网站性能测试指南(Performance Testing Guidance for Web Appl

1PerformanceTestingGuidanceforWebApplicationsFeedback/Comments:[e-mail]danceforWebApplicationspatterns&practicesJ.D.MeierCarlosFarrePrashantBansodeScottBarberDennisReaInformationinthisdocument,includingURLandotherInternetWebsitereferences,issubjecttochangewithoutnotice.Unlessotherwisenoted,theexamplecompanies,organizations,products,domainnames,e-mailaddresses,logos,people,places,andeventsdepictedhereinarefictitious,andnoassociationwithanyrealcompany,organization,product,domainname,e-mailaddress,logo,person,place,oreventisintendedorshouldbeinferred.Complyingwithallapplicablecopyrightlawsistheresponsibilityoftheuser.Withoutlimitingtherightsundercopyright,nopartofthisdocumentmaybereproduced,storedinorintroducedintoaretrievalsystem,ortransmittedinanyformorbyanymeans(electronic,mechanical,photocopying,recording,orotherwise),orforanypurpose,withouttheexpresswrittenpermissionofMicrosoftCorporation.Microsoftmayhavepatents,patentapplications,trademarks,copyrights,orotherintellectualpropertyrightscoveringsubjectmatterinthisdocument.ExceptasexpresslyprovidedinanywrittenlicenseagreementfromMicrosoft,thefurnishingofthisdocumentdoesnotgiveyouanylicensetothesepatents,trademarks,copyrights,orotherintellectualproperty.©2007MicrosoftCorporation.Allrightsreserved.Microsoft,MS-DOS,Windows,WindowsNT,WindowsServer,ActiveDirectory,MSDN,VisualBasic,VisualC++,VisualC#,VisualStudio,andWin32areeitherregisteredtrademarksortrademarksofMicrosoftCorporationintheUnitedStatesand/orothercountries.Thenamesofactualcompaniesandproductsmentionedhereinmaybethetrademarksoftheirrespectiveowners.Foreword By Alberto Savoia Testingtheperformanceofwebapplicationsiseasy.It’seasytodesignunrealisticscenarios.Easytocollectandmeasureirrelevantperformancedata.And,evenifyoumanagetodesignasoundscenarioandcollecttherightdata,it’seasytousethewrongstatisticalmethodstosummarizeandpresenttheresults.Startinginthelate90s,throughthepeakoftheInternetbubbleandbeyond,Ispentalotoftimetestingtheperformanceofwebapplications.Duringthatperiod,Idesignedandledseveralmission-criticalwebperformanceandloadtestsforhigh-profileInternetcompanies.Workingwiththein-houseperformanceexpertsateachcompanywasveryrevealing–andquitefrightening.Mostoftheengineersassignedtoworkonwebapplicationperformanceweresmart,hard-working,anddedicated;theyinvestedinexpensivesoftwareandhardware,readtherightbooks,andfollowedthebestpracticesoftheday.But,somehow,theresultsoftheirperformancemeasurementsandpredictionsdidnotmatchreality.Insomecasestheperformancetestsoverestimatedtheperformanceandscalabilityofthewebapplications–leadingtoembarrassingandcostlycrasheswhenthewebapplicationwasdeployed.Inothercases,theyunderestimatedcapacityandscalability–leadingtounnecessaryspendingonhardwareandinfrastructure.Theerrorsinthesetestswerenotsmall;sometestsoverestimatedorunderestimatedactualperformanceandcapacitybyanorderofmagnitudeormore!Howisthispossible?Basedonmyexperience,themajorityofgrosserrorsinwebapplicationperformancetestingaretheresultofoversimplification.Moreprecisely,theyaretheresultoversimplificationofuserbehaviorandoversimplificationinsummarizingandreportingtestresults.Imagineatransportationengineerestimatingtrafficpatternsforaproposedstretchofhighwaybyassumingthatmostdriverswilldriveatthesameaveragespeed,breakandacceleratewiththesameresponsetimeandatthesamerate,andneverchangelanes.Asimple–butcompletelyworthless–scenario.Orimaginethesametransportationengineerreportingthattherearenotrafficissuesbecausetheaveragespeedis57mph–withoutbringingupthatduringrush-hourtheaveragespeedis25mph.Asimple,butverymisleading,result.Unfortunately,mostwebapplicationperformancetesterscommiterrorsofoversimplificationasbad,orworse,astheonescommittedbyourhypotheticaltransportationengineer.Iamallforsimplicitybut,asAlbertEinsteinoncesaid:“Makeeverythingassimpleaspossible,butnotsimpler.”Whenitcomestotestingtheperformanceofwebapplications,that’sexactlywhatthisremarkable–andmuchneeded–bookteachesyou.Theauthorsleveragetheirpassion,experience,andhard-earnedknowledgeandprovideyouwiththebroad,thorough,andextensiblefoundationyouneedtotacklewebperformancetestingtherightway.PerformanceTestingGuidanceforWebApplicationsdoesnotgetboggeddownwithunnecessarydetails,butitdoesmakesurethatyouknowabout–anddon’toverlook–thekeyparametersandvariablesthatyouneedtotakeintoaccountindesigning,conducting,andanalyzingyourtests.Ifyouarenewtowebperformancetesting,thisbookwillgetyoustartedontherightpathandsaveyoualotoftimeandembarrassment.Evenifyouareaseasonedwebperformancetestingveteran,Iamconfidentthatthisbookwillprovideyouwithnewinsightsand,mostlikely,haveyouslapyourforeheadafewtimesasyoureadaboutsomecommonandfamiliarmistakes.Ineithercase,PerformanceTestingGuidanceforWebApplications,isamust-haveforanywebperformanceengineerbookshelf.AlbertoSavoiaFounderandCTO,AgitarSoftwareInc.July,2007Authorof:“TheScienceandArtofWebSiteLoadTesting”,“WebLoadTestPlanning”,and“TradeSecretsfromaWebTestingExpert”.Foreword By Rico Mariani It’shardtoimagi