Advanced DBA – Performance Tuning for MUF Part I
-
Upload
ca-technologies -
Category
Technology
-
view
50 -
download
1
Transcript of Advanced DBA – Performance Tuning for MUF Part I
World®’16
AdvancedDBA– PerformanceTuningforMUFPartIKevinShumaVP,ProductManagementCATechnologies
MAINFRAMEANDWORKLOADAUTOMATION
MFX66EA
2 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
Abstract
ThisisatwopartsessionthatdescribesthevarioustechniquesusedtotuneaCADatacomenvironment(MUF)forthebestpossibleperformance.Includedwillbediscussionsonavailablemonitoringtools.
ThesessionwillcovervariousutilitiesandtollsprovidedwithCADatacom®/DBandCASysview®tomeasuretheperformanceofCADatacomenvironments.
KevinShuma
VP,ProductManagement
CATechnologies
3 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
Agenda
ITUNEDLASTYEARWHYDOITAGAIN?
WHATDRIVESTCO?
TECHNIQUESTOMEASUREPERFORMANCE
SUMMARY
1
2
3
44
4 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
ITunedLastYearWhyDoItAgain?
§ Hardwarechanges– IBMz13- TheProcessorUnitchip(PUchip)measures678 mm2
andconsistsof3.99billiontransistorsfabricatedusingIBM's22nmCMOSsilicononinsulatorfabricationprocess,using17metallayersandsupportingspeedsof5.2 GHz,whichislessthanitspredecessor,thezEC12.[3][5]ThePUchipcanhavesix,sevenoreightcores(or"processorunits"inIBM'sparlance)enableddependingonconfiguration.Newforthez13isthatthePUchipcomespackagedinsinglechipmodules,whichisabigchangefromallpreviousmainframeprocessorsthatweremountedonlargemulti-chipmodules.
HowoftendoIneedtotune?
It’sreallyfast
5 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
ITunedLastYearWhyDoItAgain?
§ OperatingSystemchanges– IBM'sz/OS®V2.2operatingsystemandlatestIBM®z13server
deliverinnovationsdesignedtohelpyoubuildthehighlyscalablenext-generationinfrastructureyouneed.Together,theyofferthecapacity,scale,availability,andthroughputrequiredtoimprovebusinessperformance,meetresponsetimeobjectives,protectsensitivedataandtransactions,andminimizeoperationalriskforanexceptionalcustomerexperience.Neweconomicefficienciesallowthez13withz/OSV2.2tooffermorethroughputandcapabilitieswithlessimpacttotheITbudget.
HowoftendoIneedtotune?
It’sreally complex
6 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
ITunedLastYearWhyDoItAgain?
§ DatabaseSoftwarechanges– CADatacomV14.02andv15.0deliveredover50major
enhancementsthatmostlikelywilldirectlyaffecthowyoudatabaseapplicationsrun.
HowoftendoIneedtotune?
It’sreallydifferent
7 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
ThatSoundsReallyScary!
§ Itis,andyetitisn’t
§ Understandingwhatdrivesyourbusinessandhowthatbusinessrelatestothedatabaseapplicationshelpstoestablishthetotalcostofownership(TCO)
§ TCOcanbedifferentforeverysite– Butinmostcases,mainframesitesshareasetoftypicallyTCOdrivers
§ OneconstantismainframeistheneedtodrivedownTCO
HowoftendoIneedtotune?
8 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
WhatdrivesTCO?
9 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
WhatDrivesTCO?
§ HardwareandsoftwarechargesbasedonCPUconsumption– TypicallybasedonpeakCPU(MSUs)useduringameasuredperiod
§ IBMspecialty(zIIP)processorsnotincluded– IBMWorkloadLicenseChargesarebasedon4houraverages
CriticalfactorsformainframeTCO
WhilereducingCPUingeneralisgood,
focusingonthepeakperiodsisbetter
10 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
WhatDrivesTCO?(cont’d)
§ Memorytypicallymeasuredasa“overheadcost”– Costtypicallyspreadacrossalltasks– CanbeusedtoreduceIO,andthereforereduceCPU– Ifoverused,candriveupCPUintheformofsystempaging
§ PhysicalIOtypicallynotmeasuredasa“cost”– ButitconsumesCPUforeachIOused– Notavailablefordispatchingonspecialtyprocessor
CriticalfactorsformainframeTCO
11 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
WhatDrivesTCO?(cont’d)
§ DASDStorage– PhysicalcostsforDASD“onthefloor”– Relativelylow-costresource,but– Theamountofdataisgrowing– Mostshopsareseeinganever-increasingDASDfootprint
CriticalfactorsformainframeTCO
12 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
HowtomeasureTCO?(cont’d)
§ BusinessValueMetrics- CPU– RequestsperCPU
§ 650,000rq/(600CPCPU)=1083rqpertotalCPU
§ AdjustedvaluewithzIIPCPUata%ofCPCPU– BusinessrateszIIPatacostof10%ofCP
§ Requests/(CPCPU+(zIIPCPU*.1))§ 650,000rq/(50+(550*.1))§ 650,000rq/(105)=6190rqperadjustedCPU
MeasuringdatabaseactivityasaTCO
13 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
HowtomeasureTCO?(cont’d)
§ Rememberthisgraph?– ItisnotjustCPUbutCPUuseatthe“highperiods”– zIIPdoesnotcounttowardstheseperiods
MeasuringdatabaseactivityasaTCO
14 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
HowtomeasureTCO?(cont’d)
§ BusinessValueMetrics- IO– RequestsperIO
§ 650,000rq/(60,000IOs)=10.83rqperIO
§ (?)BusinessValueMetrics– DASDinuse– RequestperMB,TB,other?– Notsomethingregularlymeasured– Shoulditbe?
MeasuringdatabaseactivityasaTCO
15 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
What’sinYourToolbox?
§ Trustedandtrue– PXXreport
§ Newandshiny– Dynamicsystemtables
§ Everythingyoucanfind– AutoInfo,AutoStatus,AutoCollect,Autoanything…
Whattoolsareyouusing?Whattoolsareavailable?
16 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
What’sinYourToolbox?
§ Itdoesn’tmatteraslongasitmeetsyourgoals?
§ ImprovedperformanceofCADatacom– Whichshouldresultinreducedresourceconsumption(perrequest)– WhichshouldresultinareductionintheTotalCostofOwnership
Whattoolsareyouusing?Whattoolsareavailable?
17 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
Techniquestomeasureperformance
18 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
Techniquestomeasureperformance
§ Whatresourcesarebeingconsumed
§ Whataretheperformancemetrics
§ Whatarethe“keyconsiderations”– CPUconsumption– Resourceutilization– Responsetime– Locking/Logging/recovery– DASDIOs– More…
Beforeyoucanimprove,youneedtoknowwhattomeasure
19 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
Knowingwhattheenvironmentisdoing
§ ResourceconsumptionsinceMUFcameup– Generalindicationof“health”– Indicationofhigh-useortroublespots
§ Performancebenchmarkorbaseline– Alwaystakesnapshotatthe“same”period– Whenproblemsoccur:
§ Doaquickscanofsnapshotfortroublespots§ Comparecurrentsnapshotagainst“healthysnapshot”
– Lookfordifferences
ThePerformanceSnapshot
20 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
“One-time”performancesnapshot– variousoptions
§ DBUTLTYAUTOINFO(AutoInfo)– Printedreport– Optionforsequentialoutputreadytouploadintospreadsheet
§ ConsolecommandALL_INFO_REPORT– GeneratesoutputtoPXXorPXXSYSOUTdatasets– SimilartoMUFEOJplusPXXSUMMARYreport
Quickandeasy
21 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
“One-time”performancesnapshot– variousoptions(cont’d)
§ MUFEOJSYSPRINT– PrintedtoSYSPRINTatMUFendofjob
§ PXXSYSSTAT=SUMMARY– OndemandrequiresPXXSTATS=DETAIL– PXXSTATS=EOJorEOJPRTonlyavailableatMUFEOJ
22 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
“One-time”performancesnapshot– comparison
23 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
Automatedperformancesnapshotsanddeltas
§ AutoCollect– Automatedsnapshotcollection
§ SimilardataasAutoInfo§ DataplacedinasetofDatacomtables(DBID1019)
– Automatedcreationof“deltas”§ DataplacedinasetofDatacomtables(DBID1020)
– Interval- Differencebetweensnapshots– Last– Differencebetween1stsnapshotandlastsnapshot
DBUTLTYAUTOCOLL
24 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
Automatedperformancesnapshotsanddeltas
§ AutoCollect– On-demandusercreateddeltas
§ Summary– combiningdeltastogether§ Baseline– combinedeltastogetheranddividetogetanaveragebaseline§ Averageperformance– combinedeltasandcalculateaverage“hour”ofperformance
DBUTLTYAUTOCOLL
25 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
SamplesfromAutoCollectusedtomaptopeakperiods
26 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
Interactiveperformancemeasurements
§ CASYSVIEWforCADatacom– DynamicallyfindsallCADatacomMUFs
§ RetrieveskeyinformationontheMUFaddressspaces§ IntegratedwiththeCADatacomDynamicSystemsTables§ Displaysreal-timedataforproblemsolving§ Accesstohistoricalandperformancemetricsfortuning§ QuickestwaytomeasureaperformancechangewhileMUFisactive
– Batchcapabilitiesallowthecollectionofmetrics
Real-timeperformanceinformation
27 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
CASYSVIEWreal-timeperformancemonitors
28 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
SummaryAFewWordstoReview
Results– PartIWeaddressedtheneedtoregularlymonitorperformancetoimproveTCO.Wediscussedvariouson-demandandautomatedcapabilitiestobuildperformancebaselines.Wehighlightedcertainkeymeasurementsthatcanbeusedtodetermineif“deepdive”tuningisneeded.
InpartII,wewillgothroughaseriesofperformancepracticesthatcanbeappliedtoDatacomenvironmentstomeasureperformanceandleadtoimprovedTCO.
@CAWORLD#CAWORLD ©2016CA.AllRIGHTSRESERVED.29 @CAWORLD#CAWORLD ©2016CA.AllRIGHTSRESERVED.29
MainframeandWorkloadAutomation
FormoreinformationonMainframeandWorkloadAutomation,pleasevisit:http://cainc.to/9GQ2JI
30 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
RecommendedSessionsSESSION# TITLE DATE/TIME
MFX92S StrategyandVisionforCADatacomandCAIdeal 11/16/2016at12:45pm
MFX93S What’snewinCADatacom 11/16/2016at1:45pm
MFX94S ReducingTCOwithDatacomCompression(Presspack) 11/16/2016at3:00pm
MFX95SA ModernizingtheCADatacomdata – SQLAccessPartI 11/16/2016at3:45pm
MFX95SB ModernizingtheCADatacomdata – SQLAccessPartII 11/16/2016at4:30pm
MFX96S CADatacom/ADInstallation/UpgradeandMaintenance 11/17/2016at12:45pm
MFX97S CADatacom/ADReportUtilization,andinformationgatherforSupport 11/17/2016at1:45pm
MFX98SUsingSysview tomonitorandmeasureyourCADatacomenvironment– ADandDB
11/17/2016at3:00pm
MFX99S BirdsofaFeather/StumptheTechie! 11/17/2016at3:45pm
31 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
Questions?
32 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
Thankyou.
Stayconnectedatcommunities.ca.com
33 ©2016CA.ALLRIGHTSRESERVED.@CAWORLD#CAWORLD
©2016CA.Allrightsreserved.Alltrademarksreferencedhereinbelongtotheirrespectivecompanies.
Thecontentprovidedinthis CAWorld2016presentationisintendedforinformationalpurposesonlyanddoesnotformanytypeofwarranty. The informationprovidedbyaCApartnerand/orCAcustomerhasnotbeenreviewedforaccuracybyCA.
ForInformationalPurposesOnlyTermsofthisPresentation