ExtensityPerformanceBenchmarksExtensityPerformanceBenchmarksExtensityPerformanceBenchmarksExtensityPerformanceBenchmarksMarch2001BenchmarkStudyResultsExtensityPerformanceBenchmarksMarch2001BenchmarkStudyResultsIntroductionAsanantidotetotoday’schallengingeconomicenvironment,manylargecompaniesareturningtosolutionsthatcanbringnewefficienciestointernaloperationsandreducecosts.Thesebusinessrequirementsaregivingrisetoanewcategoryofenterprisesoftware,EmployeeRelationshipManagement(ERM).ERMapplicationstargettheprocessesinvolvedinacquiring,managing,trainingandretainingemployees,withafocusonbringingunprecedentedefficienciestoday-to-dayemployeetasks.SinceERMsolutionstoucheveryemployeeintheorganization,asopposedtomorespecializedERPorCRMsystems,it’sespeciallycriticalthattheseapplicationsbebaseduponafoundationarchitecturethatwillprovidemaximumscalabilitytoalargenumberofgeographicallydistributeduserswithoutcompromisingperformanceorsecurity.AsapioneerinInternet-basedERMapplications,Extensityhasdesigneditsproductsfromthebeginningtoensurethesuccessfuldeploymentoftoday’semployee-facingenterpriseapplications.Extensityapplicationsfeaturea100%Java-basedcomponentarchitecturethatsharesacommoncodebaseacrossallapplicationstoensurethehighestlevelsofperformance,enablingcodetobecompiledonceandthenreusedinmanyplaces.Inaddition,Extensityapplicationscanbedistributedacrossdatabaseservices,applicationserversandclientmachines,withprocessingandlegacysystemintegrationmanagedbytheserverwherecapacitycanbeeasilyexpanded.InMarch2001,ExtensityconductedaseriesofbenchmarkteststodemonstratethestrengthoftheExtensityarchitectureanditsabilitytosupportthescalabilityandperformancerequirementsofGlobal2000companies.Thepurposeofthisdocumentistodescribetheresultsofthisprogram;morespecifically,toshowserverresponsetimesthatresultedfromrunningExtensityapplicationsinacontrolledenvironment.ThisseriesofbenchmarksdemonstratedconsistentscalabilityoftheExtensitysuiteupto800concurrentusers.Usingrealcustomerserverdata,thesebenchmarksshowExtensityscalingtocomfortablysupportaninstalledbaseof100,000users.2BenchmarkMethodologyExtensity’sapproachtobenchmarkingistosimulateactualcustomerusageofthesystem,usinginstrumentedclientandserversoftwareonadepartmentalclassserverarchitecture1thatmimicsthetimingandactivitiesofatypicaluser.Thisseriesofbenchmarkstestsserver-sidescalabilityandserver-to-clientresponsetime,butdoesnotreflectclient-sideprocessingtime.Extensityutilizesatoolthatrecordsandreplaysclient/servercommunicationstosimulatecustomerusage.Ourscriptsexecutethefollowingoperations:TaskDescriptionLoginTimefromExtensityshortcuttologinscreenInboxTimefromLoginprompttoInboxCreateadocumentTimefromInboxtoNewDocumentscreenEditadocumentTimetoretrievesaveddocumentfromInboxinEditmodeSubmitthedocumentTimetosubmitasaveddocumentViewadocumentTimetoretrievesaveddocumentfromInboxinread-onlymodeExtensityexecutedthesescriptsinparalleltomimic300,600and800concurrentusers.Foreachthresholdofuservolume,Extensitysubmittedausertestsequenceconsistingofone15-lineexpensereportfromeachuser,one25-linetimesheetfromeachuser,oracombinationofboth.Thebenchmarkconsistedoffourtests:ApplicationNetworkEnvironmentExtensityExpenseReports™LANExtensityTimesheets™LANExpenseReports,TimesheetsLANExpenseReports,TimesheetsLANvs.dialup(28.8modem,120mslatency)1Accordingtocustomerserver...
