《信息科學類專業(yè)英語》課件第5章

Lesson5IntroductiontoCloudComputing(第五課云計算簡介)

Vocabulary(詞匯)ImportantSentences(重點句)QuestionsandAnswers(問答)Problems(問題)ReadingMaterial(閱讀材料)

NumeroussurveysreportthatCloudComputingwillbeatop10technologythatenterprisebusinessmanagersneedtobeawareoffor2010.Notthatyoucanescapethemarketingandinformationpublishedaboutthislatestsuperhypedtopic.MuchofthemessagefocusesonCloudComputingasalowercostdeliverymodelforITservices.Thismayormaynotbetrue.1SoWhatisCloudComputing?

WeseeCloudComputingasacomputingmodel,notatechnology.Inthismodel“customers”plugintothe“cloud”toaccessITresourceswhicharepricedandprovided“on-demand”.Essentially,ITresourcesarerentedandsharedamongmultipletenantsmuchasofficespace,apartments,orstoragespacesareusedbytenants.DeliveredoveranInternetconnection,the“cloud”replacesthecompanydatacenterorserverprovidingthesameservice.Thus,CloudComputingissimplyITservicessoldanddeliveredovertheInternet.RefertosectionofTypesofCloudComputing.

CloudComputingvendorscombinevirtualization(onecomputerhostingseveral“virtual”servers),automatedprovisioning(servershavesoftwareinstalledautomatically),andInternetconnectivitytechnologiestoprovidetheservice.[1]Thesearenotnewtechnologiesbutanewnameappliedtoacollectionofolder(albeitupdated)technologiesthatarepackaged,soldanddeliveredinanewway.

Akeypointtorememberisthat,atthemostbasiclevel,yourdataresidesonsomeoneelse’sserver(s).Thismeansthatmostconcerns(andtherearepotentiallyhundreds)reallycomedowntotrustandcontrolissues.Doyoutrustthemwithyourdata?2TheEconomics

EconomiesofscaleandskilldriveCloudComputingeconomics.AswithrentedRealEstate,thecostsofownershiparepooledandspreadamongalltenantsofthemulti-tenantCloudComputingsolution.Consequentially,acquisitioncostsarelowbuttenantsneverownthetechnologyassetandmightfacechallengesiftheyneedto“move”orendtheserviceforanyreason.[2]

SomethingthatisoftenoverlookedwhenevaluatingCloudComputingcostsisthecontinuedneedtoprovideLANservicesthatarerobustenoughtosupporttheCloudsolution.Thesecostsarenotalwayssmall.Forexample,ifyouhave6ormoreworkstationcomputers,youwillprobablyneedtocontinuetomaintainaserverinadomaincontrollerrole(toensurenameresolution),atleastoneswitch(toconnectallofthecomputerstoeachotherandtherouter),oneormorenetworkedprinters,andtherouterfortheInternetconnection.3WhatdoIneedtouseCloudComputing?

AllthatisreallyneededtoacquireanduseCloudComputingsolutionsisacreditcard(orotherpaymentmethod)andaLANwithanInternetconnectionrobustenoughtosupporttheClouddeliveredservice.Thesetworequirementsaredeceptivelysimple.

Fromatechnicalpointofviewthebiggestchallengeforbusinesses,particularlySMBs,maybetheneedforanappropriatelyrobustLANinfrastructureandInternetconnection.Typically,InternetaccessisprovidedbyasinglecommercialserviceISPproviderthroughasingleportonarouter.AcharacteristicofthistypeinstallationisthatallofthecomputersconnectingthroughtheLANsharetheInternetbandwidthequally.Thiscanquicklybecomeanissue.

Forexample:VerizonFiOSInternet15/2(down/up)servicemighthaveameasuredspeedof14420/1867Kbps.Thiswouldseemtobeplentyofspeed.However,supposeabusinesshad5computersusingaCloudsolutionandsendingdatatothecloudforprocessing.Thebandwidthavailabletoeachcomputerwouldbe373Kbps(up1867/5).Thatisabout46(8bit)characterspersecondtothecloudapplicationanddoesnotincludeanycommunicationorapplicationdata.Thecloudsolutionmightnotworkorresponsessoslowastobeunacceptable.Itisn’tthedownloadspeedthatbecomesalimit,butheuploadspeed.RefertoPage5foraBandwidthChart.

Theon-demandnatureofCloudComputingpresentsadilemma:Theon-demandmodelincludesaself-serviceinterfacethatallowsuserstoself-provisionservices(forexamplestorage).[3]Thisempowersusersbutcanmakeservicestooeasytoacquireandconsume.ToquoteanITadministrator“Peoplecouldcarelessaboutpolicies.Theywantwhattheywantwhentheywantit.Theydon’tinvolveIT.”

ConsiderthefacultymemberattheUniversityofMassachusettswhoquietly(withoutanyone’sknowledge)usedacloudservicetobackup20GBofdataeachnightovertheInternetbringingtheschoolLANtoitsknees.HowmanagementcontrolsCloudComputingisuniquetoeachorganizationandisanITGovernanceissue.4Conclusion

Weareoftentoldparticularlybyvendorsandevangelists,“Youdon’tlikeOutsourcedorCloudComputingsolutions”.Thisissimplynottrue.Outsourcingandusingthirdpartiesforservicecanbeveryhelpfultoclients.However,wedonotthinkthatthesesolutionsareappropriateoreffectiveineverysituationorforeveryorganization.5TypesofCloudComputing

5.1SaaS(SoftwareasaService)

Isthemostwidelyknownandwidelyusedformofcloudcomputing.Itprovidesallthefunctionsofasophisticatedtraditionalapplicationtomanycustomersandoftenthousandsofusers,butthroughaWebbrowser,nota“l(fā)ocally-installed”application.[4]LittleornocodeisrunningontheUserslocalcomputerandtheapplicationsareusuallytailoredtofulfillspecificfunctions.SaaSeliminatescustomerworriesaboutapplicationservers,storage,applicationdevelopmentandrelated,commonconcernsofIT.

Highest-profileexamplesareS,Google’sGmailandApps,instantmessagingfromAOL,YahooandGoogle,andVoIPfromVonageandSkype.

5.2PaaS(PlatformasaService)

Deliversvirtualizedserversonwhichcustomerscanrunexistingapplicationsordevelopnewoneswithouthavingtoworryaboutmaintainingtheoperatingsystems,serverhardware,loadbalancingorcomputingcapacity.ThesevendorsprovideAPIsordevelopmentplatformstocreateandrunapplicationsinthecloud(e.g.usingtheInternet).ManagedServiceproviderswithapplicationservicesprovidedtoITdepartmentstomonitorsystemsanddownstreamapplicationssuchasvirusscanningfore-mailarefrequentlyincludedinthiscategory.[5]

WellknownproviderswouldincludeMicrosoft’sAzure,Salesforce’sF,GoogleMaps,ADPPayrollprocessing,andUSPostalServiceofferings.

5.3IaaS(InfrastructureasaService)

Deliversutilitycomputingcapability,typicallyasrawvirtualservers,ondemandthatcustomersconfigureandmanage.HereCloudComputingprovidesgridsorclustersorvirtualizedservers,networks,storageandsystemssoftware,usually(butnotalways)inamultitenantarchitecture.IaaSisdesignedtoaugmentorreplacethefunctionsofanentiredatacenter.Thissavescost(timeandexpense)ofcapitalequipmentdeploymentbutdoesnotreducecostofconfiguration,integrationormanagementandthesetasksmustbeperformedremotely.

VendorswouldincludeA(ElasticComputeCloud[EC2]andSimpleStorage),IBMandothertraditionalITvendors.

1．hypevt.使增加;夸大;<俚>(用刺激針劑等)強行刺激,使興奮(up)。

2．on-demand在要求時,一經(jīng)請求;在要求支付時。

3．essentiallyadv.本質(zhì)上,本來,根本。

4．provisionn.供應,(一批)供應品,預備,防備,規(guī)定。

5．potentiallyadv.潛在地。Vocabulary

6．infrastructuren.基礎;基礎結(jié)構(gòu)[設施](尤指社會、國家賴以生存和發(fā)展的)；【建】基[底]層結(jié)構(gòu),下部結(jié)構(gòu)；【軍】永久性防御設施,永久性基地。

7．dilemman.進退兩難的局面,困難的選擇。

8．empowervt.授權(quán),準許;使能夠。

9．outsourcing[商]外部采辦,外購。

10．sophisticatedadj.詭辯的,久經(jīng)世故的。

11．downstreamadv.下游地adj.下游的n.后臺處理程序。

12．multitenantn.多承租；多租戶。

[1]CloudComputingvendorscombinevirtualization(onecomputerhostingseveral“virtual”servers),automatedprovisioning(servershavesoftwareinstalledautomatically),andInternetconnectivitytechnologiestoprovidetheservice.

云計算提供商將虛擬化(一臺計算機主機對應多個虛擬服務器)、自動運行維護(服務器具有自動安裝的軟件)以及因特網(wǎng)連接技術(shù)組合來提供這些服務。ImportantSentences

[2]Consequentially,acquisitioncostsarelowbuttenantsneverownthetechnologyassetandmightfacechallengesiftheyneedto“move”orendtheserviceforanyreason.

必然地，買進成本是低的，但是承租人從未擁有技術(shù)財產(chǎn)，并且如果他們因任何原因需要轉(zhuǎn)移或結(jié)束這項服務時可能面臨挑戰(zhàn)。

[3]Theon-demandnatureofCloudComputingpresentsadilemma:Theon-demandmodelincludesaself-serviceinterfacethatallowsuserstoself-provisionservices(forexamplestorage).

云計算的按請求服務的本質(zhì)呈現(xiàn)這樣一種尷尬：按請求服務的模型包含一種自服務的接口以允許用戶使用一些自規(guī)定的服務(如存儲)。

[4]Itprovidesallthefunctionsofasophisticatedtraditionalapplicationtomanycustomersandoftenthousandsofusers,butthroughaWebbrowser,nota“l(fā)ocally-installed”application.

它通過網(wǎng)絡瀏覽器，而不是本地安裝的應用程序，提供所有的復雜的傳統(tǒng)應用功能給許多客戶，通常有數(shù)以千計的用戶。

[5]ManagedServiceproviderswithapplicationservicesprovidedtoITdepartmentstomonitorsystemsanddownstreamapplicationssuchasvirusscanningfore-mailarefrequentlyincludedinthiscategory.

管理服務提供商將應用服務提供給IT部門來監(jiān)視系統(tǒng)及后臺處理程序，如電子郵件的病毒掃描通常就歸屬于這一范疇。

(1)?WhatkindsoftechnologiesdotheCloudComputingvendorscombinetoprovideservice?()

A.?virtualization,automatedprovisioning,andInternetconnectivity.

B.?virtualization,database,andInternetconnectivity.

C.?Graphics,database,andInternetconnectivity.

D.?virtualreality,Graphics,database.QuestionsandAnswers

(2)?Whatdoestheon-demandmodelinclude?()

A?Theon-demandmodelincludesaGUIinterfacethatallowsuserstoself-provisionservices.

B.?Theon-demandmodelincludesaself-provisioninterfacethatallowsuserstoself-serviceservices.

C.?Theon-demandmodelincludesaself-serviceinterfacethatallowsuserstoself-provisionservices.

D.?Theon-demandmodelincludesaself-serviceinterfacethatallowsserverstoself-provisionservices.

(3)?Whichkindofserviceisthemostwidelyknownandwidelyusedformofcloudcomputing?()

A.?PaaS(PlatformasaService).

B.?SaaS(SoftwareasaService).

C.?IaaS(InfrastructureasaService).

D.?Alloftheabove.

(4)?Whichkindofserviceisdesignedtoaugmentorreplacethefunctionsofanentiredatacenter？()

A.?PaaS(PlatformasaService).

B.?SaaS(SoftwareasaService).

C.?IaaS(InfrastructureasaService).

D.?Alloftheabove.

(5)?Whichkindofservicehaseliminatedcustomerworriesaboutapplicationservers,storage,applicationdevelopmentandrelated,commonconcernsofIT？()

A.?SaaS(SoftwareasaService).

B.?PaaS(PlatformasaService).

C.?IaaS(InfrastructureasaService).

D.?Alloftheabove.

1.?WhatdoweneedtouseCloudComputing?

2.?What’sthedifferenceamongthethreetypesofCloudComputing?Problems

There’sakindofsupply-and-demanddynamicthatappliestotechnicalterms—whenafewknowledgeableinsidersarehoardingaword,itmaintainsitsmeaning,butwhenthemassesgetholdofitandabuseit,it’squicklyemptiedofvalue.ReadingMaterialTheCloud:ashortintroduction

Thisiscertainlythecasewith“thecloud,”atermthatusedtomeansomething,andnowmeanseverythingandnothing.“Thecloud”issooverusedbystartupsdesperateforVCmoney,andbybigcompaniesdesperatetolooklikehipstartups,thatITprofessionalsareincreasinglywaryofanythingcloud-related.Itdoesn’thelpthattheimageconjuredbythewordisofsomethingvaporous,flimsy,andfleeting—whatevercloudis,itdoesn'tsoundlikethekindofthingyouwanttoentrustcriticalbusinessfunctionsto.

Despitethefactthateveryoneseemstoseeadifferentshapewhentheystareatit,thereissomethingworthpreservingin“thecloud”asatermthatusefullydescribesoneapproachtowhatisoftencalled“utilitycomputing”,whichlattertermisitselfametaphoricalwayofspeakingaboutabusinessmodelcenteredaroundtheideaofcomputingpowerasaservicelikeelectricalpower.

Infirstdefiningandthendescribingcloudcomputinginthisbriefarticle,myaimistoprovideausefuldefinitionforITprofessionalswhoaretaskedwithexploringcloudservicesasapotentialavenueforfindingnewefficiencies,reducingfixedcosts,tacklingscalingchallenges,andsolvingnovelproblemsatInternetscale.MysecondaryaudiencesforthispieceareITproswhoneedtoquicklyexplain“thecloud”toacluelessCIO,andcluelessCIOswho’drathernothavetorelyonITprostoexplainbuzzwordstothem.

Thisarticletakesahistoricalandcomparativeapproachtothetopicofcloudcomputing.First,I’llintroducethevenerableclient-servermodel,amodelofwhichcloudisjustthelatestinstance,andthenI’llcontrastthecloudwithitsimmediatepredecessor,thegrid.Finally,I’lldescribethethree-tieredmodelofcloudservices.

Oneofthemostcommonquestionsaskedbycloudskepticsis,“isn’tcloudjustclient-server?”Theanswertothisis,yes,itis.Therearemanyproducer-consumerrelationshipsateverylevelofcomputing,fromtheindividualsystemouttothenetwork,thatcanusefullybethoughtofinclient-serverterms.Forinstance,aPC’smainmemoryservesavarietyofclientsscatteredthroughoutthesystemviaDMArequests.Ingeneral,aclient-serverrelationshipischaracterizedbyasingleproducerthatallowsmultipleconsumersaccesstoitsresourcepool.Abriefhistoryofclient-server

Inthisrespect,cloudfitstheclient-servermodel,and,insofarasthetypicalcloudclientisthesameasthetypicalenterpriseclient(i.e.singledesktoporlaptopcomputer),someobservershaveatendencytostopatthislevelofanalysis.But,ofcourse,therealactioninthecloudhappensontheserversideoftheequation,andthat’swherethingsgetinteresting.Butbeforewegetintothecloudinearnest,let’stakeabrieflookbackatclient-server.

Thereareessentiallytwokindsofresourcesthataservercanprovidetoclients:storageandcomputecycles.Client-servermodelscangenerallybecategorizedaccordingtowhichtypeofresourcetheyprovide.

Chronologically,thefirsttypeofclient-serverpairtobecomepopularwasthemainframeandterminal.SincestorageandCPUcyclesweresoexpensive,themainframepooledbothtypesofresourcesandservedthemtothin-clientterminals.WiththeadventofthePCrevolution,whichbroughtmassstorageandcheapCPUstotheaveragecorporatedesktop,thefileservergainedinpopularityaswaytoenabledocumentsharingandarchiving.Truetoitsname,thefileserverservedupstorageresourcestoclientsintheenterprise,whiletheCPUcyclesneededtodoproductiveworkwiththoseresourceswereallproducedandconsumedwithintheconfinesofthePCclient.

The1980salsosawtheriseofthesupercomputer,whichfeaturedalarge,homogenousarrayofprocessorsandwasdesignedtoserveCPUcyclesto“fat-client”workstations.Supercomputerswerelimitedtogovernment(mostlymilitary)andgovernment-sponsoredpartsofacademia,notjustbecausethosesectorsweretheonlyoneswiththeappetiteforthatmuchnumbercrunchingpower,butbecausethosetypesofpublicinstitutionshadpocketsdeepenoughtoaffordthesemachines(itwasvery,veryexpensivetopoolCPUcyclesandservethematascalethatcouldactuallydousefulwork).

ButwhilethesupercomputermarketwasheatingupalongwiththeColdWarthatmuchofitsoutputwenttowardfighting,theseedsofthatmarket’sdestructionwerebeingsownbybothMoore'sLawandtheInternet.

Intheearly1990s,thebuddingInternetfinallyhadenoughcomputersattachedtoitthatacademicsbeganthinkingseriouslyabouthowtoconnectthosemachinestogethertocreatemassive,sharedpoolsofstorageandcomputepowerthatwouldbemuchlargerthanwhatanyoneinstitutioncouldaffordtobuild.Thisiswhentheideaof“thegrid”begantotakeshape.Thegrid,andtheriseofutilitycomputing

Theterm“grid”isametaphordeliberatelydrawnfromtherealmofelectricitygeneration,whereelectricutilitiesprovidepowerovera“grid”networktoclientswhopayonameteredbasisfortheelectricitythattheyconsume.Theideabehindthegridmodel,andtherelatedconceptof“utilitycomputing”,wasthatasufficientlylargenumberofnetworkedcomputerscouldbepooledtogetherlikeagiant,virtualsupercomputerorfileserver,andaccesstothatpoolofcomputeorstorageresourcescouldbesoldinanon-demand,meteredfashion.

Inall,gridcomputingfeaturesalargenumberofnetworked,oftengeographicallyandinstitutionallyseparatenodesthattogethermakeupasharedpoolofcomputeresources.Dataandcomputationalgridsarecharacterizedbyautonomous,homogeneousnodesthatarelooselycoupledandoftenusepublicnetworks.Notethatthegrid’sloosecouplingofnodesisamajorcharacteristicthatdistinguishesitfromthecluster,asimilarmultinodecomputingconceptwithwhichthegridisoftenconfused.Clustersfeaturenodesthatareconnectedbyveryhigh-bandwidthlinks,andthisbandwidthadvantagegivesthemalotmoreaveragecomputepowerpernodethanagridbecausenodesdon’tspendasmuchidletimewaitingondatatoarrive.

Computationalgridsaremorecommonthandatagrids,andapplicationshavetobespeciallywrittenforsuchgridsanddesignedtoscaletoalargenumberofparallelnodes.Atypicalcomputationalgridclientturnstothegridbecauseheneedstorunamassive,compute-intensivejobthatwilloccupyalargesubsetofthosenodesforagivenperiodoftime.Fig.1Thegrid.Differentcoloredjobsbelongtodifferentclients.(OneofthosejobsbelongstotheDepartmentofDefense.)

Gridjobsareoftenruninbatches,whereavailablenodesarepooledtogetherandthenassignedworkthatmonopolizesthemuntilit’sdone.(Note:manygridnodes,likethoseinvolvedintheproject,alsorunlocalclientsoftwaresimultaneouslywiththeirgridjob;butfromthepoint-of-viewofthegrid,thatnodeisstillworkingonasinglejob.)Whenthegridjobiscomplete,thenodesarereleasedbackintothepoolofavailableresources,andarereadyforsomeotherclienttouse.

Onekeyaspectofthegridisthatmultipleinstitutionscansharethesamehardwareresourceswithoutworryingaboutanyoneelseonthegridgainingunauthorizedaccesstotheirdata.Eventhoughthedataisonapubliclyaccessiblegrid,itremainsaccessibleonlytotheclientthatownsit.It’salsothecasethatthegridhardwareitselfoftenhasmanyinstitutionaland/orindividualowners—eachpartycontributescomputeresourcestoasharedpool,andinexchange,contributorscanbidforcyclesfromthatpool.

Thecloudisthesamebasicideaasthegrid,butscaleddowninsomeways,scaledupinothers,andthoroughlydemocratized.Takealookatthediagrambelow,andcontrastitwiththegriddiagramabove.ThecloudFig.2Thecloud.Differentcoloredjobsbelongtodifferentclients.(Oneofthosejobsbelongstoyour18-year-oldnephew.)

Likethegrid,thecloudisautilitycomputingmodelthatinvolvesadynamicallygrowingandshrinkingcollectionofheterogenous,looselycouplednodes,allofwhichareaggregatedtogetherandpresentthemselvestoaclientasasinglepoolofcomputeand/orstorageresources.Butthoughtheserversideofthemodelmaylooksimilar,mostthemajordifferencesbetweencloudandgridstemfromthedifferencesbetweentheirrespectiveclients.

Insteadofafewclientsrunningmassive,multinodejobs,thecloudservicesthousandsormillionsofclients,typicallyservingmultipleclientspernode.Theseclientshavesmall,fleetingtasks(e.g.,databasequeriesorHTTPrequests—thatareoftencomputationallyverylightweightbutpossiblystorage-orbandwidth-intensive).Fig.3CloudvsGrid

Anotherdifferencebetweenthecloudandthegridisthatthegridsarebiasedtowardservingcomputecycles,whilecloudstypicallyoffermoreinthewayofstoragethancycles.Indeed,mostgridswouldbeveryill-suitedtocloudworkloadslikeWebserving,andmostcloudswouldfallfarshortofgridclients’massivecomputeneeds.

Becauseofthenatureoftheirrespectiveclientprofiles,cloudsandgridsalsohavedifferentownershipcharacteristics.Inotedabovethatgridstendtobemulti-institutional,whereinstitutionsand/orindividualsallcontributehardwareresourcesthatarethensharedbyothercontributors.Acloud,incontrast,isalwaysownedbyoneinstitution,regardlessofwhetheruseofthecloudisopentoclientsoutsidethatinstitutionornot(i.e.,whetherthecloudispublic,private,orhybrid).

Cloudservicesareofferedatthreebasiclevels,ortiers,thataredistinguishedbythelevelofabstractionthateachpresentstotheclient.Thesetiersroughlymaptothethreelayersofthestandardhardware/OS/applicationsstackfamiliartoanyonewhousesaPC.Cloudservices:tiersandfears

Thelowestcloudtierisinfrastructure-as-a-service(IaaS),whichlookstotheclientlikeadynamicallyscalablepoolofcomputeand/orstorageresources.ThebasicmeteredunitofIaaSisusuallyeitherasinglevirtualmachine(e.g.,AmazonEC2)oranabstractstorageobjectofacertainsize(e.g.,AmazonS3).

Nextuptheladderofabstractionisplatform-as-a-service(PaaS),whichprovidesAPI-levelaccesstoacloudinfrastructurelayer.ExamplesofPaaSareGoogleAppEngineandF.BecausePaaSofferingsoftencomewrappedinavendor-specificAPI,theuseofthislayerprettymuchlocksyouintoaparticularvendor.It’satthistierthatenterprisecustomersmusttakeseriouslytherisk/rewardtradeoffbetweentheconvenienceandagilityaffordedbyavendor’scloudofferingandthepotentialinconvenienceofbeingunabletoeasilymoveawayfromthatvendor’splatformshouldbusinessortechnicalconsiderationsdemandit.

Thefinalandmostpopulartierofcloudserviceissoftware-as-a-service(SaaS).GoogleAppsandSarethetwoparadigmaticSaaSexamples,andthey’resoubiquitousthatnotmuchmoreneedstobesaidaboutthiscloudtier.

Insofaras