简介:ThechallengeofLHCcomputing,withdataratesintherangeofseveralPB/year,requiresthedevelopmentofGRIDtechnologies,tooptimizetheexploitationofdistributedcomputingpowerandtheauthomaticaccesstodistributeddatastorage.IntheframeworkoftheEU-DataGridproject,theALICEexperimentisoneoftheselectedtestapplicationsfortheearlydevelopmentandimplementationofGRIDServices.Presently,about15ALICEsitesaremakinuseofavailableGRIDtoolsandalargescaletestproductioninvolving9ofthemwascarriedoutwithoursimulationprogram.Resultsarediscussedindetail,aswellasfutureplans.
简介:Since1998,theALICEexperimentandtheCERN/ITdivisionhavejointlyexecutedseverallarge-scalehighthroughputdistributedcomputingexercises:theALICEdatachallenges.ThegoalsoftheseregularexercisesaretotesthardwareandsoftwarecomponentsofthedataacqusitionandcomputingsystemsinrealisticconditionsandtoexecuteanearlyintegrationoftheoverallALICEcomputinginfrastructure.ThispaperreportsonthethirdALICEDataChallenge(ADCIII)thathasbeenperformedatCERNfromJanuarytoMarch2001.ThedatausedduringtheADCⅢaresimulatedphysicsrawdataoftheALICETPC,producedwiththeALICEsimulationprogramAliRoot.ThedataacquisitionwasbasedontheALICEonlineframeworkcalledtheALICEDataAcquisitionTestEnvironment(DATE)system.Thedataaftereventbuilding,werethenformattedwiththeROOTI/OpackageandadatacataloguebasedonMySQlwasestablished.TheMassStorageSystemusedduringADCIIIisCASTOR.Differentsoftwaretoolshavebeenusedtomonitortheperformances,DATEhasdemonstratedperformancesofmorethan500MByte/s.Anaggregatedatathroughputof85MByte/swassutainedinCASTORoverseveraldays.Thetotalcollecteddataamountsto100TBytesin100.00files.
简介:导读:(赵伐,浙江外国语学院教授)本文系加拿大著名文学评论家戴维·斯特恩斯(DavidStaines,1946—)2014年2月7日在德国小城特里尔(Trier)举办的门罗小说研讨会上的演讲稿。斯特恩斯是门罗近40年的挚友,两人常一起评论作品,畅谈文学,曾作为加拿大吉勒文学奖(GillerPrize)的评委,一起遴选加拿大文学创作领域的年度佳作,还一起主编了集加拿大文学经典之大成的《新加拿大文库》(TheNewCanadianLibrary),并为许多经典作品亲笔撰写后记。可以说,斯特恩斯对门罗的创作心路了然于胸,心有灵犀。
简介:TheALICEexperiment[1]attheLargeHadronCollider(LHC)atCERNwilldetectupto20,000particlesinasinglePb-Pbeventresultinginadatarateof-75MByte/event,Theeventrateislimitedbythebandwidthofthedatastoragesystem.Higherratesarepossiblebyselectinginterestingeventsandsubevents(HighLeveltrigger)orcompressingthedataefficientlywithmodelingtechniques.Bothrequireafastparallelpatternrecognition.OnepossiblesolutiontoprocessthedetectordataatsuchratesisafarmofclusteredSMPnodes,basedonoff-the-shelfPCs,andconnectedbyahighbandwidt,lowlatencynetwork.
简介:TheALICETriggerandDataAcquisition(TRG/DAQ)Systemisrequiredtosupportanaggregateeventbuildingbandwidthofupto4GByte/sandastoragecapabilityofupto1.25GByte/stomassstorage.Thesystemhasbeendecomposedinasetofhardwareandsoftwarecomponentsandprototypesofthesecomponentsarebeingdeveloped.Itisnecessarytoveritythesystemdesign,itscapabilitytoreachtheexpectedbehaviorandthetargetperformances,discoverpossiblebottlenecksandwaystocorrectforthem,andexplorealternativealgorithmsandnewarchitectures.ToachievethisthecompleteTRG/DAQsystemhasbeenformallyspecified.andtheverificationoftheexpectedbehaviorhasbeenperformedthroughtheexecutionofthespecification,Twotoolswereusedforthis.Foresight,andPtolemy.