信息基础设施_第1页
信息基础设施_第2页
信息基础设施_第3页
信息基础设施_第4页
信息基础设施_第5页
已阅读5页,还剩30页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

InformationInfrastructure

Lecture051Withorganizations’growingneedsforacomprehensiveinformationsystemsinfrastructure,anumberofsolutionshaveemergedandarecontinuingtoemerge.Whilesomeofthesesolutionsarealreadycommonbusinesspractice,othersarejustnowstartingtobeadopted.2DesigningtheInformationSystemsInfrastructureBothbusinessesandresearchfacilitiesfaceanever-increasingneedforcomputingperformance.Forexample,automanufacturers,suchJapaneseToyota,uselargesupercomputerstosimulateautomobilecrashesaswellasevaluatedesignchangesforvibrationsandwindnoise.3ManagingtheHardwareInfrastructureWhilenoteveryorganizationfacessuchlarge-scalecomputingproblems,thedemandsforcomputingresourcesareoftenfluctuating.Thismightleadstoeitherhavingtoofewresourcesforsomeproblemsorhavingtoomanyidleresourcesmostofthetime.4Toaddressthisproblem,manyorganizationsnowturnto:on-demandcomputingforfluctuatingcomputationneedsgridcomputingforsolvinglarge-scaleproblemsautonomiccomputingforincreasingreliability.5Inalmosteveryorganization,demandforindividualISresourcesishighlyfluctuating.Forexample,somehigh-bandwidthapplications,suchasvideoconferencing,maybeneededonlyduringcertaintimesoftheday,orsomeresourceintensivedata-miningapplicationsmayonlybeusedinirregularintervals.On-demandcomputingisawaytoaddresssuchfluctuatingcomputingneeds;here,theavailableresourcesareallocatedonthebasisofusers’needs(usuallyonapay-per-usebasis).6On-DemandComputingForexample,morebandwidthwillbeallocatedtoavideoconference,whileotheruserswhodonotneedthebandwidthatthattimereceiveless.Similarly,auserrunningcomplexdataminingAlgorithmswouldreceivemoreprocessingpowerthanausermerelydoingsomewordprocessing.7Attimes,organizationspreferto“rent”resourcesfromanexternalprovider.Thisformofon-demandcomputingisreferredtoasutilitycomputing,wheretheresourcesintermsofprocessing,datastorage,ornetworkingarerentedonanas-neededbasis.Organizationreceivesabillfortheservicesusedfromtheproviderattheendofeachmonth8Formanycompanies,utilitycomputingisaneffectivewayformanagingfluctuatingdemandaswellascontrollingcosts;inessence,alltasksassociatedwithmanaging,maintaining,andupgradingtheinfrastructurearelefttotheexternalproviderandaretypicallybundledintothe“utility”bill—ifyoudon’tuse,youdon’tpay.Also,aswithyourutilitybill,customersarechargednotonlyonoverallusagebutalsoonpeakusage(i.e.,differentratesfordifferenttimesoftheday).10GridComputingAlthoughtoday’ssupercomputershavetremendouscomputingpower,sometasksareevenbeyondthecapacityofasupercomputer.Indeed,somecomplexsimulationscantakeayearorlongertocalculateevenonasupercomputer.Sometimes,anorganizationoraresearchfacilitywouldhavetheneedforasupercomputerbutmaynotbeabletoaffordonebecauseoftheextremelyhighcost.11Forexample,thefastestsupercomputerscancostmorethan$200million,andthisdoesnotrepresentthe“totalcostofownership,”whichalsoincludesalltheotherrelatedcostsformakingthesystemoperational(e.g.,personnel,facilities,storage,software,andsoon)Additionally,theorganizationmaynotbeabletojustifythecostbecausethesupercomputermaybeneededonlyoccasionallytosolveafewcomplexproblems.Inthesesituations,organizationshavehadtoeitherrenttimeonasupercomputerordecidedsimplynottosolvetheproblem.12However,arelativelyrecentinfrastructuretrendforovercomingcostoruselimitationsistoutilizegridcomputing.Gridcomputingreferstocombiningthecomputingpowerofalargenumberofsmaller,independent,networkedcomputers(oftenregulardesktopPCs)intoasolidsysteminordertosolveproblemsthatonlysupercomputerswerepreviouslycapableofsolving.1314Whilesupercomputersareveryspecialized,gridcomputingallowsorganizationstosolvebothverylarge-scaleproblemsaswellasmultiple(concurrent)smallerproblems.Tomakegridcomputingwork,largecomputingtasksarebrokenintosmallchunks,eachofwhichcanthenbecompletedbytheindividualcomputers15However,astheindividualcomputersarealsoinregularuse,theindividualcalculationsareperformedduringthecomputers’idletimesoastomaximizetheuseofexistingresources.Forexample,whenwritingareport,weusedonlyminimalresourcesonourcomputers(i.e.,wetypicallyusedonlyawordprocessor,theInternet,maybee-mail);ifourcomputerswerepartofagrid,theunusedresourcescouldbeutilizedtosolvelarge-scalecomputingproblems.16Thisisespeciallyusefulforcompaniesoperatingonaglobalscale.Ineachcountry,manyoftheresourcesareidleduringthenighthours,oftenmorethan12hoursperday.Becauseoftimezonedifferences,gridcomputinghelpsutilizethoseresourcesconstructively.OnewaytoputtheseresourcesintousewouldbetojointheBerkeleyOpenInfrastructureforNetworkComputing(BOINC),whichletsindividuals“donate”computingtimeforvariousresearchprojects,suchassearchingforextraterrestrialintelligence(SETI@home)orrunningclimatechangesimulations.17However,asyoucanimagine,gridcomputingposesanumberofdemandsintermsoftheunderlyingnetworkinfrastructureorthesoftwaremanagingthedistributionofthetasks.Further,manygridsperformonthespeedoftheslowestcomputer,thusslowingdowntheentiregrid.Manycompaniesstartingoutwithagridcomputinginfrastructureattempttoovercometheseproblemsbyusingadedicatedgrid.18Inadedicatedgrid,theindividualcomputers,ornodes,arejusttheretoperformthegrid’scomputingtasks;inotherwords,thegridconsistsofanumberofhomogeneouscomputersanddoesnotuseunutilizedresources.Adedicatedgridiseasiertosetupandmanageandisformanycompaniesmuchmorecosteffectivethanpurchasingasupercomputer.Asthegridevolvesandnewnodesareadded,dedicatedgridsbecomemoreheterogeneousovertime.19AnotherrecenttrendinIShardwareinfrastructuremanagementisedgecomputing.Withthedecreaseincostforprocessinganddatastorage,computing

tasksarenowoftensolvedattheedgeofacompany’snetwork.Inotherwords,ratherthanhavingmassive,centralizedcomputersanddatabases,multiplesmallerserversarelocatedclosertotheindividualusers.20EdgeComputingThisway,resourcesintermsofnetworkbandwidthandaccesstimearesaved.Ifacomputerneedsseveralhourstocomputeacertainproblem,itmightbeagoodchoicetosendthetaskoveranetworktoamorepowerfulcomputerthatmightbeabletosolvethatproblemfaster.However,asthecostsforcomputingpowerhavedecreasedtremendouslyoverthepastyears,manyproblemscannowbecomputedlocallywithinamatterofseconds,soitisnotcost-effectivetosendsuchproblemsoveranetworktoaremotecomputer21Tosaveresources,manybusinessesuseedgecomputingfortheironlinecommercesites.Insuchcases,customersinteractwiththeserversofanedge-computingserviceprovider(suchasAkamai).Theseservers,inturn,communicatewiththebusiness’computers.Thisformofedgecomputinghelpstoreducewaittimesfortheconsumers,asthee-commercesitesarereplicatedonAkamai’sservers,whileatthesametimereducingthenumberofrequeststothecompany’sowninfrastructure.22Thisprocessnotonlysavesvaluableresourcessuchasbandwidthbutalsoofferssuperiorperformancethatwouldotherwisebetooexpensivefororganizationstooffer.Akamai’sservicesareutilizedbyorganizationssuchasNBC,FoxSports,BMW,andVictoria’sSecret.23OnemajordrawbackofthesehardwareinfrastructuretrendsandthedemandsforISinfrastructureingeneralistheincreasedcomplexityofsuchsystems.Whereastheprimaryreasonforhavingthisinfrastructureistheutilizationoftheresources,thetimeandmoneyneededtomanagetheseresourcesdon’taddvaluetotheorganization;infact,somepeoplebelievethatthecostsofmanagingthesesystemsunderminethebenefitsthesesystemsprovide,eveniftheorganizationdecidestouseoutsideservices.24AutonomicComputingToovercomethis,academicandindustryresearchers(e.g.atIBM)havebegunworkingonautonomiccomputingsystems,whichareself-managing,meaningtheyneedonlyminimalhumaninterventiontooperate.Inotherwords,inatraditionalcomputingenvironment,systemoperatorsoftenhavetofine-tunethecomputer’sconfigurationinordertomostefficientlysolveaparticulartypeofcomplexproblem.25Inanautonomiccomputingenvironment,theultimategoalistoallowthesystemtodoeverythingelseonitsown,completelytransparenttotheuser.Inordertoachievethis,anautonomiccomputingsystemmustknowitselfandbeself-configuring,selfoptimizing,self-healing,andself-protecting.2627Inordertooptimallyperformdifferenttasks,anautonomicsystemmustknowitself;thatis,itmustknowitsconfiguration,capacity,andcurrentstatus,butitmustalsoknowwhichresourcesitcandrawon.Second,inordertobeabletousedifferentresourcesbasedondifferentneeds,thesystemshouldbeself-configuringsothattheuserdoesnothavetotakecareofanyconfigurationissues.28Further,asanypartsofasystemcanmalfunction,anautonomicsystemshouldbeself-healingsothatanypotentialproblemsaredetectedandthesystemisreconfiguredsoastoallowtheusertocontinueperformingthetasks,evenifpartsofthesystemarenotoperational.Finally,asalmostanycomputersystemcanbethetargetforanattack,autonomiccomputingsystemsmustbeawareofanypotentialdangersandmustbeabletoprotectthemselvesfromanymaliciousattacks(e.g.,byautomaticallyquarantininginfectedpartsofasystem).29CloudComputingisageneraltermusedtodescribeanewclassofnetworkbasedcomputingthattakesplaceovertheInternet,basicallyasteponfromUtilityComputingacollection/groupofintegratedandnetworkedhardware,softwareandInternetinfrastructure(calledaplatform).UsingtheInternetforcommunicationandtransportprovideshardware,softwareandnetworking

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

评论

0/150

提交评论