Network Management Appli
21. script languages like Perl could also have been used The name of the final file has the format SubDirectory finalSYEAR DATE The SubDirectory and files naming convention is discussed in Chapter 4 The flowchart of the first toolkit is illustrated in Figure 3 2 Call fucrb2a to extract successful CollCate CDRs Merge multiple CDRs files into one file Substitute every field ZED delimiter using SubsComma TAR Call TimeCalculate to calculate Call Duration Add Call Duration field FinalGenerate to the CDRs file Add HEADER to the CDRs file to generate the final file Figure 3 2 Flowchart of Toolkit 1 26 3 4 Data Collection from IBM 8265 Switches Data collection from IBM 8265 switches is done quite differently FORE ASX series switches transfer their CDRs to the network management station NMS or data server in real time The IBM 8265 switch holds a history log that contains up to 100 CDRs entries but it does not provide schemes to automatically transfer CDRs file to some external file server The history log can be sent to the NMS when a predefined threshold is reached or when it is requested by the NMS 14 We have to use a special language to talk with the IBM 8265 switch directly based on the analysis of supported MIBs including both public and private MIBs We first need to understand the MIB file including syntax and the semantics of various objects which have been discussed in Chapter
22. service category parameters The ATM Traffic Descriptor IE must be included in a SETUP message If an ABR connection is being requested this IE must be included in a CONNECT message to specify MCR value These parameters can also be specified in the Alternative ATD IE and Minimum Acceptable ATD IE The purpose of the Alternative ATD IE is to specify an alternative traffic descriptor for the negotiation of traffic parameters during call connection setup The purpose of the Minimum ATD IE is to specify the minimum acceptable ATM traffic parameter in the negotiation of traffic parameters during call connection setup These two IEs may be included in SETUP messages Not all combinations of these parameters are valid Table 5 1 shows the valid combination of traffic parameters for best effort connections Table 5 2 lists the valid combination of traffic parameters for any service category Table 5 1 Valid Combinations of Traffic Parameters for Best Effort 4 Combination of Traffic Parameters for Best Effort Connections Forward PCR CLP 0 1 Backward PCR CLP 0 1 Best Effort Indication Available 61 Table 5 2 Valid Combinations of Traffic Parameters in a Given Direction 4 Allowable Combinations of Traffic Parameters for a Given Direction PCR CLP 0 PCR CLP 0 1 PCR CLP 0 PCR CLP 0 1 Tagging 1 requested PCR CLP 0 1 SCR CLP 0 MBS CLP 0 PCR CLP 0 1 SCR CLP
23. successfully the second time then the user should use the following command touch tes TEMPBUR ASX1 K 104130600 This will manually create the file TEMPBUR ASX1K 104130600 under test where TEMP is the required string TEMPBUR ASX1K 1 is the switch s identification 35 0413 is the date string and 0600 is the time string If one file for a specific time is missing then the concatenation function which is introduced in next section will not finish successfully The touch command makes up for the machine s malfunction In more than six months of use this abnormal case has occurred three times After running data collection successfully the successful interim CDRs files are stored under directory test named as TEMP SubDirectory DATE HOUR MINUTE The variable SubDirectory is both the directory name under kibitz cdr VT and the switch s identification name stored in the raw CDRs files Just as the file name implies files under test are temporary files to be accessed by concatenate described in next section They are eventually deleted 4 1 1 2 Concatenation of CDR Files concatenate The concatenate function is also written as a Unix C shell script This code merges all files with a name of the form TEMP SubDirectory DATE HOUR MINUTE in directory test into one file The key command lines in the script are as follows cp test TEMPSSubDirectory DATESPADHOURSPADMI
24. that is at the top of the tree IBM as a network product vendor starts at the enterprise object ID 20 A simple object identifier tree is shown in Figure 2 2 14 Table 2 2 RM Cell Structure Field Octet Bit s Description Header 1 5 All ATM Header PTI 110 If the RM cell is in rate then the CLP 0 else CLP 1 ID 6 All Protocol ID Identifies the service using the RM cell ID 1 for ABR service DIR 7 8 Direction DIR 0 for source generated RM cells DIR 1 for M switch or destination generated RM cells E BN 7 7 BECN Backward Explicit Congestion Notification cell This S bit indicates whether the RM cell is a BECN cell or not BN S 0 indicates a source generated RM cell and vice versa G Cl 7 6 Congestion Indication It is set by the destination or E intermediate switches to indicate congestion in the network The source will decrease its ACR Allowed Cell Rate when receiving a RM cell with CI 1 T NI 7 5 No Increase It is set to 1 by the destination or intermediate Y switches to inform the source not to increase ACR The default P NI 0 will allow the source to increase its ACR E RA 7 4 Request Acknowledge Not used yet Reserv 7 3 1 ed ER 8 9 All Explicit Cell Rate It specifies the maximum value of ACR for the source It is initialized as PCR and can be marked down by the switches depending on the network condition CCR 10 11 All
25. the total transmitted cells for a given virtual channel connection VCC Cell Error Ratio CER CER specifies the fraction of the transmitted cells that contain bit errors when received by the destination Cell Misinsertion Rate CMR CMR specifies the ratio of cells received at an endpoint that were not originally transmitted by the source end to the total number of cells properly transmitted For all applications CER and CMR must be on the order of 10 2 The principal QoS parameters are CTD CDV and CLR Human sensory perceptions and other factors determine the acceptable values of these major QoS parameters for a given application while data communication protocol dynamics define the rest 2 2 5 ATM Traffic Control Approaches and Related Parameters 2 5 1 ATM Traffic Control Approaches A number of ATM traffic management functions defined by the ITU T 21 and the ATM Forum 5 support the standard service categories These functions are used for monitoring and controlling traffic and congestion occurring at an individual switch or a group of switches Some of the functions are listed below Connection Admission Control CAC Before establishing an SVC or a PVC the ATM network uses the CAC function to determine whether to accept or reject a connection request The CAC accepts a connection request only if sufficient resources are available at all related network elements 5 Usage Parameter Control UPC or Traffic Polici
26. thesis The second goal of this research is to provide recommendations for ATM CDR standards including methods to generate CDRs and standard objects that should be available within CDRs The recommendations are based on accomplishing the first goal It is a common goal that vendors strive to make their products have higher performance and a lower price Almost all ATM switch vendors can provide smart congestion control and traffic management algorithms in their switches But at present few vendors seem to pay attention to the overall network traffic management task The recommendations for CDR standards are intended to give new ideas to ATM switch vendors by answering the following questions e What kind of features should be included in ATM switches to help the network manager optimize a whole network s resource management e What kind of MIB data is needed besides the standard MIB data e How can a vendor make ATM switches that are superior to competitors products with respect to network management and monitoring 1 3 Thesis Organization This thesis is divided into six chapters Chapter 2 discusses ATM technology ATM technology is not easy to explain in a technical paper or even in a single book Chapter 2 covers just the basic concepts used in this thesis Topics included in Chapter 2 are justifications for choosing ATM ATM service classification quality of service and QoS parameters ATM traffic control and traffic para
27. this call may in fact have terminated abnormally If this call is successful it would be terminated at some time within the 6 minute duration though we cannot tell the exact time Based on this we can calculate the approximate value of the duration of this call 27 The second toolkit includes multiple packages As part of the second toolkit several codes have been created to finish the approximate calculation of call duration and they are subdivided into the following steps Step1 Collect in progress SVC calls automatically at some fixed time interval In the first toolkit the SVC calls are output automatically in some fixed interval that is configurable by the FORE switch s configuration command However we did not find such a function supported by the IBM 8265 switch The choice of the CDR output interval is fully controlled by the toolkit If the interval is too long some calls may be lost permanently since the history log can contain at most 100 entries If the interval is too short many in progress calls will be redundant Assume an in progress call appeared in the CDRs at 10 20 and it still lasts one more hour then any CDRs collected before 11 20 will include this particular call In the code the connection interval is set as 5 minutes which is same as the interval for the first toolkit Step 2 Add the Collection Time object to every CDRs file The Collection Time object is not included in the group describing i
28. to Hours to Human Max 1152 Kbps management weeks years intervention system Available in or permanent advance network element failure Signaling Seconds to Minutes to Signaling 56 1920 Kbps minutes hours Maybe available before a call Providing a reliable resource management strategy will be a challenge for network managers and developers for NET WORK VIRGINIA or any network to run SVC and provide QoS successfully 1 2 Research Goals There are two key goals for this research The first is to develop a toolkit to automatically process ATM call detail records CDRs extract useful information and provide a uniform format for high level analysis even though the underlying switches are different There are existing mechanisms to monitor separate circuits in the network These mechanisms are used to monitor the general network usage on a link by link basis rather than on a cell by cell basis Based on this idea we created a toolkit to automatically collect data from two different types of ATM switches from IBM and FORE Systems FORE Systems ATM switches are used both in the Virginia Tech campus backbone and in NET WORK VIRGINIA An IBM 8265 switch was also available for testing Analysis of the collected data and the monitoring of network flows will allow the development of traffic engineering models and monitoring of the network to ensure appropriate capacity although this higher level analysis is beyond the scope of this
29. with CLP 0 1 These two parameters are traffic parameters specific to ABR traffic They are included in the ATM Traffic Descriptor ATD IE Objects related to the Call ID ATM Traffic QoS AAL and ATM Service category were discussed previously Except for the connection ID objects the objects described in the other four groups cannot be mixed randomly The fields specified in the Broadband Bearer Capability BBC IE the ATM Traffic Descriptor ATC IE the Extended QoS 64 Parameters IE the End to End Transmit Delay IE and the QoS Parameter IE should be consistent All of these IEs can appear in the SETUP message If a SETUP message is received containing an invalid combination of BBC ATC and BEI the call is cleared with Cause 65 Bearer Capability Not Implemented 3 If the combination of traffic parameters QoS parameters and QoS class in a SETUP message is not a valid combination for the particular ATM service category the call is cleared with Cause 73 unsupported combination of traffic parameters 3 The valid combination of different status fields is listed in Table A9 2 of the UNI 4 0 specification 3 5 2 1 7 Statistical Parameters There are three statistical parameters None of them can be obtained from an IE It is the vendor s responsibility to provide these statistical parameters The descriptions are listed below e Cell_Received The number of received cells since the connection is up e Cel
30. 0 input file kibitz cdr VT WHI 287C BX 1 199905250010_05 cc MINUTE 15 78 input file kibitz cdr VT WHI 287C BX 1 199905250015_05 cc HOUR 23 MINUTE 50 input file kibitz cdr VT WHI 287C BxX 1 199905252350_05 cc MINUTE 55 input file kibitz cdr VT WHI 287C BxX 1 199905252355_05 cc HOUR 24 The Data Collection is Done Starting File Catenation The HOUR is set as 00 at the beginning The MINUTE is set as 00 at the beginning MINUTE 00 INPUT_FILE test TEMPWHI 287C BX 105250000 MINUTE 5 INPUT_FILE test TEMPWHI 287C BX 105250005 NUTE 10 INPUT_FILE test TEMPWHI 287C BX 105250010 NUTE 15 INPUT_FILE test TEMPWHI 287C BX 105250015 HOUR 23 NUTE 40 INPUT_FILE test TEMPWHI 287C BX 105252340 MINUTE 45 INPUT_FILE test TEMPWHI 287C BX 105252345 MINUTE 50 INPUT_FILE test TEMPWHI 287C BX 105252350 MINUTE 55 INPUT_FILE test TEMPWHI 287C BX 105252355 HOUR 24 The File Catenation is Done Waiting Please You need to further process the generated file finaldata WHI 287C BX 1final10525 STEP 4 Run SubsCo
31. 2 Then we need to select a proper language to talk with the MIBs supported by the switch We found that TCL Scotty is a particularly powerful script language Scotty is a software package that allows implementing site specific network management software using high level and string based application programming interfaces APIs 15 It is based on TCL but adds functions for the Simple Network Management Protocol SNMP that are critical for this research project It also has a special mib load command that can load any MIB data file TCL Scotty makes it much easier to collect CDRs from any network element from any vendor as long as the corresponding MIB data file for the network element is available The core files of the second toolkit are written using the Tcl Scotty package Some necessary Unix Shell and Perl script files are included to facilitate the operation Based on the analysis of IBM supported MIBs we know that only objects to describe the in progress SVC calls and failed SVC calls are provided in the MIBs We need to infer the call duration of successful SVC calls from the in progress SVC calls Assume that an SVC call is active at some fixed time since the call record of this particular call was first available in the in progress SVC call list and after some time for example 6 minutes that particular call is not shown in the in progress SVC call list Then we consider this call as a successful call in the toolkit although
32. 3 Flow chart of the calculation of approximate call duration c cece cece eee e ne eee eens 29 4 1 Flow chart Ob collect Prost alt hcywasc7sscnex pent ra aia EE EEE EE EE ARE Ea 33 4 2 Flow chatt Of SUbsSCOmA isss prore preet Enn r E Na PENO EEEE E ENEE TEESE ARER E TEE 39 4 3 Flow chart of Ben SLUM COdC wisi cass xce ria evecare conde ena eee Sea eu taup hen Cem iae eeaa Ha eeenclna te 48 5 1 A functional model for CDRs generation and collection cece cece ence eeeeeeeeeees 56 5 2 A component model for CDRs generation and collection ceeeeeee eee e eee eee eens 56 viii List of Tables L 1 Features of PVG and SMO tsi oye ed ictdcvesusdspcoes AEE E ty pease dee den OE EEEE E Eia 1 2 1 ATM Service Category Atti Diteswess u ves sencgevancadpavaneutanddgws ay cia bereose wi sea biden tenia 12 2 2 RM cell SWUCture iecssicecutauks neronen Ennni oe et EEEE EA EEEE ede EAEE ene ieee 15 2 3 UNI 4 0 Signaling Message Types for Point to Point Connection seescsccccccccrreee 20 5 1 Valid Combinations of Traffic Parameters for Best Effort 0 cece eee eee ee eeeneenens 61 5 2 Valid Combinations of Traffic Parameters in a Given Direction e cece eeeeene es 62 Cal C all Identification Parameters errea idee valet aie e ees an uiue yey Satine Eaa 88 G2 Traffic Parameters roeier riiete re todos sade eenuetnsyeuse ase di EEEREN SEE EEEN 88 C3 Class OF Service Parameters en
33. 5SvcEntry TempDirectory TEMP246SvcEntry06080000 MINUTE 5 INPUT_FILE tcl 8265SvcEntry TempDirectory TEMP246SvcEntry06080005 MINUTE 10 INPUT_FILE tcl 8265SvcEntry TempDirectory TEMP246SvcEntry06080010 MINUTE 15 INPUT_FILE tcl 8265SvcEntry TempDirectory TEMP246SvcEntry0608001 MINUTE 45 85 tcl 8265Svel You The The The The tcl 8265Svel TempData7 has b INPUT_FILE NUTE 50 INPUT_FILE NUTE 55 INPUT_FILE HOUR 24 Vaiting Please tcl 8265Svc tcl 8265Svc Entry TempDirectory TEMP246SvcEntry06082345 Entry TempDirectory TEMP246SvcEntry06082350 tcl 8265Svc Entry final The File Catenation is Done Entry final current directory is file exists string s data 24 Entry TempDirectory TEMP246SvcEntry06082355 further processing the generated file tcl 8265Svcl Waiting Please l6final10608 data 24 l6final20608 is generated are going to change TempData5 to TempData 6 tcl 8265SvcEntry finaldata Waiting Please ubstitution is done final output file is TempData6 under Entry finaldata n generated with CallDuration included The string substitution is done The files templout and temp2Zout are generated tcl 8265SvcEntry finaldata 246final0608 The final file is server tcl 8265SvcEntry gt 86
34. 5SvcLog or 8265vcxXConnectEntry tInterim as input file It is strongly recommended that the output file should have following name format 8265CxVclEntry or 8265SvcEntry or 8265SvcLog or 8265vcXConnectEntry Today s date Output File Name 8265SvcEntry20526 The final file name There are two 8265 switches This one is regarded as the second one by the researcher so 2 is added before the date 0526 The output file is 8265SvcEntry20526 Correct Y N y Yes HINT For IBM CDR data collection you need to substitute by the white space search String s 9 is the only choice GI You are going to substitute in 8265SvcEntryInterim with SPAC The string substitution is done The final output file is 8265SvcEntry20526 Note If the file wroger crontab under batch includes following lines 00 tel 8265SvcEntry Generate24682 05 tel 8265SvcEntry Generate24682 10 tcl 8265SvcEntry Generate24682 15 tcl 8265SvcEntry Generate24682 20 tcel 8265SvcEntry Generate24682 25 tcel 8265SvcEntry Generate24682 30 toel 8265SvcEntry Generate24682 35 tol 8265SvcEntry Generate24682 40 tol 8265SvcEntry Generate24682 45 x tcol 8265SvcEntry Generate24682 50 tol 8265SvcEntry Generate24682 55 x tol 8265SvcEntry Generate24682 Then step 4
35. Analysis of ATM Call Detail Records and Recommendations for Standards by Xianrui Roger Wang Thesis submitted to the Faculty of the Virginia Polytechnic Institute and State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE in Computer Engineering Scott F Midkiff Chairman Ira Jacobs Alex Q Huang June 1999 Blacksburg Virginia Keywords ATM Call Detail Records CDRs Network Management Analysis of ATM Call Detail Records and Recommendations for Standards By Xianrui Roger Wang Dr Scott F Midkiff Committee Chair Department of Electrical and Computer Engineering ABSTRACT Data network resource management and capacity planning are critical for network design operation and management Equipment vendors often provide good information for traffic management and control and associated tools but this information and the tools are based on independent individual switches or routers rather than the whole network There is a critical need for tools to monitor general resource usage in a network as a whole In this research we develop a toolkit to collect ATM Call Detail Records CDRs from two types of ATM switches from IBM and FORE Systems Data records collected by the toolkit can then be used to assess network resource utilization and traffic characteristics with the objective of predicting future needs making proper network management decisions and ultimately assisting in th
36. Backbone Switches We have been exercising the toolkit for FORE Systems ASX ATM switches using backbone switches in the Virginia Tech campus network We have not yet tested the toolkit on backbone switches in NET WORK VIRGINIA The CDRs from these backbone switches should have the same output format but will be produced at a much higher rate It may be more convenient to modify the collection period in the toolkit from 24 hours to 12 hours or less It is not difficult to re program the toolkit codes to do this but testing is needed Both the size of the final data file and the collection time need to be considered More system resources such as RAM and hard disk space are needed to open a data file with bigger size Another consideration for collecting CDRs from NET WORK VIRGINIA backbone switches is that a specific data server needs to be assigned for data collection 6 2 2 Calculation of Standard Deviation of Received Cells Another pending job is the calculation of the standard deviation of the quantity of received cells per minute in the toolkits It would be helpful to add two columns in the final generated data file The first column would be MEAN mean value of the quantity of received cells per minute The second column would be STD the standard deviation of quantity of received cells per minute The reason why the mean and standard deviation are not currently calculated is briefly discussed below 69 The calculation of mean and
37. Current Cell Rate Set by the source to its current ACR Its value is between MCR and PCR For BECN cells CCR 0 MCR 12 13 All Minimum Cell Rate It defines the minimum transmission rate to which the source is constrained For BECN cells MCR 0 QL 14 17 All Queue Length Not used SN 18 21 All Sequence Number Not used Reserved 22 51 All Reserved 52 8 3 CRC 10 52 2 1 Cyclic Redundancy Check The CRC 10 generating 53 All polynomial is 1 x x 4 x 5 x x 10 15 root ey a eee ccitt 0 iso 1 joint iso ccitt 2 org 3 dod 6 internet 1 directory 1 mgmt 2 experimental 3 private 4 mib 1 enterprise 1 ee a A system 1 interface 2 at 3 ip 4 ibm 2 ibmProd 6 atmSw 33 Figure 2 2 Object identifiers in the MIB From the simple structure above we know every object s ID in an IBM 8265 switch s private MIB begins with 1 3 6 1 4 1 2 6 33 2 9 1 Public MIBs Supported by the IBM 8265 Switch The IBM 8265 switch supports the following SNMP based public MIBs 14 e MIB II Version 1 1 and 1 2 RFC 1213 and MIB II Evolution RFC 1573 This MIB describes managed objects used for managing network interfaces e IETF AToMIB RFC 1695 This MIB describes objects used for managing ATM based interfaces devices networks and services It includes the following groups 16 1 The ATM Interface Configuration Group It contains information on ATM cell layer
38. I Best Effort Indicator This field was discussed in Section 6 2 1 2 since it is included in the ATM Traffic Descriptor IE It is used to indicate if the service is a UBR best effort service or a non UBR non best effort service 1 means 63 the service is a best effort service and 0 implies it is non best effort service Not all combinations of the above service category parameters are valid The valid combinations of Bearer Class ATC and BEI are clearly defined in Table A9 1 of the UNI 4 0 specification 3 5 2 1 6 ABR Related Objects The objects to describe ABR service are listed as a separate category because ATM Forum UNI 4 0 3 added ABR capability The nine objects are listed below e F_ICR Forward ABR Initial Cell Rate B_ICR Backward ABR ICR F_TBE Forward ABR Transient Buffer Exposure Identifier B_TBE Backward ABR TBE FRTT Cumulative RM Fixed Round Trip Time e F_RIF Forward Rate Increase Factor e B_RIF Backward RIF The above seven objects are specified in the ABR Setup Parameters IE that contains mandatory parameters UNI 4 0 also specifies an ABR Additional Parameters IE that contains optional parameters during the call connection establishment Both of these two IEs may be included in SETUP and CONNECT messages The ABR Setup Parameters IE is required if the service category is ABR e F_MCR_CLP_01 Forward ABR Minimum Cell Rate with CLP 0 1 and B_MCR_CLP_01 Backward ABR MCR
39. I cell has a General Flow Control GFC field that is in the first byte bit 8 to bit 5 An NNI cell uses all of byte 1 as the Virtual Path Identifier VPI The fields in the header are defined as follows 4 General Flow Control GFC A 4 bit field that supports simple multiplexing It is always coded as all zeroes under current ATM specifications Virtual Path Identifier Virtual Channel Identifier VPI VCI The length of the VPI field is 8 bits in UNI cells and 12 bits in NNI cells The VCI field is always 16 bits long Some values are reserved for special purposes such as signaling and some values are available upon negotiation between the user and the network and are used for routing Payload Type PT The 3 bit PT field discriminates between a cell carrying management information or a cell that is carrying user information Byte 1 Byte 2 Byte 3 Byte 4 Byte 5 Byte 6 Byte 53 8 5 4 1 8 5 4 1 8 1 8 5 4 2 1 8 1 Cc GFC VPI VPI VCI VCI VCI PT 7 HEC PAYLOAD lt UNI Header gt Byte 1 Byte 2 Byte 3 Byte 4 Byte 5 Byte 6 Byte 53 8 1 8 5 4 1 8 1 8 5 4 2 1 8 1 C VPI VPI VCI VCI VCI PT L HEC PAYLOAD P NNI Header Order of bit transmission from first to last Figure 2 1 UNI and NNI cell header structure Cell Loss
40. ITU T recommendations 6 21 22 and Internet Engineering Task Force IETF RFCs http www faqs org rfcs rfc titles html The ATM product manuals and technical documents from IBM 7 14 20 and FORE 9 12 13 are also helpful for developing the standard recommendation The CDR objects are subdivided into eight groups They are discussed in Sections 5 2 2 1 to 5 2 2 8 5 2 1 1 Call Identification Objects Call identification objects are used to specify the call type address reference number port ID and VPI VCI values There are ten objects in this group and descriptions are given below e Call_Source_Reference and Call_Term_Reference These are integer values used to uniquely identify a call The Call Reference IE is included in every message The Call Reference field identifies the call at the local and remote user network interface to which the particular message applies The call source reference value is assigned by the originating side of the interface for a call and the call term reference value is assigned by the destination site of the interface for this call They remain fixed for the lifetime of this call e Call Method This field is used to identify a call as a point to point or point to multipoint call This field can be extracted from the Broadband Bearer Capability BBC IE This IE must be and can only be included in the SETUP message e Call_Type The possible value can be PVC SVC PVP SVP etc This field is not included
41. M signaling support on the device The following information can be accessed from the MIB 1 Number of supported signaling channels 2 Range of reserved VPs and VCs 3 VPI VCI values of the signaling channel on a port 4 The Q 2931 protocol information for each signaling link defined per port 5 Details of failed calls for the ATM interface involved such as the called party and calling party call creation time and call log or collection time failed cause QoS bandwidth service category and clear cause location 6 Details of in progress calls such as signaling VPI VCI of the call the call reference the called party and calling party call creation time and VPI VCI value of the SVC call 7 Details of PVC calls including both entry and exit point of the PVC call Some other information is also included in this MIB but will not be discussed here as it is not relevant for this work 2 10 ATM Signaling Procedure Message Types and Information Elements 2 10 1 ATM Signaling Procedure for Point to Point Connection Our discussions focus on point to point SVC connections rather than point to multipoint connections The signaling procedure for setting up a point to point SVC 19 connection is shown in Figure 2 3 The signaling procedure for disconnecting a point to point connection is similar Originating UNI Calling User SETUP Originating Switch CALL PRCESSING CONNECT CONNECT ACK Destin
42. Microsoft VC 6 0 compiler and SunOS with the gcc compiler 18 The executable code TimeCalculate was generated for the SunOS platform To run TimeCalculate the user needs to input file SubDirectory final3 DATE after TimeCalculate at the command line The result is piped to file inaldata SubDirectory final4 DATE for further processing The file SubDirectory final4 DATE includes three columns Call Start Time Call Collect Time and Call Duration where the first two parameters have been changed to the Unix epoch time format 4 1 3 3 Generation of the CDR File with Call Duration Given the call duration this value must be inserted into the original CDR file In Unix cut and paste commands are useful in performing such functions The cut command can select a list of columns or fields from one or more files The default field delimiter is TAB The paste command can merge corresponding lines of one or more files into vertical columns separated by a TAB Columns 4 and 5 of file SROUTE SubDirectory final2 DATE are cut to calculate call duration and to generate the file SROUTE SubDirectory final4 DATE We use the cut command to cut the first three columns of file SROUTE SubDirectory final2 DATE to a file temp and the columns 6 through 43 inclusive to another file temp2 Then we use the paste command to merge three files templ SROUTE SubDirectory final4 DATE and temp2 to generate the file ROUTE SubDirectory final5 DATE 41 4 1 3 4 Fin
43. NUTE test OLDfinallSDATE cat test OLDfinallSDATE test TEMPSSubDirectorySDATESPADHOURSPADMINUTE gt test finallSDATE The basic function is to first create an initial file OLDfinall DATE when HOUR MINUTE 00 then concatenate the succeeding files one by one There are two while loop conditions not listed above The inter loop is to check if SMINUTE lt 55 the outer loop is to check if HOUR lt 23 The file TEMP SubDirectory DATE PADHOUR PADMINUTE is accessed sequentially according to the generated time After finishing concatenation a new file named SubDirectory finall DATE is generated under directory finaldata for the next stage 36 The strings PADHOUR and PADMINUTE appear in some toolkit codes The reason is explained in the following example Assume HOUR 00 at the beginning of data collection or file concatenation If the value of HOUR appears in a file name as TEMP HOUR then the correct file name is TEMPOO Later we will increase the value of HOUR using the C shell command set HOUR expr 1 HOUR After evaluating this command HOUR will become 1 rather than 01 So TEMP HOUR will be TEMP and TEMP PADHOUR will be TEMPO assuming PADHOUR Zero HOUR They are two different files So we have to add the 0 padding in front of HOUR For the same reason we use PADMINUTE rather than MINUTE 4 1 1 3 B Shell Version of CollCa
44. SLFQ8_temp SLBOQ9_temp SLFB10_temp SLBB11_ temp SLSC12_temp SLCCL13_temp gt 8265SvcLoginterim The latest generated file 8265SvcLogInterim contains vertical columns separated by a TAB character The format of file S265SvcLogInterim is as follows SvcLogIindex SvcLogInterfaceIndex SvcLogCallingNumber 2147483453 601 47 00 00 15 40 34 20 00 10 20 00 00 07 00 A0 B1 07 0A C6 00 2147483454 601 47 00 00 15 40 34 20 00 10 20 00 00 07 40 00 82 10 00 01 03 2147483455 601 47 00 00 15 40 34 20 00 10 20 00 00 07 00 A0 B1 07 0A C6 00 Every item in this file is followed by a useless brace which is deleted as described in the next section 50 4 2 2 4 Elimination of Braces SubsBrace The SubsBrace script is used to delete every brace character in the interim file It is written as a Perl script and is a modified version of the file SubsComma introduced in Section 4 1 2 After running SubsBrace the final CDRs file for this category is generated Note that there are four categories so there are four CDR files generated for each IBM 8265 switch 4 2 2 5 B Shell Version of Interim File Generation GeneratelpAddress The two B shell versions of 8265SvcEntryGenerate in the 8265SvcEntry directory are Generate24682 and Generate25090 In the lab the last two bytes of the two IBM 8265 switches IP addresses are setup as 246 82 and 250 90 The user can easi
45. The initial successful CDRs file SubDirectory finall DATE is automatically created under finaldata The status is sent to the user automatically as an electronic mail message The user only needs to run SubsComma and FinalGenerate to post process the collected CDR file These two functions are introduced below If the file wroger crontab is modified the user must run the command crontab wroger crontab to reactivate the entry 4 1 2 Commas Substitution SubsComma As the name implies the function of SubsComma is to substitute a TAB character for every comma in the file generated by CollCate As mentioned in Section 3 2 1 it is more convenient for high level analysis programs to have each field separated by TAB SubsComma is a simple Perl script The flow chart is shown in Figure 4 2 During execution the user is asked to input the file name with form SubDirectory finall DATE which was generated by CollCate The output file name must have the form SubDirectory final2 DATE since this file will be processed later The user should pay attention to the prompt messages when running the program 4 1 3 Generation of the Final CDR File The Unix C shell script FinalGenerate finishes the final generation of the CDR file FinalGenerate is a combination of multiple functions We introduce the individual functions below 38 Change directory to finaldata Does the file exist Yes Input the file name to be ge
46. VP Q 2931 SVC or PNNI SPVC Call Source Destination Port The switch s logical port where the call enters or leaves and ID VPI and VCI the value of VPI VCI Table C 2 Traffic Parameters Table C 3 Class of Service Parameters A binary bit 1 means the call is a best effort call and 0 means the call is not a best effort call 0 UBR 1 CBR 2 rt VBR 3 nrt VBR or ABR 4 nrt VBR Qos Table C 4 Statistical Parameters Table C 5 Miscellaneous Parameters Explanation Call Start Time Time when the call is initiated Call Collect Time Time when the call is cleared Call Duration Duration of the call Call Status Current status of the call 88 Vita Xianrui Roger Wang graduated from Department of Computer Science and Engineering Huazhong University of Science and Technology Wuhan China in July 1988 The degree he received there was Bachelor of Engineering Before studying in Virginia Tech in 1997 he had more than 8 years working experience in three different companies His last job title in China was network consultant in Timeplex Group Beijing Office He has interest in working as a network and system engineer He will be employed by Comrise Technology Inc working as a system engineer for AT amp T New Jersey 89
47. a uniform CDR output to make the switch type as transparent as possible to high level analysis applications The design of the toolkits for the two different switches and the associated collection of CDR data provided experience that can be we hope applied to other network elements such as routers switches and hubs from other vendors After understanding ATM CDRs from different vendors specifically IBM and FORE Systems we can provide recommendations for standardizing CDRs for future ATM switch products 22 In this chapter our classification of CDRs is introduced Then the implementation of the toolkits is discussed Since the toolkit supporting the IBM 8265 switch is created based on Management Information Base MIB files a basic introduction to MIB and MIBs supported by the IBM 8265 switch was discussed in Section 2 9 3 1 Classification of Call Detail Records A CDR is created when a virtual circuit is established We have interest in successful calls since only successful calls will occupy significant network resources At present we have interest primarily in successful SVC calls We classified potentially successful calls which might fail later as type A and type B We classified known successful calls as type C and type D These call types are illustrated in Figure 3 1 Test Period Figure 3 1 Classification of CDRs We record calls of type C and D since they are successful but type A and B calls add some comple
48. al Generation of CDR File with Header The file SROUTE SubDirectory finalS DATE includes only a list of CDRs without object names A header is created manually and concatenated to the beginning of file SROUTE SubDirectory final5 DATE Until now the final CDR file has been generated with the name SubDirectory finalSYEAR DATE under directory finaldata All the interim files should be purged The final file is ready for high level processing 4 1 3 5 Standardization of Final File Section 5 2 1 in Chapter 5 discusses the recommendation for CDRs objects Under directory batch we created a Standard_Header file that includes every recommended CDR object At the end of FinalGenerate evaluation following message will display Do you want to standardize the final file y n If the answer is y then a file with name SubDirectory Standard YEAR DATE is generated under finaldata This file includes some empty fields for objects whose values are not available 4 2 CDR Collection from IBM 8265 Switches The second toolkit is used to collect CDRs from IBM 8265 switches It has been mentioned that collection of CDRs from IBM 8265 switches is different from collection of CDRs from FORE switches A good understanding of MIBs supported by the IBM 8265 is critical After loading and checking every public MIB supported by the IBM 8265 switch we did not find any relevant CDR data Some objects identifiers included in public MIBs are empt
49. and step 5 will not be necessary if the user is going to collect data from switch 246 However the user still needs to check tcl 8265SvcEntry TempDirectory to make sure the required files are not empty If these files exist then the user can run FinalGenIBM under tcl 8265SvcEntry The result is shown below 84 server tcl 8265SvcEntry gt FinalGenIBM WELCOME TO USE The DATA Collection Utility Today Is Mon Jun 14 02 38 49 EDT 1999 Pleas nter the YEAR XXXX when you collect the data 1999 The year inputted YEAR 1999 nter the DATE when the data was collected he format of the input should be 4 digit Numbers The initial two s are from 01 12 month He ttt from 01 31 day 0608 the date inputted DATE 0608 You Are Going To Collect Data From One of The following The last two s are IBM switches Following IP Addresses are available BEER DAG LOZ ER DBO 90 Here the is used instead of the real value for security Please Input 246 for the first switch OR 250 for the second Switch 246 The value inputted Starting File Catenation The HOUR is set as 00 at the beginning The MINUTE is set as 00 at the beginning MINUTE 00 INPUT_FILE tcl 826
50. are partly based on the analysis of the available CDR parameters Chapter 2 Background To help readers further reading the related background knowledge is introduced in this Chapter This Chapter includes ATM technology background such as introduction to ATM the ATM cell structure ATM service classifications ATM traffic control etc This chapter also describes the basic knowledge for management information base MIB and MIBs supported by the IBM 8265 switch since the CDR collection from the IBM switch needs well understanding of them 2 1 ATM Integration of LAN WAN ATM is an integration of local area network LAN and wide area network WAN technology The major differentiation between ATM based services and other existing communication services such as IP X 25 Frame Relay leased line SMDS and Ethernet is that ATM is the first technology and protocol structure to effectively integrate voice data and video over the same communication channel with guaranteed quality of service QoS Compared with LANs such as FDDI Ethernet and Token Ring ATM can achieve more reliable QoS Compared with WANs such as Frame Relay X 25 and Switched Multimegabit Data Service SMDS ATM can provide higher access and backbone speeds ATM hardware and software platforms form a communication architecture based on the switching and relaying of cells The size of the cell is fixed at 53 bytes which makes it more convenient for hardware to do mo
51. at five minutes is the default interval value and this value can be adjusted 13 The raw CDRs from FORE switches are temporarily stored in the switch s random access memory RAM and automatically transferred to an external Unix file server in real time using the File Transfer Protocol FTP In the first toolkit we chose the test or data collection period shown in Figure 3 1 to be 24 hours There are several reasons for this The first reason is that since a large amount of data is stored in the data server and the size of the data server s hard disk is limited data more than 54 hours old is automatically purged by the data server The second reason is that we consider that some SVC calls may be long but the duration is usually less than 20 hours So few type B calls see Figure 3 1 will be seen The third reason is that if the test period is too long then the size of final CDR output file would be large and difficult to process and if the test period is too short more final CDR files would be generated Assuming the collection period is set as 12 hours for example two final CDR files will be generated per day Either case will create complexities for the high level analysis application With one day 24 hours as a test period and based on a five minute collection interval there are 288 12 x 24 raw data files that need to be processed and only one final CDR file will be generated per day 24 The first toolkit includes following st
52. ation deletion and management of soft PVCs These two MIBs can be downloaded from ftp ftp atmforum com pub approved specs Interim Local Management Interface ILMI MIB This MIB includes following groups It can also be downloaded from ftp ftp atmforum com pub approved specs 1 The System Group It gives information on the system where interfaces run ILMI 2 The Physical Port Group It gives physical layer information on a particular port such as the status transmission types and cable type 17 3 The ATM Layer Group It tells the maximum number of supported VPs and VCs on the UNI the number of VPs and VCs configured on the UNI and the number of active VP and VC on the interface 4 The ATM Statistics Group It indicates the number of cells received dropped and transmitted on the UNI 5 The VP and VC Groups They provide information on virtual paths and virtual channels such as status traffic shape parameters policing parameters and QoS parameters 6 The Network Prefix and Address Groups They include information about the network prefix in use and the ATM address in use on the user side of the UNI and its validity 7 The Service Registry Group It provides a general service registry for locating ATM network services 2 9 2 Private MIBs Supported by 8265 Switch The IBM 8265 switch supports following private MIBs 14 e IBM Hub Specific MIB Extensions This MIB has the following groups 1 The Traps C
53. ation UNI Destination Switch Called User CALL PROCESSING CONNECT ACK Figure 2 3 Signaling Procedure for a Point to Point Connection 2 10 2 Messages for ATM Call and Connection Control Table 2 3 lists signaling messages for point to point connection control defined by the ATM Forum s UNI 4 0 specification 3 Table 2 3 UNI 4 0 Signaling Message Types for Point to Point Connection Call Establishment Call Clearing Status Messages Signaling Link Messages Messages Management ALERTING RELEASE STATUS ENQUIRY RESTART CALL PROCEEDING RELEASE COMPLETE STATUS RESPONSE RESTART ACKNOWLEDGE CONNECT NOTIFY CONNECT ACKNOWLEDGE SETUP 2 10 3 Information Elements Defined by ATM Forum UNI 4 0 defines 39 information elements IEs used for various messages All defined IEs can be subdivided into the following groups connection identification IEs ABR IEs specified by UNI 4 0 AAL IEs traffic parameter IEs and QoS IEs The most important IEs for our purposes are discussed in this section The messages specified by the ATM Forum are based on the ITU Q 2931 specification Each message has a general format as shown in Figure 2 4 8 7 6 5 4 3 2 1 Protocol Discriminator byte 1 0 0 0 O Length of Call Reference byte 2 Flag Call Reference Value byte 3 Call Reference Value byte 4 Call Reference Value byte 5 Message Type byte 6 7 Message Length byte 8 9 Vari
54. cations http wwwhome cs utwente nl schoenw scotty June 1998 16 ISO8601 Data elements and interchange formats Information interchange Representation of dates and times 1988 17 The Open Group The Single UNIX Specification Version 2 http www opengroup org onlinepubs 7908799 xsh mktime html 1997 18 Free Software Foundation FSF The GCC Compiler http www fsf org software gcc gcc html 1998 72 19 W Richard Stevens TCP IP Illustrated Volume 1 Addison Wesley Longman Inc 1997 20 IBM 8265 MIB Data File Version 4 12 http www networking ibm com 1999 21 ITU T B ISDN ATM Layer Cell Transfer Performance 1 356 October 1996 22 ITU T B ISDN ATM AAL Functional Description 1 362 March 1993 23 Sidnie M Feit SNMP A Guide to Network Management McGraw Hill Inc 1995 24 D Perkins and E McGinnis Understanding SNMP MIBs Prentice Hall PTR 1997 25 D Zeltserman and G Puoplo Building Network Management Tools with TCL TK Prentice Hall PTR 1998 73 Appendix A Acronyms AAL ATM Adaptation Layer ABR Available Bit Rate ATC ATM Transfer Capability ATD ATM Traffic Descriptor ATM Asynchronous Transfer Mode BBC Broadband Bearer Capability BCOB Broadband Connection Oriented Bearer BEI Best Effort Indicator B ISDN Broadband Integrated Service Data Network CAC Connection admission Control CBR Constant Bit Rate CCI Call Connection ID CDR Call Detail Rec
55. configuration of local ATM interfaces on an ATM device in addition to the information on such interfaces contained in the interface table ifTable 2 The ATM Interface DS3 PLCP Group It contains the DS3 physical layer convergence protocol PLCP configuration and state parameters of those ATM interfaces that use DS3 PLCP for carrying ATM cells over DS3 3 The ATM Interface TC Sublayer Group It contains transmission convergence TC sublayer configuration and state parameters of those ATM interfaces which use the TC sublayer for carrying ATM cells over SONET or DS3 4 The ATM Virtual Link and Cross Connect Groups They model bi directional ATM virtual links and ATM cross connects They are used to create delete or modify ATM virtual links in an ATM host ATM switch and ATM network 5 AALS Connection Performance Statistics Group It is an AALS connection table used to provide AALS performance information for each AALS virtual connection that is terminated at the AALS entity contained within an ATM switch or host ATM Supplemental MIB draft ietf atommib atm2 This MIB is a draft from the Internet Engineering Task Force IETF that describes objects used for managing ATM based interfaces devices and services in addition to those defined in some other ATM MIB documents It expired on September 18 1998 PNNI MIB and PNNI MIB Extension The PNNI MIB contains necessary PNNI routing information parameters The PNNI MIB Extension allows the cre
56. data WHI 287C BX 1final50525 81 Final date file generating Waiting please HEADER CATENATION is finished The final generated file is finaldata WHI 287C Bx 1final19990525 This file is ready for SAS Analysis Do you want to standardize the final file y n y The file with standard header is finaldata WHI 287C BX 1Standard19990525 server batch gt To collect CDRs from Whittemore or ISB inter op lab STEP 3 is not necessary since CollCate for these switches will be automatically run every day Il IBM 8265 Toolkit User s Guide STEP 1 Login at server as wroger same as step 1 for Toolkit 1 STEP 2 Change directory to home wroger tcl server gt cd tcl server tcl gt ls 8265CxVclEntry 8265vcXConnectEntry snmp 8265SvcEntry SubsBrace 8265SvcLogEntry exercise STEP 3 Change directory to the Entry sub directory where you want to collect CDRs server tcl gt cd 8265SvcEntry We use 8265SvcEntry in the example STEP 4 Run interim file generation code server tcl 8265SvcEntry gt 8265SvcEntryGenerate Enter the ipAddress of IBM 8265 Switch 198 82 250 90 ipAddress 198 82 250 90 82 Waiting P The data Waiting P The data Waiting P The data Waiting P The data Waiting P The data Waiting P The data Vaiting P The data Vaiting P The data Vaiting P The data Vaiting P
57. de only cause number 16 normal call clearing 4 and cause number 31 normal unspecified 4 So grep normal was added to the command line to finish this function In some cases some raw data files do not exist for example because of a malfunction somewhere in the switch the data server or the network In this case the following error message appears KKK K K K K K K K K K K K K K K K K FK K K DATE 0413 HOUR 6 MINUTE 00 INPUT_FILE 199904130600_05 cc COULDN T FIND ABOVE DATA FILE You Need to Double Check The Directory kibitz cdr VT BUR ASX1K 2 to Make Sure the Input File Exists There KKK K K K K K K K K K K K K K K K K K K K K K In the example above the raw data file 199904130600_05 cc under kibitz cdr VT BUR ASXIK 1 is not available The user should check if the next raw data file 199904130605_05 cc is available under the directory kibitz cdr VT If it is available which means that perhaps only one specific data file was lost then the user has to re start the data collection process Special attention has to be paid to specifying the time input In the example after starting the collection the second time the user needs to tell the program to start at HOUR 06 rather than the default value of 00 and to start at MINUTE 05 rather than the default value of 00 In this example after restarting the data collection process and if the CDRs files are collected
58. e ability to provide reliable quality of service QoS in the network In addition we examine current call detail records and requirements for more comprehensive network management and make recommendations for a standardized CDR Acknowledgments This work has been supported by the IBM Corporation through a University Partnership Program grant The support of Dr Andy Rindos and Mr Jonathan Knop of IBM in securing this grant are gratefully acknowledged This research was conducted as a collaborative project between the Bradley Department of Electrical and Computer Engineering and Communication Network Services CNS Ideas guidance and technical advice provided by Mr Eric Brown and Mr Clark Gaylord of CNS were essential to the success of this research Dr Scott Midkiff had been giving inspiring ideas from the very beginning of the research through the completion of thesis writing and preparation of defense slides His patience and his precision will leave me with a lasting impression Without his help I could not have finished my thesis I would like to give my most sincere thanks to him Dr Ira Jacobs and Dr Alex Huang took their valuable time to read my thesis and contributed valuable suggestions I appreciate their help very much I would like to give special thanks for John Nichols who was my supervisor when I worked in CNS in summer 1998 His consistent support and care for my study at Virginia Tech will be unforgettable I feel
59. eps Step 1 Collect raw CDRs and generate one successful CDR file As part of its private network management system FORE Systems provides a CDR file post processing utility fyvcrb2a 12 which can convert a file of CDRs from machine recognizable binary format to an ASCII format which is recognizable by human beings The toolkit calls this utility to generate the raw ASCII CDR files Then it extracts successful CDRs from the available raw ASCII data files by using the Unix grep command After this by calling the Unix cat command the toolkit merges all the new generated data files into one integrated file for further processing The executable file CollCate written in Unix C shell script implements these functions The detailed function of CollCate is introduced in Section 4 1 1 Step 2 Re format the CDR file generated in step 1 The newly generated data file includes records with fields separated by commas For high level processing it is more convenient to separate each item with a TAB character The Perl script utility SubsComma furnishes this function To run SubsComma the user needs to specify the following information e The input file name which was generated from step 1 e The string needing to be changed which for this specific application is the character oo 99 s e The name of the output file Note that the format of the output file name is strictly defined If the user types the improper output file name the program
60. erent For FORE ASX switches the CDRs are transferred to an external server in real time so the first toolkit needs to post process the available raw CDR file For IBM 8265 switches the CDRs are stored in the switches RAM and will not be transferred to an external server automatically The second toolkit works as an SNMP manager to talk with the switches first The second toolkit also includes similar functions finished in the first toolkit The approach used in the second toolkit is more complex than that used in the first toolkit We recommend that the CDRs should be generated periodically and automatically by the switch The generated CDRs should also be updated frequently to provide usage details on an end to end basis regardless how long the connections may last A recommended functional model for generation and collection of CDRs is shown in Figure 5 1 The components to match functions shown in Figure 5 1 are shown in Figure 5 2 The following commands related to CDR generation and conveyance should be supported by the ATM switch configuration commands 55 Raw Raw Binary Necessary ASCII CDRs Format CDRs Data Conversion Data Server Server Toolkit Application Server Display Database amp Application Access Server Figure 5 1 A functional model for CDRs generation and collection External Data CDRs Server Internet or Private Network External Application
61. ex puts S SvcLogIndex The format of this command is snmp walk varName varbindlist body The walk session traverses or walks a whole MIB subtree The command repeats sending getbulk getnext requests until the returned varbind list a list of MIB instances is outside of the subtree given by varbind list If varbind list is retrieved from the agent the Tcl script body is evaluated In this example the varbindlist is just one simple instance atmSvcLogIndex The varName SvcLogIndex will contain the value of atmSvcLogIndex After evaluating the body puts SvcLogIndex the value of atmSvcLogIndex is printed on the screen 4 2 2 2 Parameter Generation gen_SL 1 The file gen_SLIJ as its name implies generates a parameter file SLIZ that includes the results of atmSvcLogIndex collected by coll_SLI The flow chart of gen_SLII is shown in Figure 4 3 The following code determines if the IP address is active set active catch exec ping SipAddress 2 result if active 0 puts stderr ipAddress is not active exit 47 address and name of loutput file address active Is the file name correct Call coll_SLH and write result to the output file Figure 4 3 Flow chart of gen_SLIJ code In the above set catch and exec are TCL commands and ping is the standard TCP IP command embedded in most Unix systems The catch command in TCL is used to evaluate and trap errors on a specified scri
62. g started cxVclInDiscards The number of cells discarded on this connection since monitoring started cxVcOutCells The number of cells transmitted on this connection since monitoring started cxVclOutDiscards The number of cells to be transmitted and discarded on this connection since monitoring started cxVclRowStatus An entry in this table starts the corresponding monitoring counters for this connection The scripts under 8265vcXConnectEntry collect information about the cross connections established within the switch for all existing Virtual Channel Link VCL based PVCs and SVCs The following information can be collected vcXIndex The connection entrance interface number for this ATM port vcXInVpi The VPI value for this connection in the entrance vcXInVci The VCI value for this connection in the entrance vceXOutIndex The connection exit interface number for this ATM port vcXOutVpi The VPI value for this connection in the exit vceXOutVci The VCI value for this connection in the exit vcXType Indicates if the call is unicast or multicast veXDirection Indicates the direction of the connection upstream versus downstream as seen from the root 4 2 2 The Code Introduction The file structure under each of the four directories is the same except that two B Shell version scripts and FinalGenIBM are included in directory 8265SvcEntry which includes code to collect CDRs of in progress calls As introduced
63. generation and processing is given and some commands related to CDRs are suggested The standard CDR should include 56 objects that are distributed in eight categories Some objects can be extracted from ATM information elements and some objects can be obtained from embedded software within the switch 67 Chapter 6 Conclusions and Future Research 6 1 Summary As part of a research project investigating network management and traffic characterization for networks with dynamically allocated resources such as ATM two toolkits for collecting call detail records were developed In addition based on requirements and the experience of understanding and collecting the CDRs recommendations for a standard method to generate and convey CDRs and for standard CDR objects have been developed The two toolkits developed work with FORE Systems ASX ATM switches and IBM 8265 ATM switches The toolkit for the FORE Systems ASX switches has been exercised using switches in the Virginia Tech campus backbone and should also work with FORE Systems switches in NET WORK VIRGINIA The toolkit for the IBM 8265 ATM switch has been exercised using two switches in the CNS Interop Lab Since the mechanisms to generate CDRs are different for the two types of switches the toolkit designs vary accordingly FORE Systems switches transfer their CDRs to an external data server in real time The toolkit for the FORE Systems switch needs to extract the successful or interesti
64. he following values should be input by the user YEAR 1999 DATE 0413 yesterday HOUR the start hour 00 MINUTE the start minute 0 32 Display welcome message The starting date and time when the CDRs generated The building name The No of fabrics No Print error message Is the raw data file existed Generate temp file with only successful CDRs MINUTE lt 0 HOUR lt HOUR 1 Print message successfully finished End Figure 4 1 Flow chart of collect program 33 After inputting time information the program asks the user to input the location of the switch Presently the user can choose from BUR CAS ISB OWE SHA or WHI as prompted After this the data collection begins or the program asks the user to input either the fabric number or room number This depends on which building the user chooses For example if the user chooses building ZSB then the following prompt will appear Input the fabric 1 2 3 or room 118 208 If the user chooses building SHA then the data collection will begin immediately without further questions since there is only one switch with just one fabric in Shanks Hall Before running this toolkit the user should know exactly from which switch he or she is going to collect data After the user provides time and location information the CDR collection starts without further intervention The key function in the code is to
65. he user accesses the service and presumably recorded in the management information base Every CDR object can be extracted either from the switch s private MIBs or from public MIBs supported by the switch This section provides recommendation for standard CDR objects The related background knowledge about ATM messages information elements and signaling procedure has been described in Chapter 2 The reader should understand how objects in the first group are extracted upon understanding the background knowledge The extraction of objects in the second group is based on embedded software in the switch 5 2 1 Recommendations for Standard CDR Objects Parameters One goal of our research is to provide recommendations for standard CDR objects There is presently no such standard because of several reasons Compared with other services mentioned in Chapter 2 ATM technology is newer ATM backbone networks are not widely deployed yet There exist some backbone switches but most network managers still have not attempted to optimize network management from a higher level such as on a per connection basis Accordingly most vendors have not paid much attention to the inclusion of CDR objects in their ATM switch products We hope that the recommended CDR standard will be helpful for both the public network managers and product vendors 58 The standard CDR objects are recommended based on the collection of CDRs and ATM Forum documents 3 4 5
66. i eases enea Pi e E E TEEN EEEE cause ees aes 88 C4 Statistical Parameters o ceisio sioiias tn E E EEEE EEEE RE E EERE N a 88 C 5 Miscellaneous Parameters ssssssssssreesssssesssrrrrrrerrsrrssseerrrrrereererssereereererreers 88 Chapter 1 Introduction 1 1 Background Asynchronous Transfer Mode ATM has emerged as a promising technology to support advanced broadband multimedia communication services Accordingly its emergence has generated new challenges for data network designers and managers One of these challenges lies in network resource management to guarantee the reliable transfer of traffic for critical applications such as video conferencing distance learning and telemedicine NET WORK VIRGINIA 1 is a Virginia statewide public broadband network It is regarded as one model for the Next Generation Internet NGI It supports video conferencing and traditional data networking and is based on ATM technology Data networking is based on best effort service within the network while video conferencing relies on the inherent quality of service QoS mechanisms in the network At present video conferencing is shifting from using permanent virtual circuit PVC service to switched virtual circuit SVC service The basic features of PVC and SVC are listed in Table 1 1 Table 1 1 Features of PVC and SVC VC Setup mode Setup time Duration Disconnect Speed Type mode Supported by NET WORK VIRGINIA PVC Network Minutes
67. ield should be seconds 5 2 2 CDR Output Format Each CDRs output file should have a standard header with the TAB character as the field delimiter All objects parameters discussed in Section 5 2 1 should be included in the header The standard header file that includes 56 fields is shown below Call_Source_Reference Call_Term_Reference Call_Method Call_Type Call_Start_Time Call_Collect_Time Call_Duration Call_In_Port VPI VCI Call_Status Calling_Party_Address Calling_Party_Subaddress Called_Party_Address Called_Party_Subaddress Cells_Received Cells Rejected Cells_Transmitted F_PCR_CLP_O F_PCR_CLP_O1 B_PCR_CLP_0O B_PCR_CLP_0O1 F_SCR_CLP_0O F_SCR_CLP_01B_SCR_CLP_0 B_SCR_CLP_0O1 F_MBS_CLP_0O F_MBS_CLP_01B_MBS_CLP_O B_MBS_CLP_01 F_QoS B_QoS Cause Call_Out_Port VPI VCI BEI Tagging_Forward Tagging_Backward C_CTD M_CTD AF_CDV AB_CDV CF_CDV CB_CDV AF_CLR AB_CLR AAL_ Type Bearer_Class Service_Type F_ICR B_ICR F_TBE B_TBE FRTT F_RIF B_ RIF F_MCR_CLP_0O1 B_MCR_CLP_01 Call_Termination_Location This standard header file was discussed in Section 4 1 3 5 and Section 4 2 2 6 For any given call some fields must be empty For example for a UBR call the objects related to ABR parameters will be empty 66 5 3 Summary This chapter recommended standards for both the generation and conveyance of call detail records and objects that should be contained by call detail records An ideal model for CDR
68. in any IEs It is the vendor s responsibility to provide the proper values e Called_Party_Address This is the destination s ATM address This field is specified in the Called Party Number IE This IE must appear in SETUP LEAF SETUP REQUEST and ADD PARTY messages Its appearance in LEAF SETUP FAILURE is optional 59 Called_Party_Subaddress The format of this field is similar to Called Party Address This field is specified in the Called Party Subaddress IE The subaddress of the called party is used only to convey an ATM address in the ATM Endsystem Address format across a public network that supports the E 164 format The ATM Endsystem Address is based on the ISO NSAP format but is not an NSAP This IE can be included in any message that includes the Called Party Number IE Calling Party_Address It is the source s ATM end system address This field is specified in the Calling Party Number IE This IE can be included in the SETUP message Calling Party_Subaddress The format of this field is similar to Calling Party Address This field is specified in the Calling Party Subaddress IE This IE can be included in SETUP LEAF SETUP REQUEST and ADD PARTY messages Its appearance is optional Call_In_Port VPI VCI the source port ID and VPI VCI value of source port and Call_Out_Port VPI VCI the destination port ID and VPI VCI value of destination port The port ID can be obtained through public MIB IT ETF RFC 1213 The VPI VCI values a
69. interesting information includes e The service categories that the network supports e The bandwidth QoS and traffic parameters to describe each successful connection e The duration of each individual successful connection e The number of cells which have been transferred or received for a successful connection As a statewide public network NET WORK VIRGINIA provides us with a good research and test bed We need to create a toolkit to collect call records that can include interesting objects High level statistical analysis can be exercised on the available data The toolkit has components that are specific to the particular ATM switches of interest We initially collected CDRs from FORE Systems ATM switches in the Virginia Tech campus backbone switches The architecture of Virginia Tech s campus backbone network is similar to that of NET WORK VIRGINIA which is the ultimate specific target for this work Most of the backbone switches are FORE Systems ASX series switches The campus network carries traffic for a variety of applications including local and Internet data videoconferencing for distance learning and voice NET WORK VIRGINIA carries similar information although relative utilization may differ After exercising the first toolkit developed for FORE Systems ASX switches we developed a second toolkit to collect CDRs data from IBM 8265 ATM switches which are available in a laboratory environment within CNS The toolkits are to provide
70. ioritization and intelligent buffer allocation The buffer architectures include First In First Out Strict Priority Queuing Fair Queuing Weighted Round Robin Queuing and Weighted Fair Queuing A detailed description is provided in 8 Route Management ATM switches maintain a detailed and frequently updated topology of the network including information such as cost bandwidth congestion and connection status Neighboring ATM switches communicate with each other using the PNNI protocol After receiving a call request the routing algorithm strives to supply the optimal end to end route through the network 2 21 2 5 2 ATM Traffic Parameters The ATM traffic contract is an agreement between a user and a network The network provides a specified QoS as mentioned in the previous section if and only if the user s cell flow conforms to a negotiated set of traffic parameters The ATM Forum 5 defined the following traffic parameters Peak Cell Rate PCR PCR expressed in cells per second characterizes the maximum source transmission rate Accordingly 1 PCR is the minimum inter cell interval time for a given virtual circuit The UPC function is responsible for policing the PCR of each virtual circuit over the access circuit Any cell that does not conform to PCR is discarded Cell Delay Variation Tolerance CDVT CDVT expressed in seconds is used for CBR traffic to specify the acceptable tolerance to cell by cell variations of the CDV
71. ircuit SPVP Smart Permanent Virtual Path 75 STD Standard Deviation SVC Switched Virtual Circuit TBE Transient Buffer Exposure TC Transmission Convergence TDM Time Division Multiplex UBR Unspecified Bit Rate UNI User Network Interface UPC Usage Parameter Control VBR Variable Bit Rate VCI Virtual Channel Identifier VPI Virtual Path Identifier VPN Virtual Private Network WAN Wide Area Network 76 Appendix B User s Guide for the Toolkits Fore Systems ASX Toolkit User s Guide STEP 1 Login at server as wroger UNIX r System V Release 4 0 server login wroger Password Last login Tue May 25 23 16 23 from as5300 5 s1016 c Sun Microsystems Inc SunOS 5 5 Generic November 1995 This is server Unauthorized access or use is prohibited You have new mail STEP 2 Change directory to batch server gt cd batch server batch gt STEP 3 Run CollCate server batch gt CollCate WELCOME TO USE The DATA Collection Utility Today Is Wed May 26 10 55 48 EDT 1999 Pleas nter the YEAR XXXX when you collect the data 1999 s The year YEAR 1999 Enter the DATE when the data was collected The format of the input should be 4 digit Numbers T f he initial two s are from 01 12 month The last two s are rom 01 31 day 0525 The date DATE 0525 Enter the HOUR 00 23 when you collect the data Usually the beginning value
72. is 32 bytes long It is difficult for the toolkit described in this thesis to extract calls with same CCI in the final files and to keep track of them Because of the above reasons the mean and standard deviation for this call cannot be calculated by the existing toolkit However it is possible to calculate them with the help of advanced database system such as Oracle which has been shown in the ideal functional model of a CDR collection system in Figure 5 1 To further explain the calculation of the mean and standard deviation the related formulas to calculate the MEAN and STD values of a successful call are provided below 70 xX Mean value of the quality of received cells per unit time in the ith interim file x The quantity of received cells of the ith interim file S fa Standard deviation of the quantity of received cells per unit time in the ith interim file C The quantity of total received cells of until the ith interim file T CDR generation time of the i 1 th interim file T CDR generation time of the ith interim file The initial values x C n C ITa Ty Sy 0 k l 1 1 X x kx x 1 k 1 an i EL k ka where x C C k k 1 S Fig a 2 x tkr x kx i 1 i l i 1 i l k gt Vx k DS tka 2 i l k 1 as k 1 _ 2 kSta k D Xe Pere ae Y x k G Xp i l i l ee 2 _ 2 1 2_ 2 2 xX K S kx Srg Xy PLR FX k
73. is the sum of the fixed and propagation delays from the source to a destination and back The value ranges from 0 to 16 7 seconds TBE The Transient Buffer Exposure is the maximum number of cells a source can initially transmit prior to the first RM cell returning It is expressed as cells ranging from 0 to 16 777 215 CDF The Cutoff Decrease Factor controls the decrease in ACR associated with CRM It has value 0 or is the inverse of a power of 2 from 1 64 to 1 TCR The Tagged Cell Rate limits the rate at which a source can transmit extra forward RM cells TCR is a constant fixed at 10 cells second 2 8 2 RM Cell Structure Table 2 2 shows the structure of RM cell including each field position and necessary description 3 2 9 Introduction to Management Information Base MIB The MIB is the database of information maintained by the agent that a manager can query or set The agent is the software in the network element that runs the management software The manager here refers to the network management system Objects in the MIB are defined using a subset of Abstract Syntax Notation One ASN 1 defined in the Structure of Management Information SMI which is a collection of documents 19 An object identifier is a sequence of integers separated by decimal points specifying an object defined by some organization These integers traverse a tree structure similar to a Unix file system The object identifiers start at an unnamed root
74. jects to describe successful SVC calls We exercised the second toolkit in a lab environment where there are fewer calls compared with production switches It would be hard to select the collection interval and collection period without a comprehensive test As mentioned before the second toolkit includes multiple packages All functions included in the second toolkit are discussed further in Chapter 4 3 5 Summary This chapter introduced approaches to collect CDR from ATM switches To collect CDRs from the FORE ASX switches as used in NET WORK VIRGINIA and the Virginia Tech campus backbone we need to configure the switches properly so that the CDRs are transferred to the file server automatically To collect CDRs from IBM 8265 switches we must first analyze MIB data files supported by IBM and then write code using a script language to collect the interesting CDRs The call duration of successful 8265 SVC calls is estimated so only approximate values are calculated Chapter 4 provides detailed descriptions of the toolkits 30 Chapter 4 Description of the Toolkits Chapter 3 discussed the general approaches for data collection from different ATM switches One toolkit is used to collect CDRs from FORE switches while the second toolkit is used to collect CDRs from IBM 8265 switches This chapter discusses the detailed implementation of the toolkits and includes flow charts and discusses problems encountered during programming and testing
75. l_Transmitted The number of transmitted cells since the connection is up e Cell_Rejected The number of rejected cells since the connection is up 5 2 1 8 Miscellaneous Parameters Objects not included in other groups are classified as miscellaneous parameters The six parameters are described below e Call_Status The vendor should provide this object to describe the status of the call The possible values can be Call Setup Call In progress and Call Terminated e Cause The integer value of Call Termination Cause This field can be found in the Cause IE The Cause IE is mandatory in STATUS DROP PARTY LEAF SETUP FAILURE and ADD PARTY REJECT messages Its appearance in RELEASE and RELEASE COMPLETE messages is optional More than 40 possible values are defined in ATM Forum UNI 3 1 4 and 4 0 3 specifications 65 e Call_Termination Location This field is used to tell where the disconnection occurred It is included in Cause IE too There are seven possible values defined in UNI 3 1 such as User Private Network Serving the Local User Public Network Serving the Local User Transit Network etc e Call_Start_Time and Call_Collect_Time These two parameters cannot be obtained from any IEs It is the vendor s responsibility to provide these two fields The format of these fields can be YYYYMMDDThhmmss where T is a field deliminator to separate date and time e Call_Duration The vendor should provide this field The unit of this f
76. ly recognize the code with the corresponding switch if the user knows the switch s IP address After properly configured in wroger crontab file which was introduced in Section 4 1 1 3 these two files can collect in progress calls in 5 minute intervals The collected files are stored in directory tcl 8265SvcEntry TempDirectory with name format as TEMP246SvcEntry DATE TIME or TEMP246SvcEntry DATE TIME There are 11 objects in the SvcEntry group Call collection time is not given and it cannot be given for an in progress call These temporary files include collection time as the 12th object Compared with the C shell version of this file the inclusion of collection time field is new The following command is used to first create a temporary file which includes one line echo SYEARSDATESTSHOURSMINUTE gt CollectionTimeTemp Then the caf command is used to merge CollectionTimeTemp multiple times the number of in progress calls A temporary file named CollectionTimeTemp2 is generated which includes exactly one column The number of rows in this file is equal to the number of in progress calls 51 4 2 2 6 Final Generation of CDRs File with Call Duration and Standard Header FinalGenIBM As shown in Figure 3 3 FinalGenIBM code contains multiple functions Most of them such as file concatenation and standardization of file header have been introduced during the discussion of the first toolkit Three new commands are i
77. me formats into time since the epoch where structure tm is defined in the standard header file Time h as shown below ifndef TM DEFINED struct tm inttm_sec seconds after the minute 0 59 inttm_min minutes after the hour 0 59 inttm_hour hours since midnight 0 23 inttm_mday day of the month 1 31 inttm_mon months since January 0 11 int tm_year years since 1900 inttm_wday days since Sunday 0 6 int tm_yday days since January 1 0 365 int tm_isdst daylight savings time flag s define TM DEFINED endif The key function in TimeCalculate c is fill It fills the fields tm_year tm_mon tm_mday tm_hour tm_min and tm_sec in the tm object To do that two standard C library functions memcpy copying characters between buffers and atoi converting a 40 string to an integer are used Note that m_year is the number of years after 1900 so 1900 should be subtracted after changing YYYY to integer Otherwise this code will have trouble for calls started in 1999 and terminated in 2000 Also the value of tm_mon is 0 to 11 where 0 represents January and 11 represents December accordingly so 1 should be subtracted after changing MM to integer This problem existed in the source code until testing found some negative values for call duration in some final CDR files TimeCalculate c has been compiled successfully on both Windows 95 with the
78. meters ATM adaptation layer ABR service parameters and resource management cells MIBs ATM signaling message types and information elements This chapter is intended to help readers understand the succeeding parts of the thesis Chapter 3 describes the research approach It introduces the classification of CDRs data collection approaches required for different switches and the related toolkit This chapter discusses the types of CDRs of interest collection of CDRs from switches from different vendors and programming languages and techniques Chapter 4 discusses the detailed implementation of the toolkit used to collect CDRs from the FORE Systems ASX series ATM switch and the IBM ATM 8265 switch To help readers understand the toolkit program flow charts technical tips and important code segments are given Chapter 5 presents the recommended standards for CDRs including generation methods and objects The researcher hopes these recommendations can improve the design of next generation ATM switches Chapter 6 draws conclusions about the research The work that needs to be continued in the future is also described Appendix A lists most of the acronyms used in the thesis The toolkit user manual is given in Appendix B Appendix C includes the classification of the collected CDRs such as identification parameters QoS parameters traffic parameters and statistical parameters It provides background for Chapter 5 since the recommended standards
79. mma server batch gt SubsComma WELCOME 79 You are going to process a special file To substitute a string in this file with TAB The current directory has been changed to finaldata Hint The possible input file name could be SubDirectoryfinallDATE under current directory such as WHI 287C BX 1finall0507 or any other FILE existed SubDirectory is any directory name under kibitz cdr VT Make sure you input the correct file name input file name WHI 287C BX 1final10525 The file name generated in step 3 If you are using SubDirectoryfinallDATE as input the output file should have name SubDirectoryfinal2DATE Such as WHI 287C BxX 1final20507 Output File Name WHI 287C BxX 1final20525 The output file name as requested The output file is WHI 287C BX l1final20525 Correct Y N y Yes HINT For CDR data collection you need to substitute chracter by the TAB 6699 search String Character is the only choice You are going to substitute in WHI 287C BX 1final1l0525 with TAB The string substitution is done The final output file is WHI 287C BxX l1final20525 STEP 5 Run FinalGenerate server batch gt FinalGenerate Pleas nter the YEAR XXXX when you collected the data 1999 YEAR 1999 The year nter the DATE when you collected the data he format of the input should be 4 digi
80. n progress calls For the first toolkit collection time means the time when the call is terminated normally However here this object means a specific SVC call is not terminated at this time It will be used for the calculation of call duration In the second toolkit the collect interval is set to 5 minutes so for any given day the Collection Time object can only be 00 00 00 05 23 55 Step 3 Merge all files collected during collection period The collection period has been temporarily set as 24 hours in the code This value should be adjusted in a real production switch This function was discussed during for the first toolkit Step 4 Make all records collected within the collection period unique Based on the approach used many CDRs will be redundant although the collection time value is different for the redundant records The toolkit needs to extract only the calls with latest collection time 28 Step5 Calculate the call duration and standardize the output format In the latest file generated in step 4 assume a call s call start time is 10 30 and the call collection time is 13 55 Then the call duration is 3 hours and 25 minutes plus 2 5 minutes Adding 2 5 minutes accounts for the fact that the call was terminated between 13 55 and 14 00 We cannot consider calls with collection time equal to 23 55 as successful call yet since these calls may last to the next day Then we have t
81. n sepine nane EENE Or EIEE PEIA E EENE ENA 45 4 22 V Object Collection scol SLIM i rsrrr o ae n A E TS 46 4 2 2 2 Parameter Generation gen SIU oc sccssveuvessn oxy ave vwastweieewe teiee tieaawenvareveaatecssinass 47 4 2 2 3 Interim File Generation 8265SvcLogGenerate ccceeece ence ence cette eneeenaeeees 49 4 2 2 4 Elimination of Braces SUBSBIrace 2 0 vis os ardaces vesdnesoatep sits eduasiece rea es peas esgeereuecae ts 51 4 2 2 5 B Shell Version of Interim File Generation GenerateIpAddress 20cee ee ees 51 4 2 2 6 Final Generation of CDRs File with Call Duration and Standard Header Pal Gen LBM 28 inside ie bact Gioiiha he A ino ash eurceuiledewbacutelamaddiatedawabiadescdss 52 REESI EIE OT 1 ae e out hace va wich yeas baa T AT a bebe a neee as ety he datonssegeebstla 53 Chapter 5 Recommendations for Standards for CDR Generation and OD ICC creiser a asa asa ea EE EA PE 55 5 1 Recommendation for CDR Generation and Conveyance sssseseseeseressesesssssereseee 55 Dall CDR Output On ON artes cleat ide e e a heel Renate et loti aN 56 Os 2 CEDR GD eranen awe wey ance E E a A a ates tel LE Nica aia lea 56 5 1 3 CDR Output Destination Configuration 0 cece cee ee cece ee ene ence ene eneenaeeneenaees 57 5 1 4 CDR Generation Interval Setup ssrsenuuueeesessssssssssrssssrerrrrrrrrrerereresessssen 57 5 1 5 CDR Configuration Display iaviivieveceicues es cx eia cc ieies av dcoke
82. ndently Some parameters are related to some particular categories Table 2 1 lists the relations among them 5 Table 2 1 ATM Service Category Attributes Attribute ATM Layer Service Category Traffic Parameters PCR CDVT Specified Specified Specified SCR MBS CDVT N A Specified QoS Parameters Peak to peak CDV Specified Unspecified MaxCTD Specified Unspecified oe Unspecified Other Attributes Attributes ee EE Specified Notes CLR is low for sources that adjust cell flow in response to control information Whether CLR is specified is network specific 2 8 ABR Service Parameters and Resource Management Cells ABR service is introduced in ATM Forum UNI specification 4 0 3 Compared with other service categories ABR s flow control scheme and its use of resource management cells are unique There are some parameters specifically defined for ABR service category We discuss ABR service in this section 12 2 8 1 ABR Service Parameters The following parameters are used to implement ABR flow control on a per connection basis Note that except for the PCR parameter that we introduced in Section 2 5 2 the parameters are specific to ABR service Peak Cell Rate PCR PCR is the cell rate that the source may never exceed It is expressed as cells per second Minimum Cell Rate MCR MCR is the rate at which the source is always allowed to send It is expressed as cells per second Initial Cell Rate ICR
83. nerated Does the file with the same name exist Substitute all with TAB Figure 4 2 Flow chart of SubsComma 4 1 3 1 Generation of Interim File before Call Duration Calculation The data file SubDirectory final2 DATE described in Section 4 1 2 includes 43 CDR parameters columns or fields The file includes call start time at column 4 and call collect time at column 5 It does not include call duration an important parameter Before calculating the call duration an interim file is generated with name of the form SubDirectory final3 DATE It includes only two columns Call Start Time and Call Collect Time 39 The following Unix command is used to create this file cut f 4 5 SROUTE SubDirectoryS final2SDATE gt SROUTE SSubDirectory final3 DATE It is much easier to calculate the call duration with only the two necessary columns available 4 1 3 2 Generation of Call Duration The TimeCalculate function is written in C The input time format ISO 8601 16 is of the form YYYYMMDDTHHMMSS S where character T separates the values of Date and Time It is easier to calculate call duration after changing the time format above to some standard epoch based time format We chose the Unix epoch time format as our object time format The Unix time epoch is 00 00 January 1 1970 17 The standard C library provides a function mktime struct tm timeptr that converts various ti
84. ng The UPC function monitors and ensures that a connection complies with the traffic contract and the source traffic parameters so that the QoS provided to other connections is not jeopardized It is performed at the UNI To determine compliance with a traffic contract the UPC function implements the leaky bucket algorithm also called the Generic Cell Rate Algorithm GCRA 5 Traffic Shaping Traffic shaping is a method for smoothing bursty traffic that might arrive on a virtual circuit to present a more uniform traffic stream to the network Frame Discard A network element can discard cells on a frame i e AAL protocol data unit The two frame discard mechanisms are Early Packet Discard EPD and Partial Packet Discard PPD 2 5 Explicit Forward Congestion Indication EFCI EFCI is a 1 bit field in the PTI field of the ATM header of a data cell as opposed to an operation and maintenance cell to identify whether congestion has been experienced by the cell A network element in a congested state may set the EFCI bit for use by other network nodes or the destination equipment to avoid further congestion It is used in resource management RM cells in ABR services 5 21 Buffer Management A switch s output port buffer can provide temporary storage for the cells traveling to this port and exceeding the port s transmission capacity Efficient buffer management can maintain the appropriate QoS for each traffic type through pr
85. ng CDRs from the raw data files merge all CDR files generated from the previous stage and generate the call duration of each successful call The languages used to develop the first toolkit are Unix Shell C and Perl The IBM 8265 switch does not automatically move CDRs to an external server so this toolkit needs to talk with the switch directly as an SNMP manager and collect CDRs from the switch at some fixed interval The main language and system used in the second toolkit is the Tcl Scotty script language Based on the approaches we used to extract CDRs from IBM and FORE Systems switches we should be able to develop other toolkits to collect CDRs data from other network elements from other vendors Experience from developing CDR collection procedures from various switches also allowed us to develop a recommended standard for CDR generation and conveyance Study of the CDR objects that were available allowed us to provide recommendations for standard CDR fields We hope that this 68 recommendation will help vendors to consider resource management from both the switch level and the whole network level Based on the recommended standards and the quantitative analysis of the usage distribution for each class which will be provided in a later thesis the network manager should be able to get better ideas about network resource utilization and requirements 6 2 Future Work and Research 6 2 1 Data Collection from NET WORK VIRGINIA
86. ntroduced below e awk T print 612 Gly 2 3 S4 5 66 SI S87 9 STO STI FEN SRoute finaldata SSwitch finallSDATE gt SRoute finaldata templ Awk is a utility to search and process a pattern in a file in Unix 11 The above command re formats the output file to make the last field become the first field e sort n r 4 5 SRoute finaldata temp2 gt SRoute finaldata temp3 Sort is a Unix command to sort and or merge files 11 The option n makes the file sorted by numeric values of the specific fields The option r keeps the sorted list in reverse order for example z precedes a and 2 precedes 7 The field arguments 4 and 5 make the file sorted according to the values in the fifth column and sixth column e unig 1 S Route finaldata temp3 sort n gt SRoute finaldata SSwitch final2SDATE Uniq is a Unix command to display lines of a file that are unique 11 The field argument 1 skips the first field in each line The input file temp3 s first field is collection time However temp3 contains the same record multiple times with different values of collection time For example a call from 12 00 to 13 01 will appear at least twelve times with different collection time given polling interval as 5 minutes We are only interested in the call with collection time equal to 13 00 The unig command only keeps the latest record by deleting the redundant records without comparing the firs
87. o compare files collected in successive days This adds complexity to the toolkit The flow chart of the calculation of call duration is shown in Figure 3 3 Automatically generate TEMP DATE Generate24682 _ file under a special directory Modify OR crontab entry file to run Generate25090 Generate24682 or Generate25090 every 5 minutes Include Collection Time field in every file FinalGenIBM Merge all files collected within collection period FinalGenIBM Make the records unique Is there any call collected FinalGenIBM before 23 25 Calculate the call FinalGenIBM duration and standardize the output format Figure 3 3 Flow chart of the calculation of approximate call duration 29 The above discussion about the approach to calculate the call duration applies to IBM 8265 switches which do not provide objects describing successful SVC calls For any vendor including a group to describe successful SVC calls in the products private MIB should not be a problem Once the successful SVC calls are described the inference of call duration will not be necessary A successful call s call duration should be calculated based on the call start time and call termination time of this call However if such a group is not included in the MIBs supported by the product the calculation of call duration will be a hurdle for the creation of a toolkit We recommend that the switch vendor include necessary ob
88. o that a get next request on the table will retrieve the latest calls first e atmSvcLogInterfaceIndex The ifIndex value of the ATM interface used by this SVC e atmSvcLogCallingNumber The ATM address of the calling party e atmSvcLogCalledNumber The ATM address of the called party o atmSvcLogCreationTime The time when the call was placed e atmSvcLogTime The time when the call was cleared atmSvcLogClearCause The cause of the clearing of the SVC e atmSvcLogForwardQoS QoS with 0 for unspecified 1 for class A 2 for class B 3 for class C and 4 for class D e atmSvcLogBackwardQoS QoS with 0 for unspecified 1 for class A 2 for class B 3 for class C and 4 for class D e atmSvcLogForwardBW The forward bandwidth requested for this call e atmSvcLogBackwardBW The backward bandwidth requested for this call e atmSvcLogServiceCategory The service category with 1 for other none of the following 2 for CBR 3 for rt VBR 4 for nrt VBR 5 for ABR and 6 for UBR e atmSvcLogClearLocation The originator that caused this call to be cleared The scripts under 8265CxVclEntry collect counters for each VC connection following CDRs parameters can be collected e cxVclIndex The interface index ifIndex value of MIB II e cxVclVpi The VPI value for this connection e cxVclVci The VCI value for this connection 44 The cxVclInCells The number of valid cells received on this connection since monitorin
89. on A network transit delay of even a few seconds is not a problem here But low cell loss rate and cell error rate are required Unspecified Bit Rate UBR The UBR service category is for best effort delivery of data No QoS parameters are specified so the network provides no quality commitment The only traffic parameter specified is peak cell rate PCR and all submitted traffic has the CLP bit in the ATM header set to 1 It is suitable for traditional data applications such as LAN emulation LANE and IP transport Available Bit Rate ABR The ABR service category guarantees a minimum cell bit rate and allows a higher bit rate up to PCR if there is available capacity in the network ABR provides a flow control mechanism to regulate the source s traffic rate according to the existing network conditions It also attempts to keep the cell loss rate as low as possible 2 4 Quality of Service Classes and Related Parameters 2 4 1 QoS Classes Defined by the ATM Forum Five QoS classes are defined by the ATM Forum 3 4 Class 0 No equivalent ITU T Service Class Class 0 provides best effort service with no objectives specified for the performance parameters i e it provides UBR service Class 1 ITU T Service Class A Class 1 service is CBR service with end to end timing Class 2 ITU T Service Class B Class 2 service is VBR service with end to end timing It is intended for packetized video and audio in teleconferencing
90. ontrol Group It allows the user to configure traps to be sent 2 The Physical Group This group can be subdivided into module table port table interface table and global throughput statistics 2 1 The ATM Modules Table gives details about the maximum number of supported VPs and VCs per module and the number of VPs and VCs in use It also includes the type of module the number of ports on a module cable type status and supported interfaces e g private UNI private NNI public UNI 2 2 The ATM Ports Table gives information such as the number of ports on a module cable type status and supported interfaces e g private UNI private NNI public UNI 2 3 The ATM Interfaces Table gives information about MIB II interface index and the physical slot and port numbers 18 2 4 The Global Throughput Statistics can be turned on to monitor the total number of cells transmitted through the switch If the monitoring is on the system performance will be negatively influenced 3 The Cross Connect Group It contains the cross connections set up in the switch for all existing virtual channel link VCL based PVCs and SVCs 4 The Neighbor Devices Group contains basic characteristics of adjacent ATM devices attached to this switch Some other groups are also supported but are not particularly relevant to this research so they are not described here Details are in reference 14 e IBM Signaling Extensions MIB This MIB extension defines AT
91. ord CDVT Cell Delay Variation Tolerance CLP Cell Loss Priority CLR Cell Loss Rate CMR Cell Misinsertion Rate CRC Cyclic Redundancy Check CTD Cell Transfer Delay EFCI Explicit Forward Congestion Indication EPD Early Packet Discard FDDI Fiber Distributed Data Interface FRTT Fixed Round Trip Time FT1 Fractional T1 GCRA Generic Cell Rate Algorithm GFC Generic Flow Control HDLC High level Data Link Control HEC Header Error Check ICR Initial Cell Rate 74 IE Information Element ILMI Interim Local Management Interface IP Internet Protocol ISO International Standards Organization ITU International Telecommunications Union LAN Local Area Network LANE LAN Emulation MBS Maximum Burst Size MCR Minimum Cell Rate MIB Management Information Base NNI Network Network Interface nrt VBR Non real time Variable Bit Rate OAM Operations Administration and Maintenance OC Optical Carrier PCR Peak Cell Rate PLCP Physical Layer Convergence Protocol PNNI Private Network Network Interface PPD Partial Packet Discard PT Payload Type PVC Permanent Virtual Circuit QoS Quality of Service RIF Rate Increase Factor RM Resource Management rt VBR Real time Variable Bit Rate SCR Sustainable Cell Rate SEAL Simple and Efficient Adaptation Layer SMDS Switched Multi Megabit Data Services SNMP Simple Network Management Protocol SONET Synchronous Optical Network SPVC Smart Permanent Virtual C
92. ous length IEs as required Figure 2 4 General Message Format The four IEs protocol discriminator call reference message type and message length are mandatory for every message type The protocol discriminator 1 byte is the first part of every message It is used to distinguish Q 2931 messages from other messages The call reference field 4 bytes identifies the call to which this message is related because one user may have multiple simultaneous calls The message type 2 bytes identifies various message types such as those shown in Table 2 3 The message length 2 bytes identifies the length of the whole message Detailed descriptions of ATM messages and information elements are available in the ATM Forum s UNI 3 1 4 and UNI 4 0 3 specifications 2 11 Summary This chapter provided basic background knowledge of ATM needed to follow the remainder of this thesis All ATM technology related discussions are rooted from ATM Forum UNI 4 0 3 and ATM Forum Traffic Management Specification V4 0 5 The discussions on Management Information Bases MIBs are based on 14 23 24 For further information on ATM technology refer to 2 3 4 and 5 21 Chapter 3 Approach Through this research we hope to provide some fresh idea about dynamic network resource management and to optimize large network design Statistical information about successful connections should be helpful The
93. pt It catches any errors that may occur in the script If a script executes without error the result is 0 Otherwise the result is 1 The character 2 in the code is the timeout value If there is no reply from the IP address within two seconds this address is considered inactive The code to write the final result in the output file is shown below set fileId open SoutputFile RDWR CREAT set result exec coll_SLI1 SipAddress puts fileId result close S fileld 48 Here the name outputFile must be SLI for this example More generally gen_parameter calls coll_parameter to generate a new temporary file The name of this file must be parameter It is transparent to the final user since both gen_parameter and coll_parameter scripts are transparent to the user 4 2 2 3 Interim File Generation 8265SvcLogGenerate As indicated above there are 13 objects parameters related to the category SvcLogEntry so there are 13 pairs of coll parameter and gen_parameter files After running each pair of collect and generate functions 13 files will be generated Each file has a header containing the parameter names and multiple records in rows These files are meaningful but the information included in each file is not user friendly The Unix Shell script 8265SvcLogGenerate is used to extract one specific column in every file and merge all the extracted columns into one interim file Thi
94. re specified in the Connection Identifier IE This IE can appear in SETUP CALL PROCEEDING and CONNECT messages If this IE appears in SETUP message then it must be included in the other two messages Some switches may not support the full range of VPI values 0 255 since some VPI bits may be not active VCI values 0 15 are reserved by ITU T VCI values 16 31 are reserved by ATM Forum Only values 32 65 535 are available for users VPI VCI values used by a PVC cannot be used by an SVC 5 2 1 2 ATM Traffic Related Objects There are fourteen objects to describe ATM traffic They are shown below F_PCR_CLP_0 Forward PCR with CLP 0 F_PCR_CLP_01 Forward PCR with CLP 0 1 B_PCR_CLP_0 Backward PCR with CLP 0 60 B_PCR_CLP_01 Backward PCR with CLP 0 1 F_SCR_CLP_0 Forward SCR with CLP 0 F_SCR_CLP_01 Forward SCR with CLP 0 1 B_SCR_CLP_0 Backward SCR with CLP 0 B_SCR_CLP_01 Backward SCR with CLP 0 1 F_MBS_CLP_0 Forward MBS with CLP 0 F_MBS_CLP_01 Forward MBS with CLP 0 1 B_MBS_CLP_0 Backward MBS with CLP 0 B_MBS_CLP_01 Backward MBS with CLP 0 1 e Tagging Forward 0 tagging not requested 1 tagging requested e Tagging Backward 0 tagging not requested 1 tagging requested All of the above parameters are specified in the ATM Traffic Descriptor ATD IE This IE also includes a one byte Best Effort Indicator BEI that is an ATM service category parameter It will be further discussed with the
95. red to operate with switches in the Virginia Tech campus backbone in the following locations e WHI Whittemore Hall a FORE ASX 200BX switch e BUR Burrus Hall a FORE ASX 1000 switch e CAS Cassell Colliseum a FORE ASX 1000 switch e ISB Information System Building a FORE ASX 1000 switch and an ASX 200BX switch e OWE Owens Hall a FORE ASX 200BX switch e SHA Shanks Hall a FORE ASX 200BX switch Most of above switches are campus network backbone switches This code plays an important role in the whole toolkit The flow chart is shown in Figure 4 1 As illustrated in Figure 4 1 the user has to input the starting date and time when the CDRs were generated The program has a special format requirement for the date YEAR and DATE and time HOUR and MINUTE input The format for YEAR is YYYY e g 1999 or 2000 The format for DATE is MMDD e g 0403 or 1126 The format for HOUR is HH e g 00 01 02 23 The format for MINUTE is MM e g 00 05 10 55 Note the MINUTE has to be a multiple of 5 assuming the default collection interval is set to be 5 minutes as explained in Section 3 2 1 This code should typically be run once a day to collect the previous day s data As an example assume a user is running the code on April 14 1999 On this day the user needs to collect CDR data from 00 00 April 13 1999 to 23 55 April 14 1999 since the test period we suggest is 24 hours as explained in Section 3 2 1 T
96. rrrrrrreren 14 2 9 1 Public MIBs Supported by the IBM 8265 Switch sssssssssrrssesssssrrerrrrreerrrserssr 16 2 9 2 Private MIBs Supported by 8265 Switchx v5 20 a jasseuyes vee seats yasiecnes isu aesreeeaseeey ee 18 2 10 ATM Signaling Procedure Message Types and Information Elements 45 19 2 10 1 ATM Signaling Procedure for Point to Point Connection cece eee eee eee e ees 19 2 10 2 Messages for ATM Call and Connection Control cee cceeee sees eee ee ene eeeeeneees 20 2 10 3 Information Elements Defined by ATM Forum ccceeeeeeneeeeeeeeneeeneeenaes 20 ZAM SUMMA a 3c52 vie seds Goede ndeaecessak vod E A E E E O E E A mabe E ENE eee 21 Chapter S Approael sissies ceeseds yids oan oder cub es cond hevy do ee ein eked ee ow TES ESR 22 3 1 Classification of Call Detail RECOrds gs tescinsaesds lt dean cou snan cence sev lee aibenc eunes es cease eeawes eas 23 3 2 Creation Ob TOOK eoe aieea n raea dees seen proc E SEE OERE O TOREEN A ee EREEREER 24 3 3 Data Collection from FORE ASX Switches ssssessessssssssrrrrererrrressrrrrrrrrrrreeeen 24 3 4 Data Collection from IBM 8265 Switches cc ce cee ccc eeee ence cence eens eens eeaeeenaeeeeaes 27 JA UMN ALY yee e AIE NE EERE E AE O AR a a a 30 Chapter 4 Description of the Toolkits 0 00 ccc cece ccc eee e eee e ence eee e ene teen eeeeeee es 31 4 1 CDR Collection from FORE Systems ASX Switches
97. s described below AAL_Type This field is specified in the AAL Parameter IE The AAL IE can be included in messages SETUP CONNECT ADD PARTY and ADD PARTY ACK The inclusion of this IE is not mandatory The possible AAL values and associated meanings are 0 AAL Type 1 3 AAL Type 3 4 5 AAL Type 5 16 undefined AAL 5 2 1 5 ATM Service Category Objects There are three objects to describe the ATM service category They are described below Bearer_Class The possible values are BCOB A Broadband Connection Oriented Bearer Class A BCOB C BCOB X or Transparent VP Service BCOB A is used for connection oriented CBR service BCOB C is used for connection oriented VBR service A network supporting BCOB A or BCOB C may perform inter networking based on the AAL information element BCOB X is used for ATM transport service where AAL traffic type and timing requirements are transparent to the network If Transparent VP Service is specified the user is requesting an ATM only service from the network This service differs from BCOB X in that with the Transparent VP Service both the VCI field and PT Payload Type field will be transported transparently by the network It is included in the BBC Broadband Bearer Capability IE This IE must be and can only be included in the SETUP message Service_Type or ATC ATM Transfer Capability The possible values for this field are CBR rt VBR nrt VBR and ABR It is included in the BBC IE too BE
98. s script file 8265SvcLogGenerate shields all the gen_parameter and coll_parameter functions from the user The user only runs 8265SvcLogGenerate which generates an interim CDR file Some sample commands in the 8265SvcLogGenerate script are listed below e gen_SLI1 ipAddress SLI1 ipAddress is the IP address of 8265 switch The file SLI will be generated with the following format SvcLogIndex 1 3 6 1 4 1 2 6 33 1 9 1 6 1 1 2147483453 INTEGER 2147483453 1 3 6 1 4 1 2 6 33 1 9 1 6 1 1 2147483454 INTEGER 2147483454 1 3 6 1 4 1 2 6 33 1 9 1 6 1 1 2147483455 INTEGER 2147483455 Only the last column is the value of atmSvcLogIndex cut d 3 SLIL gt SLI1_temp This command is used to cut the third field separated by space to generate a temporary file SLI _temp with format as follows SvcLogIindex 2147483453 49 N 147483454 147483455 N Note the brace character is not useful and it is later deleted Similarly the format of other temporary files such as SLII2_temp atmSvcLogInterfaceIndex is shown below SvcLogInterfaceIndex 601 601 501 After generating all thirteen temporary files the following command merges them to form an interim file paste SLIl_temp SLII2_temp SLCN3_temp SLCN4_temp SLCT5_temp SLT6_temp SLCC7_temp
99. should be set as 00 If you don t begin with 00 just type the double digit number like 04 11 etc 77 00 The starting hour HOUR OO Enter the MINUTE 00 05 10 55 when you collect the data Usually the beginning value should be set as 00 Tf you don t start at 00 type 05 10 15 50 55 00 The starting minute MINUTE 00 You Are Going To Collect Data From One of The Following Six Buildings BUR Burrus Hall CAS Cassell Colliseum SB Information System Building OWE Owens Hall SHA Shanks Hall WHI Whittemore Hall Please Input The Building Name Where Your Switch Is HEE a EE HEE EEE EH HEE EE EE EE EEE EEE EEE EE EE EEE EE EE EEE H Make Sure The Building Name You Input is CAPITAL LETTERS HEE a EH HE aE EE EE aE EE EE EE EEE HEE EE EE EEE EE EEE EEE EE H WHI The building name BUILDING WHI You can collect data from following switches Ha EAE EH HH H WHI 287C Bx WHI 287C BxX 2 Ht HH Ht HEH HHH Please input the fabric number 1 2 1 The fabric number You are going to collect data from WHI 287C BX 1 HOUR 00 MINUTE 00 input file kibitz cdr VT WHI 287C BX 1 199905250000_05 cc MINUTE 5 input file kibitz cdr VT WHI 287C BX 1 199905250005_05 cc MINUTE 1
100. st of the switching 2 There are business drivers application drivers technology enablers information and computer networking trends leading to the full deployment and development of ATM technology 2 The basic motivating forces for ATM technology are given below Business Drivers e The move from centralized to distributed operations e Client server applications the need for LAN MAN WAN interconneciton e Bandwidth on demand e Enterprise networking world flexibility and investment protection Application Drivers e Distance learning training and video conferencing e Remote medicine and remote distributed database access e Multimedia applications e Various consumer applications Technology Enablers e Development of hardware technology processors memory electronics e Availability of digital transmission media with higher performance price ratio e Worldwide support of ATM technology Information and Computer Networking Trends e Crisis of routing and need of switching e Deployment of virtual private networks VPNs intranets and extranets e Security and bandwidth requirement 2 2 ATM Cell Header Structure ATM cells have a fixed length of 53 bytes including a 5 byte header 3 4 The header is different depending on whether the cell is used at the User Network Interface UND or Network Network Interface NND The UNI and NNI cell header structures are shown in Figure 2 1 The difference between them is only in byte 1 A UN
101. standard deviation of the quantity of received cells per second for a specific successful call depends on the previous calculation of mean and standard deviation for this call in the previous file Given the five minute collection interval if a call is started and terminated within the collection interval then the MEAN is the received cells divided by call duration and the STD value of this call is zero However most calls will last more than 5 minutes To make the discussion less complex we assume that a successful SVC call lasts 14 minutes and that it spans three CDR files Note that this call spans at least three CDR files since a temporary CDR file is generated in the toolkits every five minutes To calculate the final mean and standard deviation of the quantity of received cells per minute for this call we have to calculate mean and standard deviation for the second interim CDR file of this call Similarly we have to calculate the mean and standard deviation of the first interim file for this call To accomplish this the existing toolkit needs to be modified to collect CDRs of both in progress and successful SVC calls For a campus backbone switch in Burrus hall with collection every five minutes the size of a CDR file with both in progress and successful calls included is more than 30 kilobytes To merge large files makes the toolkit run extremely slowly Each call is uniquely identified by the Call Connection ID CCI The CCI for each call
102. t field For example assumes the temp3 file contains following lines collection_time interface_index VPI VCI 19990608T0925 601 0 33 19990608T0925 601 0 33 52 19990608T0920 601 0 33 19990608T0920 601 0 33 19990608T0915 601 0 33 19990608T0915 601 0 33 19990608T0925 601 0 36 After running the unig temp3 command the following result will be displayed collection_time interface_index VPI VCI 19990608T0925 601 0 33 19990608T0925 601 0 36 To run this code the user needs to input the year and date when the data was collected The IP address of the switch is also input by the user The file will first check whether the required files in tcl 8265SvcEntry TempDirectory are readable or not In the lab environment the two 8265 switches are sometimes turned off due to various reasons This added some difficulty during testing The final generated files with name formatted as Switch finalSDATE are saved in the tcl 8265SvcEntry finaldata directory where Switch can be the string 246 or 250 depending on from which switch the data was collected 4 3 Summary This chapter introduced scripts and programs included in the two toolkits The first toolkit for the FORE Systems ASX switches is stored under directory batch For every switch exactly one final CDR file will be generated To date we have been automatically collecting CDRs from both Whittemore Hall and the Interop Lab in the ISB Every day
103. t Numbers he initial two s are from 01 12 month The last two s are from 01 31 day HH 80 0525 The date DATE 0525 You Are Going To Collect Data From One of The Following Six Buildings BUR Burrus Hall CAS Cassell Colliseum SB Information System Building OWE Owens Hall SHA Shanks Hall WHI Whittemore Hall Please Input The Building Name Where Your Switch Is FE E ae aE aE EH FE FE aE EEE HEE aE EE HE EE EE EEE E AE AE EEE EE EEE EE EE H Make Sure The Building Name You Input is CAPITAL LETTERS HEE a EE HEE EEE EH aE EE EE EE EEE EE EE EEE EE E EE EE EEE EE WHI The building name BUILDING WHI You can collect data from following switches FEAE EAE ET HH H WHI 287C BX 1 WHI 287C BX 2 HEREHERE HHH Please input the fabric number 1 2 1 The fabric number You are going to collect data from WHI 287C BX 1 INPUTFILE finaldata WHI 287C BX 1final20525 please wait File Columns cut is done The new generated file is finaldata WHI 287C BX 1final30525 Starting generating call_duration Waiting please Time Duration calculation is finished The new generated file is finaldata WHI 287C BX 1final40525 Starting pasting Waiting please File paste is finished The new generated file is final
104. te CollCateBshell Before concluding Section 4 1 1 we briefly introduce CollCateBshell which is another version of CollCate written as a B shell script We created a B shell script CollCateBshell and a cron entry file wroger crontab to automatically collect CDRs data at a fixed time every day A B shell script is required since the B shell is the default shell for cron The use of cron eliminates the need to manually log in every day to run CollCate A crontab file consists of lines with a fixed format as follows Each line consists of six fields separated by spaces or TABs The first five fields are integer patterns that specify the minute 0 59 hour 0 23 day of the month 1 31 month of the year 1 12 and day of the week 0 6 with 0 Sunday All five fields can be set to meaning all legal values The sixth field is a command that is executed by the shell at the specified times The following two lines in file wroger crontab are related to the automatic execution of CollCateBshell 00 12 batch CollCateBWHI 10 12 batch CollCateBISB 37 As explained the B shell script file CollCateBWHI is evaluated at 12 00 and CollCateBISB is evaluated at 12 10 every day The first command is to collect CDRs from a switch in Whittemore Hall and the second command is to collect CDRs from a switch in the Interop Lab in the ZSB The user can modify the code to collect data from any switch automatically
105. try The following sections discuss the available CDR parameters in the four different directories Details of the toolkit are also provided 4 2 1 Introduction to CDR Parameters The scripts under 8265SvcEntry are used to collect CDRs for in progress SVC calls The final generated CDR file includes the following parameters e atmSvclInterfaceIndex The index value of the ATM interface used by this SVC e atmSvcSiVpi The VPI value of the signaling channel for this entry e atmSvcSiVci The VCI value of the signaling channel for this entry usually there is one signaling channel per interface defined by VPI VCI 0 5 e atmSvcCallReference The Q 2931 call reference value for this SVC e atmSvcEndPointReference One of the Q 2931 end point reference values used by this SVC e atmSvcCallingNumber The ATM address of the calling party e atmSvcCalledNumber The ATM address of the called party e atmSvcClear This variable allows a network manager to clear this SVC e atmSvcCreationTime The time when the call was placed e atmSvcVpi The VPI value used by this SVC 43 e atmSvcVci The VCI value used by this SVC The scripts under 8265SvcLogEntry collect CDR information about the latest unsuccessful SVC completed in this node The following CDR parameters are collected for SVC calls finished abnormally e atmSvcLoglIndex An ID value for this entry assigned by the local SNMP agent It is allocated in decreasing order s
106. two CDRs files are automatically generated under finaldata The second toolkit for IBM 8265 switches is stored under tcl For each switch there are four categories of parameters so four different CDR files are generated These CDR files provide parameters for in progress SVC calls failed SVC calls counters for each VC connection and information about the cross connection setup in the switch for all existing VCL based PVCs and SVCs The values of call duration of possibly 53 successful SVC calls are estimated These CDR files have not to date been used for high level analysis However the method used to create toolkits should be helpful for further research and data collection from IBM and other switches User manuals for both toolkits are included in Appendix B The classification of available CDRs is shown in Appendix C 54 Chapter 5 Recommendations for Standards for CDR Generation and Objects We have created toolkits to collect CDRs from two different types of switches Based on the different methods used to collect CDRs the various objects present in the CDRs and the requirements for analysis and network management we have developed a set of recommendations for both the generation and conveyance of CDRs and objects within CDRs 5 1 Recommendation for CDR Generation and Conveyance The approaches used in the collection of CDRs from two different ATM switches are different since the CDR generation methods are diff
107. xages or veebarnivearceehawieeseabes 57 5 1 6 Additional Commands ssssssseserssssessssserseresreresssestseeeeeseressssesorreeesrseesese 57 5 2 Recommendations for CDR Object and Output Format Standard 58 5 2 1 Recommendations for Standard CDR Objects Parameters cecceeee cence eeeetees 58 5 2 1 1 Call Identification ODjeCts sx conse neis iaa CE E E ep RANGN FRC Ss 59 5 2 1 2 ATM Traffic Related Opies wc srce asides wu elesaperdany Wits ad swe cuena per a aE 60 52 1 3 Q0S Relat ed Objects o il rn aebherens e040 aries sees Meo er das UN E ss i aE e 62 5 2 1 4 ATM Adaptation Layer AAL Object ccc cee cece cece eee eeeeeeeneeeeneeeneeeenaees 63 5 2 1 5 ATM Service G ate cory ODjeCis acd seu ece sted fesse ceed ted Daw Soe Aas ee iki Rad Ds dowd 63 5 24 62 A BR Related ODiecisa ds stcca e EE N E arses lake antes E EERE eoeees 64 5 2A 7 StatsticalPatamieterS 3 dice ie eis seins Sale ie lhe ice Dati hs oC oN as 8a 65 5 2 8 Miiscellaneous Parameters d 22i2secscespsredenolyad is EIEEE EEOAE E eae E ean ees 65 22 2 CDR Output Format seere enee ea En E EE EA a EE 66 6 3 SUMUMALY enr ca vais atc savanna REAR Caney was das a A cau eve EANES EEEE KART OERE ER EATE 67 Chapter 6 Conclusions and Future Research 0 0 cccee cece ence ences eens ena reese eens 68 6 1 Summarys a a AA ISE AEE E ewes yaw EE AREE E EEEE R eels 68 6 2 F t re Work and Resear ein cyt oo oe Siu 0559s hel tue N E E E E E 69 6
108. xity We ignore type A calls in our analysis In equilibrium type C calls balance type A calls i e the number of type A calls created within the test period that terminate outside of the period should be statistically equivalent to the number of type C calls created outside the period but that terminate within the period Type B calls are less easily dealt with It is not feasible to collect data for type B PVC calls since the duration of such calls may be infinite If type B calls are SVC calls we can ignore them as long as our test period is long enough Few SVC calls observed to date from the Virginia Tech campus backbone lasted more than 20 hours However we can reexamine or reprogram 23 the duration of test period in the future as we broaden the scope of our data collection The test period in Figure 3 1 is explained further in Section 3 3 3 2 Creation of Toolkit Source files in the first toolkit that supports FORE Systems ASX switches are written in Perl 10 Unix Shell 11 and C languages Source files for the second toolkit that supports the IBM 8265 switch are written in Tool Command Language TCL as part of the Scotty system 15 Perl and Unix Shell script languages The following sections explain why multiple toolkits are required and why multiple programming languages were used 3 3 Data Collection from FORE ASX Switches The FORE ASX switch generates the raw call record data file in a 5 minute interval Note th
109. y The IBM 8265 private MIB data file V4 12 the latest version at the time of this writing did not contain information about successful SVC calls However the private MIB does contain some CDR parameters for in progress SVC calls failed abnormal SVC calls counters associated with the virtual connections and information about the switch s internal cross connections Therefore the second toolkit generates CDR files including this information The second toolkit also includes two B 42 shell scripts Generate24682 and Generate25090 which are used as entries in the crontab file so the in progress CDRs can be collected automatically from two 8265 switches The code FinalGenIBM is used to post process these automatically collected CDR files estimate the call duration and standardize the final file Although the generated CDRs will not be as useful for high level analysis as those collected by the first toolkit we are able to accumulate valuable experience from the process of CDR collection and programming that will be helpful for collection of CDRs from other network elements from other vendors Also the CDRs generated from the IBM 8265 switch together with CDRs generated from FORE Systems ASX switches provide a basis for recommended standards for CDR objects as discussed in Chapter 5 The second toolkit is in directory tcl There are four independent sub directories amp 265SvcEntry S8265SvcLogEntry 8265CxVclEntry and 8265vcXConnectEn

Analysis of ATM Call Detail Records and Recommendations for

Contents

Download Pdf Manuals

Related Search

Related Contents