Domov Technika distribuováno

distribuováno



Overview

Whencomputerprogramsanddataaredistributedonmorethanonecomputerthroughanetwork,computingbecomes"distributed."Thepreviouscalculationswereusuallydonebythecomputingcenter.Althoughtherearestillsuchcomputingcenters,largeandsmallenterprisesaregraduallymoving(distributing)applicationstothemostefficientlocationintheenterprise,usuallydesktopworkstations,localareanetworkservers,departmentalservers,Webservers,andotherservers.mix.Themorepopularistheclient/servermodel.Theclientonlyhascertainfunctions,andotherfunctionsneedtobeobtainedfromtheserverthatprovidestheservice.TheHTTPprotocolisanexample.Inadistributedcomputingenvironment,datastorageandprocessingcanbeperformedonlocalworkstations.

Distributedcomputing

Distributedcomputingisanewcomputingmethodproposedinrecentyears.Theso-calleddistributedcomputingisthesharingofinformationbetweentwoormoresoftware.Thesesoftwarecanrunonthesamecomputeroronmultiplecomputersconnectedthroughanetwork.Distributedcomputingisacomputingmethod,asopposedtocentralizedcomputing.Withthedevelopmentofcomputingtechnology,someapplicationsrequirehugecomputingpowertocomplete.Forexample,centralizedcomputingrequiresalongtimetocomplete.Distributedcomputingdecomposestheapplicationintomanysmallparts,whicharedistributedtomultiplecomputersforprocessing.Thiscansavetheoverallcalculationtimeandgreatlyimprovethecalculationefficiency.

Comparedwithotheralgorithms,distributedcomputinghasthefollowingadvantages:

1.Vzácné zdrojelze sdílet;

2.DistributedcomputingcanbeusedinmanyBalancethecomputingloadonacomputer;

3.Program můžete zadat na počítač, který je pro jeho spuštění nejvhodnější;

Amongthem,sharingrareresourcesandbalancingtheloadarecomputerdistributedcomputingOneofthecoreideas.

Infact,gridcomputingisatypeofdistributedcomputing.Ifwesaythatacertainjobisdistributed,thenitmustbenotjustacomputer,butacomputernetworkthatparticipatesinthisjob.Obviously,this"antmovingmountain"approachwillhavestrongdataprocessingcapabilities.Theessenceofgridcomputingistocombineandshareresourcesandensuresystemsecurity.

Metoda distribuovaného počítače

DistributedcomputingisakindofcomputingthatusestheidleprocessingpoweroftheCPUofthecomputerontheInternettosolvelarge-scalecomputingproblemsscience.Below,let’sseehowitworks:

Firstofall,wemustdiscoveraproblemthatrequiresahugeamountofcomputingpowertosolve.Suchproblemsaregenerallyinterdisciplinary,extremelychallenging,andurgentresearchtopicsforhumanbeings.Themorefamousonesare:

1. Vyřešte složitější matematické problémy, jako je GIMPS (Najít největší Mersenneprime).

2. Prozkoumejte a najděte nejbezpečnější kryptografický systém, například: RC-72 (prolomení hesla).

3.Biologický patologický výzkum, např.:Folding@home (pro studium skládání proteinů, nedorozumění, polymerace a souvisejících nemocí způsobených tímto).

4.Hledání léků na různé nemoci, například:UnitedDevices (hledání účinných léků proti rakovině).

5.Zpracování signálu, například: SETI@Home (hledání mimozemské civilizace doma).

Fromthesepracticalexamples,itcanbeseenthattheseprojectsareverylargeandrequireanastonishingamountofcalculation.Itisabsolutelyimpossibleforasinglecomputerorindividualtocompletethecalculationinanacceptabletime.possible.Inthepast,theseproblemsshouldbesolvedbysupercomputers.However,thecostandmaintenanceofsupercomputersareveryexpensive,whichisnotwhatanordinaryscientificresearchorganizationcanafford.Withthedevelopmentofscience,acheap,efficient,andeasy-to-maintaincalculationmethodhasemerged-distributedcomputing!

DistributedcomputingandthepopularizationoftheInternet

Withthepopularizationofcomputers,personalcomputershavebeguntoentermillionsofhouseholds.Accompanyingitistheuseofcomputers.Moreandmorecomputersareinanidlestate,andthepotentialoftheCPUisfarfrombeingfullyutilizedevenwhenitisturnedon.Wecanimaginethatahomecomputerspendsmostofitstime"waiting".Evenwhentheuseractuallyusestheircomputer,theprocessorisstillsilentconsumption,andtherearestillcountlesswaiting(waitingforinput,butactuallydoingnothing).TheemergenceoftheInternethasmadeitarealitytoconnectandcallallthesecomputersystemswithidlecomputingresources.

DistributedComputingProject

So,someproblemsthatareverycomplexinthemselvesbutsuitablefordividingintoalargenumberofsmallercomputingfragmentsareproposed,andthensomeresearchTheorganizationhasdevelopedacomputingserverandclientthroughalotofhardwork.Theserverisresponsiblefordividingthecalculationproblemintomanysmallcalculationparts,andthenassigningthesepartstomanycomputersconnectedtothecalculationforparallelprocessing,andfinallycombiningthesecalculationresultstoobtainthefinalresult.

Ofcourse,thisseemsprimitiveanddifficult,butasthenumberofparticipantsandthenumberofcomputersinvolvedinthecalculationcontinuestoincrease,thecalculationplanhasbecomeveryrapid,andithasbeenproventobefeasibleinpractice.of.Theprocessingpowerofsomelargerdistributedcomputingprojectscanreachorevenexceedthefastestsupercomputersintheworld.

YoucanalsochoosetoparticipateincertainprojectstodonatetheCPUcoreprocessingtime.YouwillfindthattheCPUcoreprocessingtimeyouprovidewillappearintheproject'scontributionstatistics.Youcancompetewithotherparticipantsfortherankingofcontributiontime,youcanalsojoinanexistingcomputinggrouporformacomputinggroupyourself.Thismethodisveryhelpfultomobilizetheenthusiasmoftheparticipants.

Participateincomputing

Asthenumberofprivateteamsgraduallyincreased,manylargeorganizations(suchascompanies,schools,andvariouswebsites)havealsobeguntoformtheirownteams.Atthesametime,alargenumberofcommunitieswiththethemeofdistributedcomputingtechnologyandprojectdiscussionshavealsoformed.Mostofthesecommunitiesaretranslatingandproducingtutorialsfordistributedcomputingprojects,publishingrelatedtechnicalarticles,andprovidingnecessarytechnicalsupport.

Sowhomightjointheseprojects?Ofcourse,anyonecan!Ifyouhavealreadyjoinedaprojectandconsideredjoiningthecomputingteam,youwillbeintheChinaDistributedComputingCenterandFindyourhomeintheforum.Anyonecanjoinanydistributedcomputinggroupformedbyoursite.IhopeyouwillfindfuninChinaDistributedCentralStationandforums.

Participateindistributedcomputing-themostmeaningfulchoicethatcangivefullplaytotheusevalueofyourpersonalcomputer-onlyneedtodownloadtherelevantprogram,andthenthisprogramwillbeinthecomputerwiththelowestpriorityRunonthecomputer,whichhasalmostnoeffectonthenormaluseofthecomputer.Ifyouwanttouseyourfreetimeonyourcomputertodosomethinguseful,whydoyouhesitate?Actnow,yourinsignificantcontributionmaybeabletomakeyouleaveafortuneinthehistoryofthedevelopmentofhumanscience.

Distributedstoragesystem

Distributedstoragesystemistostoredatainmultipleindependentdevices.Thetraditionalnetworkstoragesystemusesacentralizedstorageservertostorealldata.Thestorageserverbecomesthebottleneckofsystemperformanceandthefocusofreliabilityandsecurity,whichcannotmeettheneedsoflarge-scalestorageapplications.Theadventoftheeraofbigdatamakestheresearchondistributedstoragesystemsofgreatsignificance:

Formassivedatastorage,distributeddatastoragehasgoodscalability.Robustnessandefficiencysurpassthetraditionalcentralizedstoragetechnology,butmanyofitsownperformanceindicatorssuchasdataredundancy,dataaccessspeed,bandwidthoccupancy,storagecostandreliability,etc.makedifferentsystemsanddifferentIndividualsandenterpriseshavedifferentemphasisonstoragerequirements.Datastoragemostlyconsidersaccessefficiencyandstoragecost,andthereisverylittleresearchondatainvulnerability.

Forthemanagementandmaintenanceofmassivedata,maintainingdataconsistencyisakeydirectionfordistributedstoragesystemstomaintaindata.Duetotheever-changingInternetenvironment,dataupdatespeedandconversionfrequencycontinuetoaccelerate,resultingindataconsistencymaintenanceFacingmanyproblems,suchasreliabilityproblems,dataredundancyproblems,networkturbulenceproblems,andmaliciousattacks,haveseriouslyaffectedtheformulationanddevelopmentofconsistencymaintenancestrategies.

P2Pdatastoragesystem

P2PdatastoragesystemadoptsthecharacteristicsofP2Pnetwork,thatis,eachuseristheacquirerandproviderofdata,andthereisnocentralnode,soeachuserTheyallexistonanequalfooting.AP2Pdatastoragesystembuiltwiththisfeaturecanstoredataonmultiplepeernodes.Whendataisneeded,afixedresourcesearchalgorithmcanbeusedtofinddataresourcestoobtainthedesireddata.

ThisfeatureofP2Pdatastoragesystemmakesitverysuitableforstoringlargeamountsofdata.Firstofall,becausethereisnocentralserver,thedataisdistributedandstoredoneachpeernode,sothatthereisnoproblemofexcessiveloadonacertainnode,andthescalabilityisgood;second,theabilitytoresistcyberattacksisstrong.Whenthereisanetworkattack,thedatalostbytheattackednodeisonlyasmallpartoftheentiredatastoragesystem,andmostoftheresourcesarestillinasafestate;finally,thankstotheexcellentscalabilityandeasymaintenanceoftheP2Pnetwork,P2PDatastoragesystemsalsohavetheseadvantages.

Cloudstoragesystem

Cloudstoragesystemisakindofnetworkstoragesystem,whichintegratesalargenumberofdatastorageserversandmanifestsasmultiplestorageserversworkingtogetherinternally.Thetaskofdatastoragedividesthetaskofdatastorageintomultiplesubtasksforparallelstorage,therebyreducingthetimeofdatastorageandincreasingdatasecurity.Toputitsimply,cloudstorageistostoredataorfilesinthecloud.DatauserscaneasilyaccessdatathroughtheInternetatanyplace,anddatastorageinthecloudhastheadvantagesofhighsecurityandlowcost.

Application

Theso-calleddistributedmeansthatthedataandprogramscannotbelocatedononeserver,butdistributedtomultipleservers,withgeographicalinformationdatadistributedonthenetworkandaffectedbyit.Theaffecteddatabaseoperationisatheoreticalcalculationmodeloftheresearchobject.Distributedisconducivetotaskallocationandoptimizationontheentirecomputersystem,overcomestheshortcomingsoftraditionalcentralizedsystemsthatcausecentralhostresourcetensionandresponsebottlenecks,andsolvesthedataheterogeneity,datasharing,andcomplexcalculationsthatexistinthenetworkGIS.Theproblemisamajoradvancementingeographicinformationsystemtechnology.

TraditionalcentralizedGISisdifficulttoapplytoatleasttwotypesofgeographicinformationsystems,andadistributedcomputingmodelisrequired.Thefirstcategoryisalarge-scaleprofessionalgeographicinformationsystem,thematicgeographicinformationsystemoraregionalgeographicinformationsystem.Thespatiotemporaldatasources,types,andstructuresoftheseinformationsystemsarediverse,andonlybydistributedcanthedivisionoflaborandcooperationindataresourcesharinganddataprocessingberealized.Forexample,inthecomprehensivemunicipalundergroundpipenetworksystem,thedataoftapwater,gas,andsewagearedistributedintheirrespectivemanagementorganizations.Tocollect,edit,store,extract,andanalyzethesedata,itisnecessarytousedistributedcomputingandmakethesetasks.Theyareallcarriedoutintheirrespectiveinstitutions,andtheirrespectivemanagementsystemsareestablishedassubsystemsoftheintegratedsystemtocompletethemanagementwork.Thetraditionalcentralizedtypecannotprovidethisnecessarydivisionoflabor.Thesecondcategoryisacomprehensiveinformationmanagementsystemwithinascope.Urbangeographicinformationsystemisaveryrepresentativeexampleofthiskindofsystem.Urbanmunicipalmanagementaccountsforalargeproportionofmanagementworkinallcountriesintheworld.Thedistributioncharacteristicsofurbaninformationandthegeographicaldispersionofurbaninformationmanagementdepartmentsdeterminethatmulti-level,multi-component,andmulti-contenturbaninformationmustadoptadistributedprocessingmodel.

Itisobviousthattraditionalcentralizedgeographicinformationsystemscannotmeettheneedsofamodernsocietywithacleardivisionoflabor,andthefurtherdevelopmentofdistributedgeographicinformationsystemshasanunstoppablemomentum.Moreover,distributedGISiscloselyrelatedtonetworkGIS,client/serverGIScomputingmodel,andWWWcomputingmodel.DistributedGISisawaytorealizenetworkGIS,anditisanimportantcalculationmodelforNGIS;CISmodelisactuallyaspecificcalculationmodelthatcanbeusedbydistributedGIS;WWWmodelisalsoakindofspecificcalculationmodelthatcanbeusedbydistributedGIS.ItisaspecificmodelofGIS,anditisalsoadistributedGISmodelwithconsiderabledevelopmentprospects.ThecloseconnectionbetweendistributedGISandthetechnologythatdominatesthedevelopmentdirectionofgeographicinformationsystemsmakesdistributedGIScorrespondinglythemaindevelopmenttrendofgeographicinformationsystems.

Tento článek je ze sítě, nereprezentuje pozici této stanice. Uveďte prosím původ dotisku
HORNÍ