sugi 83 06 jobusch

Upload: helloapurba

Post on 03-Apr-2018

235 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/28/2019 Sugi 83 06 Jobusch

    1/7

    Evaluating the Performance of VM/SP Systems with SAS

    Peter L. Jobusch, SAS Institute Inc.This report describes a system of SASprograms fo r first level analysis of CP Monitordata collected by th e VM/SP operating system.This self-contained system calculates a completerange of standard performance variables an darchives them fo r further SAS based capacityplanning analysis.Before discussing specific examples of how SAScan be used to evaluate the performance of aVM/SP computer system, le t us se e what dataare available from which we may extractperformance information.

    CP monitor dataVM/SP Release 2 trace table recordingCP accounting recordsVM Realtime Monitor (SMART, 5796-PNA)CPWATCH, etc. on the Waterloo tape

    A description of CP monitor data is in AppendixC "VM/sP Monitor Tape Format and Content" ofthe IBM V M I S P ~ System Programmer's Guide.Appendix C is particularly useful because itdescribes th e data fields and gives a crossreference between th e variable name in th eDSECT an d th e name of th e CP variable storedin the records. In th e IBM VMISP: Dota Areasand Control Block Logic th e DSECTS describingthe records are listed.CP Monitor data collection can be controlled intwo ways. Th e SYSMON macro in th e DMKSYSfile establishes the default parameters fo rautomatic monitoring, and th e CP MONITORcommand can ,be used by a privileged user todynamically alter the classes of data beingcollected, an d to manually control th e datacollection process. The SYSMON macro isdocumented in th e IBM VM/SP: Planning andSystem Generation Guide while th.e IBM VM/SP:System Programmer's Guide an d th e IBM VM/SP:Operator's Guide document Use of th e MONITORcommand.

    Copyright 1982 by SAS Institute Inc.All rights reserved. Printed in the UnitedStates of America. No part of this publicationmay be reproduced, stored in a retrievalsystem, or transmitted in an y form or by anymeans, electronic, mechanical, photocopying, orotherwise, without th e prior written permissionof th e publisher, SAS Institute Inc.SAS is th e registered trademark of SASInstitute Inc., Cary, North Carolina, USA.SAS/GRAPHTM is a trademark of SAS InstituteInc.

    27

    The CP Trace Table Recording Facility is a newfeature of VM/SP Release 2. The facility,using th e CPTRAP command, creates a spool fileof selected entries from th e CP trace table, an drecords written throu9h special interfaces tothis facility in both CP an d from virtualmachines. While it is designed primarily as adiagnostic tool to assist Field Engineering andsystem programmers in problem determination,th e contents of th e trace table could be usedfor detailed analysis of specific performanceproblems as well. The trace table recordingfacility J th e CPTRAP command which controlsthe data collection, an d th e CMS TRAPREDcommand to display the data for problemdetermination purposes are documented in th eIBM VM/SP: System Programmer's Guide an d th eIBM VM/SP: Operator's Guide.At th e opposite end of th e spectrum from th etremendous detail available from trace tablerecording are th e global views of the systemrecorded in th e CP accounting data. While th eaccounting records do no t contain enough detailfo r direct use in performance tuning, they doserve as a source of information on averagesystem loads useful fo r projecting futuregrowth. Th e primary source of informationabout CP accounting files is the IBM VM/SP:System Programmer's Guide.A fourth source of performance data are th erecords produced by th e VM Realtime Monitor(an IUP available from IBM, more popularlyknown as SMART). While SMART's emphasis ison realtime performance problem diagnosis, itcan produce automatic hardcopy output ofsummary displays, and can spool a display to avirtual reader fo r offline problem analysis.Because SMART runs in a virtual machine anduses th e DIAGNOSE command to examine realstorage it can collect data no t available in th eCP monitor records.Finally do no t overlook software on the SHAREtape from th e University of Waterloo. Insituations where other data sou rces areinadequate there may be a program on the tapewhich could be used as distributed, or serve asa model fo r locally developed software to meetspecial needs. A good example of softwareavailable from this source is th e CPWATCH realtime monitor.While all of th e above can provide valuable datato ou r performance evaluation activity, we willconcentrate on th e reduction of CP monitordata. CP monitor data contains the informationneeded for normal analysis, an d can becollected easily and with little overhead. Th eraw data are, however, useless without a gooddeal of processing because of th e volume ofdata collected and th e need for combining datafrom several record types across a time intervalin order to produce meaningful information.

  • 7/28/2019 Sugi 83 06 Jobusch

    2/7

    There are a number of important factorsfavoring th e use of SAS as th e tool to choosefo r YM/SP performance a n a ~ y s i s . First, SAS isa very high level language with both th e powerto process th e relatively complicated CP Monitordata formats and a large library of proceduresfo r statistical analysis, graphic presentation ofresults, and time-series forecasting. Second,SAS has become the standard language forperformance analysis of MVS based systems,facilitating th e pooling of information inmixed-SCP environments. And finally,beginning with release 79.6 of CMS SAS, alibrary of VM/SP performance analysis softwareis supplied with th e system.The availability of first level performancemeasu rement software in SAS gives th e VM usercommunity a base from which to build flexibletools fo r detailed performance evaluation andcapacity planning systems tuned to local needs.The high level, self-documenting, nature of th eSAS language offers not only improvedproductivity for the individual performanceanalyst, but also facilitates th e interchange ofprograms among users.Appendix A contains examples of th e reportsavailable from th e programs distributed as partof CMS SAS. These are briefly describedbelow:Th e first report is a simple summary page,produced during the first phase of datareduction by th e MONDATA SAS program. Itdisplays various fields from th e CP Monitorheader records and record counts fo r th edifferent types of records found in the inputfile. Th e second report is produced by PROCMEANS a'nd shows a number of descriptivestatistics for each variable in th e SAS data se tCURR.MONDATA1. These same variables arealso stored in CURR.MONAVGEO, averaged on a10 minute interval throughout the samplingperiod. The MONTREND data library(MONTREND.MEANS, MONTREND.MAXIMA, andMONTREND.MINIMA) contains th e daily means.maxima, and minima fo r this same set ofvariables.

    1. SAS uses th e notation ODname.Membernameto refer to specific SAS data sets, I n th eOS version of SAS, th e DOname refers toth e DO statement defining th e SAS datalibrary, a di rectly accessed OS data set thatmay contain severai logical SAS data sets,or members. In eMS SAS, DDname refer:;to th e CMS filetype of th e files in a SASdata library, and th e Membername is th eCMS filename. For example, SAS wouldrefer to th e eMS file MARY SMITH asSMITH.MARY.

    28

    Th e next report summarizes resource utilizationby userid. Th e information shown in th ereport is saved in th e SAS fileCURR,MONUSER. PROC SORT can be used inconjunction with PROC PRINT to produce thisreport sorted by any variable.Th e report of 1/0 activity in th e system isproduced by a SAS DATA step, MONIO SAS.This information is saved in th e CURR.MONIOdata set.Next is an example of the type ofcross-tabulation which can be generated withtwo SAS statements. This report shows abreakdown of privileged instruction use fo reach userid active during the collection of class5 CP Monitor records.The plots which follow illustrate the type ofoutput one can produce on either line printersor (with SAS/GRAPHUI!) on any of a variety ofpresentation quality graphics devices.The CP Monitor analysis software included withCMS SAS is not intended to be a totalperformance analysis and capacity planningpackage. The SAS prog rams on th edistribution tape are a combination of a completesystem fo r first level analysis and examples ofmore detailed analysis programs. The strengthof this software is its providing ready access toth e full range of standard performance datafrom SAS which allows th e performance analystto develop tools to fi t local needs, merging datafrom a variety of sources to create anintegrated system.

    Because of space limitations this paper is agreatly reduced version of th e materialdistributed at SUGI. To receive a copy of all43 pages of the handout materials contact th eTechnical Support division of SAS Institute.

  • 7/28/2019 Sugi 83 06 Jobusch

    3/7

    g\/I II) e~ " , ""- .

    ! ~ ; C o)C'O .,

    Q.E

  • 7/28/2019 Sugi 83 06 Jobusch

    4/7

    'lM/370 MO,.ITOfI; ('I"TA COLLECTED FQOOO( 2.222 1.396 ... '" 1652 , , '" '" '" , '" 94 " 0 , " 0 o 0.115 XXXlOOlXX '" '" ... '" 1652 , , " '" "' , ,. " , 0 , " 0 o 0.056 X)(J()(XX)()( " " >. 0 '50 1644 , , " '" , 0 " " " , 0 " 0 13 0.231 )()()()oo()(X " " .., 1007 1616 , , '" '" , 0 " " " 0 " " 0 o O.le8 XXXXXXXX ,0> '" .. , '" 2176 " , '" "" " , '" " , 0 , , 0 89 0.88, ~ ~ " " ,. , '" '" , 0 " " , 0 '" " , 0 , 0 0 980 .9610 XXXXXX)()( 2.427 2.109 .. , '" 56211 " " '" "'" " 0 '" " , 0 , 0 74 0.77" 0 0 '" '" 0 0 " " 0 0 '" , 0 0 0 , 0 100 1.00" 0 0 "0 '" 0 0 0 0 0 0 , " 0 0 0 0 0 50 1.00" 0 0 II: 7 ,so ,,, 0 0 0 0 0 0 , " 0 0 0 0 0 50 1.00" , 0 '" "0 0 0 , " 0 0 , " 0 0 0 0 0 88 1.00" 0 0 ,., '" '" 0 0 , , 0 0 , " 0 0 0 0 0 15 1.00" '" '" ,., " " , , " " , 0 '" '" 0 0 , 0 0 980 .98" "' " , . " '" , , " " , 0 '" " , 0 0 , 0 980 .99" " " '.0 " '" , , '" '" , 0 '" " 0 0 , , 0 980.93" " , , . .. '" , , " " , 0 '" " 0 0 , 0 0 99 0.99,,= , , ' - ' ", "0 0 0 so. m , 0 " .. , 0 0 1 0 900.87,,= " 0 ]: 7 '" '" 0 0 30 " 0 0 '" " 0 0 0 0 0 100 1.00,,= " " " '" , 0 " " " 0 '" " 0 0 0 0 0 99 1.00,,= '" " .., " '" " " " " " , '" "0 , 0 0 0 91 0.91211 XXXXXXXX " " '.0 '" 560' , , " '" 0 '" " , 0 , 0 9"; 0.95,,= , . " '.0 ." 1812 , , '" '" " , '" " , 0 , , 0 920.8526 XXXXXXXX " " ,. , '" ,so , 0 " '" 0 " " 0 0 0 " 0 511 0.81127 XXXXXXXX , , ' - ' '" '" , , '" '" , 0 " " 0 0 , 0 93 0.9528 )()(X>QOOO( '" " u '" '" , '" ,,, , 0 '" " 0 0 , , 0 91 0.90" .. "' '.0 ", '" , , " m " 0 3H " , 0 , , 0 970 .96" " " .., '" ,,, , , " " , 0 >0, " , 0 0 , 0 940 .94" " " ,., ". '" , , " '" , 0 '30 " 0 0 , , 0 970 .93" " " .., '" ,,, , '" m , 0 '" " 0 0 , , 0 860 .97'. .. " ... m '" , , " '" , 0 " " , 0 0 , 0 82 0.91" '" " .., " m , , " '" , 0 '" " , 0 , , 0 970 .96" " " ,. , '" ,so , , " ", 0 '" '" , 0 0 , 0 990 .9836 '" '" ... '" '96 , "0 no " '" .. , 0 , , 0 92 0.8831 ". >50 ... '" '" , , " " , 0 '" " , 0 , 0 0 890 .98" " " ' - ' '" 2192 , , '" '06 , , OS, 53 , 0 , , 0 95 0.94" '" '" ,., '" '" " , " '" " , '" " , 0 , 0 67 0.85'10 )()(XXXXXX , '"' " ", 0 0 " " 0 0 '" " 0 0 0 0 0 99 1.0041 XXXlOOlXX ." '" ... '" '" , , " ,,, " " '" " 0 , , 0 85 0.8642 XXXXXXXX " 60 .., ", '" , , '" '" " 0 '" " 0 0 , 1 0 97 0.92, , ~ , , '"' "0 ." 0 0 eo '" " 0 63 " 0 0 0 0 0 98 1.0044 XXXlOOlXX " " ... '" '96 , " '" , 0 '" " 0 , , 0 92 0.894S )()()()()()()(x , , ,., '" '" 0 0 " '" 0 0 '" " 0 0 0 100 0.9646 >oooo(XXX n " ,., m 2008 , , '" 80. , 0 '" " , 0 , , 0 93 0.9041 JOOOOO()()( " " , . '" "0 , " '" , 0 353 " , 0 , , 0 97 0.9148 X)()(XXXXX , ... .0> ,,, 0 0 0 0 , 0 , " 0 0 0 0 0 75 1.0049 XXXXXXXX , , ... '" ." 0 0 >4. '" , 0 '" , 0 0 0 0 100 0.9750 )()()()Q(XXX '" " ... '" .60 " , '" ", " 0 '" " , 0 , , 0 96 0.94, . ~ " " ... ,., ." , '" '" , 0 ". " , 0 1 0 96 0.84XXXXXX)()( .. " ,.. '" 1116 , , '" 00 0 , 0 '" "' 0 0 " 0 92 0.8753 xxxxxxxx " " .. , '" so. , , " n, , , ,n " 0 0 , , 0 96 0.9554 XXXXIOOO( " " , . '" 2504 , '" .0> 0 '" " 0 0 , , 0 91 0.95

    '1M/370 !mITOR DATA COLLECTED FOR 43111-02 SERIAL 999999 FROM 10/11/8209:00:00 THRU 17:00:00 OBS[RVrO "l Of I / O ~ , ceu TYPE SERIAL , ACTV COUNT IlATE ARATE CHAN SYS. BliSY A'IG. MAX HO. PCl. A\lO. "',"0. PO" NO. NOH-O A'IG.120 3370 XXXXXX '" '" 167122 , . , . lCl.l , . 18.0 0 . ' , " , . 0. 0 , , 0. ' 5889 4195 19.5'" 3310 X)()(XXX 111999.8 1075116 .., .. , 12.9 '.0 13.3 o. , , " , . 0. 0 , , 0. ' 362J 2942 15.3122 3170 XXXXXX 117999.6 125051 ,., ,., 15.0 ,., 15.0 o. , , " , .0 0. 0 , , '"' 1112!J 3111 H .8'" 3370 XXXXXX '" '" 129011 '1.5 ,., 15.5 ' - ' H .7 o. , " , . 0. 0 , , '"' 6518 4HO 39.4124 3370 xxxxxx 419 99.6 1171108 .. , ,., 14.1 ,., 111.1 o. , 1 " ,., 0 .0 , , ,., 318J 3029 111.83310 XXXXXX ". '" "1115116 ,., ,., 11.11 .. , 11.9 o. , 30 .. , 0 .0 , , ,., 42:51 31111 24.6126 3370 XXXXXX 479 15.7 37978 ... .., ,. , .. , 0. 0 , , , , 0 .0 , , '"' 1202 '" '.0127 3370 XXXXXX '" , , 11946 " ' - ' 0 .' 0.' o. , 0. 0 0 0 0.0 0 .0 , , ,. ,834176 46.3"0 3375 xxxxxx 117941.2 50831 ... .. , , . ,.. 0. 0 0 0.0 0 .0 0 0 0. 0 1812 .OS ,.,'" 3115 XXXXXX 47996 .5 160295 , . , . 11.7 ,., 16.0 0-.0 " ,. 0 0 .0 0 0 0. 0 6519 1850 )0 . J'" n f5 JOQO\XX 41915.11 141209 ,., .. , 15.6 ... 13.2 0. ' " ,.. 0 .0 0 0 0. 0 2175 "0 ,.,'" 3375 XXXXXX 47999 .0 1291102 ,., '"' 14.3 ,., 11.9 0 .0 , ,., 0. 0 0 0 0 .0 3820 1812 36.1:':24 ]175 XXXXXX 47992 .3 1911012 , . , . 21.11 10.8 15.6 0. 0 , 0. ' 0. 0 0 0 0 .0 7992 2116 :9.1225 3375 lOOlXXX 47998 .5 130107 ,., ... 14.3 ' - ' 12.7 0. 0 , .., 0 .0 0 0 '.0 4676 1958 24 . '2110 3350 xxxxxx '" 0. ' " ' .0 o. , 0. 0 0.0 0 .0 0. 0 0 0. 0 0 .0 , , '" .... 2:11., 3350 xxxxxx ..79 "Z.8 1011"32 ,., ,., 11.5 ,., ,., o. , " ,., 0 .0 , '.0 24"2113 3350 XXxxxx '" 0. ' .. ' .0 0.' 0.0 0 .0 , . 0 0. 0 0 0. 0 0 .0 , '"118 3350 XXXXXX

    '"0. '

    "' .0 0.' 0. 0 0.0 0 .0 0.' 0 0. 0 0 .0 , J,OO 0_A 3350 xxxxxx '" .., " ' .0 0. ' 0. 0 0.0 0 .0 0 .0 0 0.0 0 .0 , 4 57.1 0 .02_8 3350 XXXXXX '" '"' , ' . 0 0 . ' 0. 0 0. 0 0 .0 0 .0 0 0.0 0 .0 , ."H 3H O XXXXXX '" .., """ ,. , '.0 0.' o. , .., 0. 0 0 0. 0 0 .0 , .. 20.0 1i ; 11453 3350 xxxxxx 479 111.4 ". 0. 0 0. ' o. , 0. 0 0. ' 0. 0 0 0.0 0 .0 , , , .55 3350 xxxxxx '" 0. ' " 0. 0 o. , 0.0 0 .0 0.0 0.0 0 0.0 0 .0 , , '" , 0 0 .0901888 50.4

    11'0 32110 479 15.11 59859 ,. , 13.5 100.1 0. 0 0 0. ' 0. 0 , U591110 '.1

    30

  • 7/28/2019 Sugi 83 06 Jobusch

    5/7

    ,",,,A,,,v,A,A",,,,,VA,

    USER I 0

    \/M/370 MONITOR 0A1A COLI EC1EO FOR 11341-02 SERIAL 999999 fROM 10/11/82 lHRU 17:00:00lABlE Of U51::RItI BY OrcOOE

    USER I 0 OPCOOE PRIVilEGED OPI::RA1l0N CODEFREQUENCYIPtRCENl rROW PCT ICOL PeT ISSt( IISK IHOY IS1CKC ISPkA ILPS\/ ITIO IHIO !STell ILCll I

    ~ - i - - - - - - i i - r - - - - - i i - r - - - - - i i - r - - - - - i i - r - - - - - i i - r - - - - - i i - r - - - - - ~ - r - - - - - i i - r - - - - - ; ; - r - - - - - 4 - iI g:gg I g:gg I 8:8g I 8 : ~ g I 8:88 I 4 8 : 8 ~ I 2 8 : 8 ~ I I 28:86 I 28:86 I________ J ___ : ~ ~ _ L __ : ? ~ _ L __ : ~ ~ _ L __ : ~ ~ _ L _ 0.00 .L __ ~ : ~ ~ _ L __ : ? ~ _ ! . __ : ~ . ! 1 . ! _ ! . __ : ~ ~ _ l)()()()()()(X)( t 0 I 0 I 0 I 21 I 0 I 0 I 0 I 0 I 0 I 0 II 8:88 I 8:88 I 8:88 1108:3g I 8:88 I 8:88 I 8:8g I I 8:88 I 8:88 II 0.00 I 0.00 I 0.00 t 100.00 I 0.00 I 0.00 I 0.00 I 0.00 I 0.00 I 0.00 I---------..-------- ...----- ..-_ .._------- ..-------- ...--------"--------.o.--------.o.--------.o.--------.o.--------.o.)()()()O()OO( t 82 4 I 0 lot 0 I 0 I 261 I 0 I 0 I 0 I 0 II 2.37 I 0.00 I 0.00 I 0.00 I 0.00 I 0.75 I 0.00 I 0.00 I 0.00 I 0.00 II 7) .94 I 0.00 I 0.00 I 0.00 I 0.00 I 2 ~ . 0 6 I 0.00 I 0.00 I 0.00 I 0.00 II 3.92 I 0.00 I 0.00 I 0.00 I 0.00 I 6.98 I 0.00 I 0.00 I 0.00 I 0.00 I---------.-------- ...--------.o.--------.o.-------- ...--------+ ________ o. ________+_______ o. ________ ..________+)()()()O()OO( I 1929 I 0 I 1 I 0 I 0 I 0 I 0 I 0 I 2 I 2 II ~ . 5 4 I 0.00 I 0.00 I 0.00 I 0.00 I 0.00 I 0.00 I 0.00 I 0.01 I 0.01 II 9 ' .74 I 0.00 I 0.05 I 0.00 I 0.00 I 0.00 I 0.00 I 0.00 I 0.10 I 0.10 II 9.18 I 0.00 I 3.1l I 0.00 I 0.00 I 0.00 I 0.00 I 0.00 I 0.54 I 0.49 I---------+--------...--------.o.--------.o.-------- ...--------+ ________ o. ________ o. ________ o.________ o.________+)()O(X)O(XX f 122B I 0 I 1 I 0 I 26 I 3 I 609 I 0 I 0 I 0 II 3.53 I 0.00 I 0.00 I 0.00 I 0.07 I 0.01 I 1.15 I 0.00 I 0.00 I 0.00 II 6 ~ J Z I 8:88 I ~ : ~ ~ I 8 : ~ 8 I 1 ~ : ~ ; I 8:6g I l ~ : ~ ~ I I 8:88 J 8:88 I________h _______ ..________ ..________ o. ________ ..________+_______ o. ________ o. ________ o. ________ o.________+XXXXXKXX I 0 I 0 I 0 I 0 I 0 I 0 I 81 I 0 I 0 I 0 II 0.00 I 0.00 1 0.00 I 0.00 I 0.00 I 0.00 I 0.23 I 0.00 I 0.00 I 0.00 II 8:88 I 8:88 I ~ : 8 8 I 8:38 I 8:88 I 8:88 I 8 : ~ g I I 8:g8 I 8:88 I---------.--------.o.-------- ...--------.o.-------- ...--------+________ o. ________ o.________ o.________ o.________+)()()OOOO()( I 74 I , I 0 I 0 I 0 I 307 I 151 I 0 I 1 I 1 II 0.21 I 0.00 I 0.00 I 0.00 I 0.00 I 0.88 I 1.01 I 0.00 I 0.00 I 0.00 II 10.07 I 0 . 1 ~ I 0.00 I 0.00 I 0.00 I 41.71 I 41.76 I 0.00 I 0.14 I 0.14 Ir 0.35 I lO.OO I 0.00 r 0.00 I 0.00 I 8.22 r 3.89 t 0.00 I 0.27 I 0.25 I---------+--------+--------...--------.o.--------.o.--------+________ o.________ o. ________ ..________ o.________+JOOOOOO

  • 7/28/2019 Sugi 83 06 Jobusch

    6/7

    VM/310 M O N I T ~ DATA COLLECTW fOR 1i341-o:i! SERIAL 999999 fROM 10/11/82 09:00:00 THRU 17:()0:0()PERCENTAGE BAR CHARTMIDPOlliTQISEC AVG SECS IN 111+El BETIIEEN D R O ~ : < ;

    D.OOl0.004

    O . O O ~0.0100.0150.025

    o . o ~ o0.0600.1000.150

    o . ~ , o0.4000.6001.0001.')00

    5004.0005.000

    10.00015.000

    II.j I . ... ..1_ . . .. ..., . . . . . . . . . ._ ............... * . . . . . . . . . . . . . . . . . . . . . . .1_ .... .. ..1. ....... ._ ........ .I. .. .1 .-I"""I.,III----+---+---+---+---+---+---+---+---+---+---+---+---6 e 10 14 16 Ie ~ ~

    PERCENTAGE

    fR[Q CUM. PERCENT CUH.ooo

    ,..'"'""""""

    ooooo

    FREQ PERCENTooo

    """,'"'"'".,'"'"7'"'"07 .'"07 .

    0.00O.O()0.000.210.631.889.21

  • 7/28/2019 Sugi 83 06 Jobusch

    7/7

    Number of eMS Users Each Weekday............ 120

    to .

    o.

    o.

    4.2.

    1 2 3 8 7 B 9 10 13 15 18 17 20 21 22 23 24 27 28 29 30 DAVf - - - - - - - - Septellbel" ---------1 MONTH

    LEGEND: USERS ~ " v e r ' a g . ~ T D t a l

    eMS User Load Throughout the Day

    9: 00 11: 00 13: 00 15: 00 17: 00Time of OillY

    LEGEND: USERS -- Act ive -- Logged On33