系统性能测试报告

A시스템성능테스트작업결과보고서PoweredbyLoadRunner™목차1.성능시험의개요2.성능평가기준3.성능시험진행순서4.테스트사이트구성현황5.Workload모델링–Weblog분석6.Workload모델링–CurrentWorkload산정8.Workload모델링–TargetWorkload산정9.Performance모델링–테스트대상단위어플리케이션선정10.Performance모델링–테스트대상혼합BP선정11.Performance모델링–ResourceMonitoring항목선정12.PerformanceTestResults–단위어플리케이션테스트결과13.PerformanceTestResults–단위어플리케이션권고안14.PerformanceTestResults–혼합BP테스트결과및권고안성능시험의개요시험목적시험범위A사이트(EMP와Portal)의오픈전시스템성능측정시스템에서의병목구간점검시스템에서최대수용가능동시단말사용자수산정내부성능시험은예상되는최대수용가능동시단말사용자에대한서버의처리능력검증목적외부성능시험은예상되는최대부하에대한네트워크인프라의처리능력검증목적WebServerApp.ServerDatabaseInternetFirewall성능평가기준Scalability다수의사용자가접속했을때,Performance와functionality를모두만족할것인지?계속증가하는사용자를언제까지대응할수있을까?시스템증설시가장효과적인투자부분은?Infrastructure현재Infrastructure는최적화되어있는지?현재Infrastructure내에병목구간은없는지?ISP와같은외부서비스가약속한성능을지키고있는지?PerformanceApplication혹은Infrastructure의변경이성능에어떠한영향을줄것인지?사용자증가시나타날장애유형과대처방안은?실제운영환경에서사이트의성능은어떨것인지?가상사용자를이용한부하시험사용자증가시나타나는1,2,3차병목추적실제운영환경의성능검증/개선Infrastructure병목의개선을통한최적화성능시험진행순서실제사용자와유사한패턴의부하생성A1)WebLog분석A2)PerformanceModel선정A1)시나리오구현및ScriptingA2)시스템사용현황분석P1.WorkloadModelingP2.PerformanceModelingP3.Scripting&ExecutionA3)CurrentWorkload산정A3)테스트대상BP최종선정A2)ExecutionA3)장애분석/수정,실행결과분석부하의당위성결과의신뢰성검증변경사항발생A1)성능테스트환경구축A4)TargetWorkload산정               테스트사이트구성현황IDSServer(WebMailStorage)UltraSCSIFWDSCSI-/SEInternetDiskStorageDBServerIntegrationServer(XML)FastEthernetNetworkCardUltraSCSIFWDSCSI/SEFastEthernetNetworkCardUltraSCSIFWDSCSI/SEREPxxx.xxx.xxx.100MURCURY1xxx.xxx.xxx.101MURCURY2xxx.xxx.xxx.102FastEthernetNetworkCardUltraSCSIFWDSCSI-2/SEApp.Server)*2100BaseUltraSCSIFWDSCSI-/SEDirectoryFastEthernetNetworkCardUltraSCSIFWDSCSI/SEWebServer(Rep.Srv.)*2UltraSCSIFWDSCSI2/SEFireWallFastEthernetNetworkCard100BaseDMZZONEDummyHub(100M)L4SwitchS/WHUBINTERNETROUTER외부FireWallS/WHUB(100M)WorkloadModelingWorkloadModeling–Weblog분석10월23일자웹서버로그(Murcury1,Murcury2)현황분석WorkloadBasicComponent선정클라이언트에의한동적컨텐츠트랜잭션WorkloadParameter선정시간당동적컨텐츠호출건수(TPH)분당동적컨텐츠호출건수(TPM)Peak치초당동적컨텐츠호출건수(TPS)Peak치동시단말사용자수사용자당평균호출간격시간당동적컨텐츠호출건수분당동적컨텐츠호출건수분당동시단말사용자수동적컨텐츠호출건수대비상위24개의트랜잭션WorkloadModeling–CurrentWorkload산정일일총호출건수(동적컨텐츠)45,575건시간당최대호출건수(TPH)3,460tphPeak시분당평균호출건수(TPM)57tpm(0.95tps)Peak시분당최대호출건수(TPMmax)87tpm(1.45tps)일일총방문자수2,373명Peak시(11:xx)평균동시단말사용자수32명Peak시(11:xx)최대동시단말사용자수36명사용자당평균호출간격32(명)/0.95(tps)=33.7(sec)사용자당평균방문시간T/(총방문자수/평균동시단말사용자수)3600/(196/32)=10분사용자당평균총호출건수총호출건수/총방문자수45,575/2,373=19건10월23일자CurrentWorkloadPeak치(11:xx)시간당동적컨텐츠호출건수–3,460(TPH)Peak치(11:xx)분당평균동적컨텐츠호출건수-57(TPM)Peak치(11:xx)초당평균동적컨텐츠호출건수–0.95(TPS)Peak치(11:xx)동시단말사용자수–32명사용자당평균호출간격–33.7(초)사이트Workload분석지표사용자당평균호출간격은평균응답시간과평균ThinkTime의합입니다.WorkloadModeling–TargetWorkload산정TargetWorkload산정Peak치동시단말사용자수–200명Little’sLaw(동시단말사용자수=TPS*사용자호출간격)적용TPS=200(명)/33.7(초)=5.93(TPS)부하목표인200명에해당되는TPS수치동시단말사용자수는ActiveUser와In-ActiveUser를포함한일반적인의미의동시접속자수입니다.5.93TPS200 PerformanceModelingPerformanceModeling–테스트대상단위어플리케이션선정호출건수를기준으로상위80%를차지하는단위어플리케이션단위어플리케이션의최대성능을구하기위해서ThinkTime을0초로설정하였습니다.PerformanceModeling–혼합BusinessProcess선정사용자의BusinessProcess에따른업무선정BP마다설정된thinkTime30초는사용자당평균호출간격33.7초를환산하여적용하였고실제테스트시에는L/R의“userandompercentageofrecordedthinktime”옵션을사용하여(50%~150%)의범위를가집니다.PerformanceModeling–ResourceMonitoring항목선정ResourceMonitoring항목PerformanceTestResults단위어플리케이션테스트PerformanceTestResults-단위어플리케이션테스트결과단위어플리케이션성능모델도표PerformanceTestResults-단위어플리케이션테스트결과(예상Ratio/최대Throughput의합)