摘要
TheBaBarExperimentcollectedaround20TBofdataduringitsfirst6monthsofrunning.Now,after18months,datasizeexceeds300TB,andaccordingtoprognosis,itisasmallfractionofthesizeofdatacominginthenextfewmonths,Inordertokeepupwiththedatasignificanteffortwasputintotuningthedatabasesystem,Itledtogreatperformanceimprovements,aswellastoinevitablesystemexpansion-450simultaneousprocessingnodesaloneusedfordatareconstruction.Itisbelieved,thatfurthergrowthbeyond600nodeswillhappensoon.Insuchanenvironment,manycomplexoperationsareexecutedsimultaneouslyonhundredsofmachines,puttingahugeloadondataserversandincreasingnetworktrafficIntroducingtwoCORBAservershalvedstartuptime,anddramaticallyoffloadeddatabaseservers:dataserversaswellaslockserversThepaperdescribesdetailsofdesignandimplementationoftwoserversrecentlyintroducedintheBabarsystem:conditionsOIDserverandClusteringServer,Thefirstexperienceofusingtheseserversisdiscussed.AdiscussiononaCollectionServerfordataanalysis,currentlybeingdesignedisincluded.
出版日期
2001年01月11日（中国期刊网平台首次上网日期，不代表论文的发表时间）
