英语毕业论文成都信息工程学院银杏酒店管理学院本科毕业论文ArchitectureDesignofGlobalDistributedStorageSystemforDataGridLongboRan,HaiJin,ZhipingWang,ChenHuang,YongChen,andYongjieJiaInternetandClusterComputingCenterHuazhongUniversityofScienceandTechnology,Wuhan430074,ChinaEmail:hjin@hust.edu.cnAbstractDatagridsarebecomingincreasinglyimportantforsharinglargedatacollections,archivinganddisseminating.Inthispaperwedescribearchitectureofglobaldistributedstoragesystemfordatagrid.WefocusonthemanagementandthecapabilityforthemaximumusersandmaximumresourcesontheInternet,aswellasperformanceandotherissues.Keywords:Datagrids,Matchtree,Metadata,NamespaceIntroductionData-intensive,high-performancecomputingapplicationsrequiretheefficientmanagementandtransferofterabytesorpetabytesofinformationinwide-area,distributedcomputingenvironmentsREF_Ref34142592\r\h\*MERGEFORMAT[1]REF_Ref34236428\r\h\*MERGEFORMAT[28].Examplesofdata-intensiveapplicationsincludeexperimentalanalysesandsimulationsinseveralscientificdisciplines,suchashigh-energyphysics,climatemodeling,earthquakeengineeringandastronomyREF_Ref34142612\r\h\*MERGEFORMAT[2]REF_Ref34142614\r\h\*MERGEFORMAT[3].Theseapplicationsshareseveralrequirements.Massivedatasetsmustbesharedbyalargecommunityofhundredsorthousandsofusersdistributedaroundtheworld.Datagridsarebecomingincreasinglyimportantforsharinglargedatacollections,archivinganddisseminating.Researchesonmassivestoragesystemhavegainedsignificantachievements.Therearealreadyanumberofstoragesystemsusedbythegridcommunity,eachofwhichwasdesignedtosatisfyspecificneedsandrequirementsforstoring,transferringandaccessinglargedatasets.TheseincludeDistributedParallelStorageSystem(DPSS)andHighPerformanceStorageSystem(HPSS),whichprovidehighperformanceaccesstodataandutilizeparalleldatatransferand/orstripingacrossmultipleserverstoimproveperformanceREF_Ref34142704\r\h\*MERGEFORMAT[4]REF_Ref34236428\r\h\*MERGEFORMAT[28].DistributedFileSystem(DFS)supportshigh-volumeusage,datasetreplicationandlocalca