基于网格计算的ACM任务调度系统的设计与实现

《基于网格计算的ACM任务调度系统的设计与实现》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、西南交通大学硕士学位论文基于网格计算的ACM任务调度系统的设计与实现姓名：赵惊申请学位级别：硕士专业：计算机应用技术指导教师：黄海于20090501西南交通大学硕士研究生学位论文第ll页Abs仃actACM／ICPC(InternationalCollegiateProgrammingContest)isacontestofprogrammingheldbyACMorganizationyearly．Inordertoholdthiscontest,manyuniversitiesaroundtheworldhavedevelopedtheirownACM／ICPCc

2、ontestplatforms．Mostoftheseplatformscollecttasksviawebservice，andrimthetasksonanindependentserver,thensendtheresultstousersafterrunning．However,becausethesystemcouldonlyrunthetasksonebyone，whenalargeamountofconcurrenttasksaresubmittedtothesystem,itcall’tfinishalltaskintime，thustheusersh

3、avetospendtoomuchtimewaitingfortheresultoftheirtasks．Aimedto．solvetheproblem,thedesignalsocollectstasksviawebservice．Ituses鲥dtasksschedulingtechniqueto0btaillreasonableallocationofallthetaskstomultipleservers，achievingabalancedloadandhighersystemefficiency．Thesystemincludesthreeparts．Th

4、efirstpartisPortalEntrancemodule，thesecondpartisSystemManagementmodule，andthelastpartisComputeResourcemodule，whicharefunctionalindependent．CorefunctionsofthePortalentrancemoduleareforthecollectionoftasksandthepublicationofinformation．CorefunctionsoftheSystemManagementmoduleareforthedete

5、ctionandmanagementofresources，theforecastingandschedulingoftasks，anddatamanagement．TheSystemManagementmodulemainlyinclude：theobtainingofremoteserverload,themaintenanceofmachineavailabletime，theforecastoftaskrun-timeandthecurrentmainstreamgridtaskschedulingalgorithm．CorefunctionsoftheCom

6、puteResourcemodulearefortheacquisitionofthetasks，thejudgingandthefeedbackofthetasks．Thisthesisintroducesthebackgroundsituationandresearchathomeandabroad,andthenintroducestheGridbasicenvironmentonwhichthedesignisbased．Theprocessingoftask,systemarchitecture，modulesandfunctionsofeachmodule

7、aredescribedintheoveralldesignchapter．Inthedetaileddesignchapter,designandimplementationofthesystemandeachfunctionalmoduleare西南交通大学硕士研究生学位论文第llI页described．Inthechapterofperformanceevaluation,thethesisanalysesthedatageneratedbyrunning，evaluateseachtaskschedulingalgorithms，andpro