Sei sulla pagina 1di 3

5/31/2016

CompletereferencetoInformatica:OptimizingtheBottlenecks

Complet referenc t Informatic


Home

Data WareHousing

INTRODUCTION
ETLLifeCycle
WhatIsInformatica
ClientComponents
ServicesBehindScene
TryURHand'sonAdmin
Console
DifferenceBetween7.1
and8.6
Informatica8.6
Installation

ADVANCED CONCEPTS
MappingParameter's&
Variable
Mapplets
Partitioning

Unix-Shell Scripts

PL SQL

Contact Us
PERFORMANCE TUNING

Sunday,17April2011

OptimizingtheBottlenecks

NecessityOfPerformancetuning

1.Ifthesourceisaflatfile,ensurethattheflatfileislocaltotheInformaticaserver.If
sourceisarelationaltable,thentrynottousesynonymsoraliases.

IdentificationOfbottleNecks
OptimizationofBottlenecks
PerformancetuningofLookup
PushDownOptimization

2.Ifthesourceisaflatfile,reducethenumberofbytes(Bydefaultitis1024bytes
perline)theInformaticareadsperline.Ifwedothis,wecandecreasetheLine
SequentialBufferLengthsettingofthesessionproperties.
3.Ifpossible,giveaconditionalqueryinthesourcequalifiersothattherecordsare
filteredoffassoonaspossibleintheprocess.
4.Inthesourcequalifier,ifthequeryhasORDERBYorGROUPBY,thencreatean
indexonthesourcetableandorderbytheindexfieldofthesourcetable.
PERFORMANCETUNINGOFTARGETS
Ifthetargetisaflatfile,ensurethattheflatfileislocaltotheInformaticaserver.Iftarget

TESTING
UnitTesting
IntegrationTesting
UAT
InformaticaTesting
Debugger
ConstraintBasedloading
TargetLoadPlan

isarelationaltable,thentrynottousesynonymsoraliases.

INTERVIEW ZONE

Workingwithlinks

1.Usebulkloadwheneverpossible.

Scheduler

2.Increasethecommitlevel.

InformaticaRealTimeInterviewQuestions

3.Dropconstraintsandindexesofthetablebeforeloading.

InformaticaExperiencedInterview
Questionspart1

TypesofTask's1
TypesofTask's2

PERFORMANCETUNINGOFMAPPINGS

IndirectMethodfor
Loading

Mappinghelpstochanneltheflowofdatafromsourcetotargetwithallthetransformations

SCDType1

inbetween.MappingistheskeletonofInformaticaloadingprocess.

SCDType2

1.Avoidexecutingmajorsqlqueriesfrommappletsormappings.

SCDType3

2.Useoptimizedquerieswhenweareusingthem.

IncrementalAggregation

3.Reducethenumberoftransformationsinthemapping.Activetransformationslike

MappingTemplates
GridProcessing
FrequentlyUsed
Functions
WorkFlowVariables

TRANSFORMATION
Filter
Expression
Router
Sorter
Rank
TransactionControl
SourceQualifier
StoredProcedure
SQLTransformation
Normalizer
SequenceGenerator
Aggregator
Union
Joiner
UpdateStrategy
LookUp

rank,joiner,filter,aggregatoretcshouldbeusedaslessaspossible.

InformaticaExperiencedInterview
Questionspart2
InformaticaExperiencedInterview
Questionspart3
InformaticaExperiencedInterview
Questionspart4
DataWarehousingconceptBasedInterview
Questions

4.Removealltheunnecessarylinksbetweenthetransformationsfrommapping.
5.Ifasinglemappingcontainsmanytargets,thendividingthemintoseparate

17

mappingscanimproveperformance.
6.Ifweneedtouseasinglesourcemorethanonceinamapping,thenkeeponlyone
sourceandsourcequalifierinthemapping.Thencreatedifferentdataflowsas

TOTAL VISITS

requiredintodifferenttargetsorsametarget.
7.Ifasessionjoinsmanysourcetablesinonesourcequalifier,thenanoptimizing
querywillimproveperformance.
8.InthesqlquerythatInformaticagenerates,ORDERBYwillbepresent.Removethe
ORDERBYclauseifnotneededoratleastreducethenumberofcolumnnamesin
thatlist.Forbetterperformanceitisbesttoorderbytheindexfieldofthattable.
9.Combinethemappingsthatusesamesetofsourcedata.

FEEDS

10.Onamapping,fieldwiththesameinformationshouldbegiventhesametypeand
lengththroughoutthemapping.Otherwisetimewillbespentonfieldconversions.
11.Insteadofdoingcomplexcalculationinquery,useanexpressiontransformerand
dothecalculationinthemapping.
12.Ifdataispassingthroughmultiplestagingareas,removingthestagingareawill
increaseperformance.
13.Storedproceduresreduceperformance.Trytokeepthestoredproceduressimplein

DISCUSSION FORUM
DiscussionForum

SUBSCRIBE TO LEARN INFORMATICA


Like

Share 512people
likethis.

themappings.
14.Unnecessarydatatypeconversionsshouldbeavoidedsincethedatatype
conversionsimpactperformance.
15.Transformationerrorsresultinperformancedegradation.Tryrunningthemapping
afterremovingalltransformations.Ifitistakingsignificantlylesstimethanwiththe
transformations,thenwehavetofinetunethetransformation.

http://informaticatutorialsnaveen.blogspot.in/2011/04/optimizingbottlenecks.html

1/3

5/31/2016

CompletereferencetoInformatica:OptimizingtheBottlenecks
16.Keepdatabaseinteractionsaslessaspossible.

Enteryouremailaddress:

PERFORMANCETUNINGOFSESSIONS
Asessionspecifiesthelocationfromwherethedataistobetaken,wherethe
transformationsaredoneandwherethedataistobeloaded.Ithasvariouspropertiesthat
helpustoscheduleandrunthejobinthewaywewant.

Subscribe
DeliveredbyFeedBurner

1.Partitionthesession:Thiscreatesmanyconnectionstothesourceandtarget,
andloadsdatainparallelpipelines.Eachpipelinewillbeindependentoftheother.
Buttheperformanceofthesessionwillnotimproveifthenumberofrecordsisless.
Alsotheperformancewillnotimproveifitdoesupdatesanddeletes.Sosession
partitioningshouldbeusedonlyifthevolumeofdataishugeandthejobismainly
insertionofdata.
2.Runthesessionsinparallelratherthanserialtogaintime,iftheyareindependentof
eachother.
3.Dropconstraintsandindexesbeforewerunsession.Rebuildthemafterthesession
runcompletes.DroppingcanbedoneinpresessionscriptandRebuildinginpost
sessionscript.Butifdataistoomuch,droppingindexesandthenrebuildingthem
etc.willbenotpossible.Insuchcases,stagealldata,precreatetheindex,usea
transportabletablespaceandthenloadintodatabase.
4.Usebulkloading,externalloadingetc.Bulkloadingcanbeusedonlyifthetable
doesnothaveanindex.
5.InasessionwehaveoptionstoTreatrowsasDataDriven,Insert,Updateand
Delete.Ifupdatestrategiesareused,thenwehavetokeepitasDataDriven.But
whenthesessiondoesonlyinsertionofrowsintotargettable,ithastobekeptas
Inserttoimproveperformance.
6.Increasethedatabasecommitlevel(ThepointatwhichtheInformaticaserverisset
tocommitdatatothetargettable.Fore.g.commitlevelcanbesetateveryevery
50,000records)
7.Byavoidingbuiltinfunctionsasmuchaspossible,wecanimprovethe
performance.E.g.Forconcatenation,theoperator||isfasterthanthefunction
CONCAT().Souseoperatorsinsteadoffunctions,wherepossible.Thefunctions
likeIS_SPACES(),IS_NUMBER(),IFF(),DECODE()etc.reducetheperformance
toabigextentinthisorder.Preferenceshouldbeintheoppositeorder.
8.Stringfunctionslikesubstring,ltrim,andrtrimreducetheperformance.Inthe
sources,usedelimitedstringsincasethesourceflatfilesorusevarchardatatype.
9.ManipulatinghighprecisiondatatypeswillslowdownInformaticaserver.Sodisable
highprecision.
10.Localizeallsourceandtargettables,storedprocedures,views,sequencesetc.Try
nottoconnectacrosssynonyms.Synonymsandaliasesslowdownthe
performance.
DATABASEOPTIMISATION
TogainthebestInformaticaperformance,thedatabasetables,storedproceduresand
queriesusedinInformaticashouldbetunedwell.
1.Ifthesourceandtargetareflatfiles,thentheyshouldbepresentinthesystemin
whichtheInformaticaserverispresent.
2.Increasethenetworkpacketsize.
3.TheperformanceoftheInformaticaserverisrelatedtonetworkconnections.Data
generallymovesacrossanetworkatlessthan1MBpersecond,whereasalocal
diskmovesdatafivetotwentytimesfaster.Thusnetworkconnectionsoftenaffect
onsessionperformance.Soavoidnetworkconnections.
4.Optimizetargetdatabases.
PostedbyNaveenatSunday,April17,2011

Nocomments:
PostaComment

http://informaticatutorialsnaveen.blogspot.in/2011/04/optimizingbottlenecks.html

2/3

5/31/2016

CompletereferencetoInformatica:OptimizingtheBottlenecks

Enteryourcomment...

Commentas:

Publish

GoogleAccount

Preview

Linkstothispost
CreateaLink

http://informaticatutorialsnaveen.blogspot.in/2011/04/optimizingbottlenecks.html

3/3

Potrebbero piacerti anche