Você está na página 1de 4

07/10/2015

ETLTestScenariosandTestCases

21stNovember2012 ETLTestScenariosand
TestCases
BasedonmyexperienceIpreparedmaximumtestscenariosand
testcasestovalidatetheETLprocess.Iwillkeeponupdatethis
content.Thanks
TestScenario

TestCases

Mappingdocvalidation

VerifymappingdocwhethercorrespondingETLinformationisprovidedor
not.Changelogshouldmaintainineverymappingdoc.
DefinethedefaultteststrategyIfmappingdocsaremissedoutsome
optionalinformation.Ex:datatypeslengthetc

Structurevalidation

1.Validatethesourceandtargettablestructureagainstcorresponding
mappingdoc.
2.SourcedatatypeandTargetdatatypeshouldbesame.
3.Lengthofdatatypesinbothsourceandtargetshouldbeequal.
4.Verifythatdatafieldtypesandformatsarespecified
5.Sourcedatatypelengthshouldnotlessthanthetargetdatatypelength.
6.Validatethenameofcolumnsintableagainstmappingdoc.

ConstraintValidation

Ensuretheconstraintsaredefinedforspecifictableasexpected.

DataConsistencyIssues

1.Thedatatypeandlengthforaparticularattributemayvaryinfilesor
tablesthoughthesemanticdefinitionisthesame.
Example:Accountnumbermaybedefinedas:Number(9)inonefield
ortableandVarchar2(11)inanothertable

2.MisuseofIntegrityConstraints:Whenreferentialintegrityconstraints
aremisused,foreignkeyvaluesmaybeleftdanglingor
inadvertentlydeleted.
Example:Anaccountrecordismissingbutdependentrecordsarenot
deleted.

DataCompletenessIssues

Ensuresthatallexpecteddataisloadedintotargettable
1.Comparerecordscountsbetweensourceandtarget.Checkforany
rejectedrecords.
2.CheckDatashouldnotbetruncatedinthecolumnoftargettable.
3.Checkboundaryvalueanalysis(ex:only>=2008yeardatahastoload
intothetarget)
4.Comparinguniquevaluesofkeyfieldsbetweensourcedataanddata
loadedtothewarehouse.Thisisavaluabletechniquethatpointsouta
varietyofpossibledataerrorswithoutdoingafullvalidationonallfields.

DataCorrectnessIssues

1.Datathatismisspelledorinaccuratelyrecorded.
2.Null,nonunique,oroutofrangedatamaybestoredwhentheintegrity
constraintsaredisabled.

http://testingdwh.blogspot.in/2012/11/etltestscenariosandtestcases.html

1/4

07/10/2015

ETLTestScenariosandTestCases

Example:Theprimarykeyconstraintisdisabledduringanimport
function.Dataisenteredintotheexistingdatawithnulluniqueidentifiers.

DataTransformation

1.Createaspreadsheetofscenariosofinputdataandexpectedresults
andvalidatethesewiththebusinesscustomer.Thisisanexcellent
requirementselicitationstepduringdesignandcouldalsobeusedaspart
oftesting.
2.Createtestdatathatincludesallscenarios.UtilizeanETLdeveloperto
automatetheentireprocessofpopulatingdatasetswiththescenario
spreadsheettopermitversatilityandmobilityforthereasonthat
scenariosarelikelytochange.
3.Utilizedataprofilingresultstocomparerangeandsubmissionofvalues
ineachfieldbetweentargetandsourcedata.
4.ValidateaccurateprocessingofETLgeneratedfieldsforexample,
surrogatekeys.
5.Validatethatthedatatypeswithinthewarehousearethesameaswas
specifiedinthedatamodelordesign.
6.Createdatascenariosbetweentablesthattestreferentialintegrity.
7.Validateparenttochildrelationshipsinthedata.Createdatascenarios
thattestthemanagementoforphanedchildrecords.

DataQuality

1.Numbercheck:ifinthesourceformatofnumberingthecolumnsareas
xx_30butifthetargetisonly30thenithastoloadnotpre_fix(xx_).we
needtovalidate.
2.DateCheck:TheyhavetofollowDateformatanditshouldbesame
acrossalltherecords.Standardformat:yyyymmddetc..
3.PrecisionCheck:Precisionvalueshoulddisplayasexpectedinthetarget
table.
Example:Insource19.123456butinthetargetitshoulddisplayas19.123
orroundof20.
4.DataCheck:Basedonbusinesslogic,fewrecordwhichdoesnotmeet
certaincriteriashouldbefilteredout.
Example:onlyrecordwhosedate_sid>=2008andGLAccount!=CM001
shouldonlyloadinthetargettable.
5.NullCheck:FewcolumnsshoulddisplayNullbasedonbusiness
requirement.
Example:TerminationDatecolumnshoulddisplaynullunless&untilifhis
ActivestatusColumnisTorDeceased.
Note:Datacleannesswillbedecidedduringdesignphaseonly.

NullValidation

Verifythenullvalueswhere"NotNull"specifiedforspecifiedcolumn.

Duplicatecheck

1.Needstovalidatetheuniquekey,primarykeyandanyothercolumn
shouldbeuniqueasperthebusinessrequirementsarehavingany
duplicaterows.
2.Checkifanyduplicatevaluesexistinanycolumnwhichisextracting
frommultiplecolumnsinsourceandcombiningintoonecolumn.
3.Sometimeaspertheclientrequirementsweneedsensurethatno

http://testingdwh.blogspot.in/2012/11/etltestscenariosandtestcases.html

2/4

07/10/2015

ETLTestScenariosandTestCases

duplicatesincombinationofmultiplecolumnswithintargetonly.
Example:Onepolicyholdercantakemultiplepolicesandmultiple
claims.InthiscaseweneedtoverifytheCLAIM_NO,CLAIMANT_NO,
COVEREGE_NAME,EXPOSURE_TYPE,EXPOSURE_OPEN_DATE,
EXPOSURE_CLOSED_DATE,EXPOSURE_STATUS,PAYMENT

DATEValidation

DatevaluesareusingmanyareasinETLdevelopmentfor:
1.Toknowtherowcreationdateex:CRT_TS
2.IdentifyactiverecordsaspertheETLdevelopmentperspectiveEx:
VLD_FROM,VLD_TO
3.IdentifyactiverecordsasperthebusinessrequirementsperspectiveEx:
CLM_EFCTV_T_TS,CLM_EFCTV_FROM_TS
4.Sometimesbasedonthedatevaluestheupdatesandinsertsare
generated.
PossibleTestscenariostovalidatetheDatevalues:
a.From_DateshouldnotgreaterthanTo_Date
b.Formatofdatevaluesshouldbeproper.
c.Datevaluesshouldnotanyjunkvaluesornullvalues

CompleteDataValidation
(usingminusand
intersect)

1.Tovalidatethecompletedatasetinsourceandtargettableminusquery
isbestsolution.
2.Weneedtosourceminustargetandtargetminussource.
3.Ifminusqueryreturnsanyvaluethoseshouldbeconsideredas
mismatchingrows.
4.Andalsoweneedstomatchingrowsamongsourceandtargetusing
Intersectstatement.
5.Thecountreturnedbyintersectshouldmatchwithindividualcountsof
sourceandtargettables.
6.Ifminusqueryreturnsorowsandcountintersectislessthansource
countortargettablecountthenwecanconsideredasduplicaterowsare
exists.

SomeUsefultestscenarios

1.Verifythatextractionprocessdidnotextractduplicatedatafromthe
source(usuallythishappensinrepeatableprocesseswhereatpointzero
weneedtoextractalldatafromthesourcefile,buttheduringthenext
intervalsweonlyneedtocapturethemodified,andnewrows.)
2.TheQAteamwillmaintainasetofSQLstatementsthatare
automaticallyrunatthisstagetovalidatethatnoduplicatedatahavebeen
extractedfromthesourcesystems.

Datacleanness

Unnecessarycolumnsshouldbedeletedbeforeloadingintothestaging
area.
Example2:Ifacolumnhavenamebutitistakingextraspace,wehaveto
trimspacesobeforeloadinginthestagingareawiththehelpof
expressiontransformationspacewillbetrimmed.
Example1:SupposetelephonenumberandSTDcodeindifferentcolumns
andrequirementsaysitshouldbeinonecolumnthenwiththehelpof

http://testingdwh.blogspot.in/2012/11/etltestscenariosandtestcases.html

3/4

07/10/2015

ETLTestScenariosandTestCases

expressiontransformationwewillconcatenatethevaluesinonecolumn.

Posted21stNovember2012bysatishkumar
0 Addacomment

Enteryourcomment...

Commentas:

Publish

Unknown(Google)

Preview

http://testingdwh.blogspot.in/2012/11/etltestscenariosandtestcases.html

Signout

Notifyme

4/4

Você também pode gostar