计算机网络技术的_历史和新进展

计算机网络技术的历史和新进展 高速计算机信息网络是信息社会的神经和血管体系结构 网络的骨架和神经协议 网络的心脏和血液 主要内容 网络概述Internet的发展和成功经验计算机网络技术的历史回顾国际高速计算机网络研究计划中国高速计算机网络研究计划 WhatisaNetwork fromendsystempointofview Networkoffersaservice moveinformationbird fire messenger truck telegraph telephone Internet anotherexample transportationservice moveobjectshorse train truck airplane Whatdistinguishdifferenttypesofnetworks TheservicestheyprovideWhatdistinguishtheservices latencybandwidthlossratenumberofendsystemsserviceinterface howtoinvoke otherdetailsreliability unicastvs multicast real time messagevs byte WhatisaNetwork InfrastructureCentricView ElectronsandphotonsascommunicationmediumLinks fiber copper satellite Switches mechanical electronic optical crossbar BanyanProtocols TCP IP ATM MPLS SONET Ethernet PPP X 25 FrameRelay AppleTalk IPX SNAFunctionalities routing errorcontrol congestioncontrol QualityofService QoS Applications FTP WEB Xwindows TypesofNetworks GeographicaldistanceLocalAreaNetworks LAN Ethernet Tokenring FDDIMetropolitanAreaNetworks MAN DQDB SMDSWideAreaNetworks WAN X 25 ATM framerelayInformationtypedatanetworksvs telecommunicationnetworksApplicationtypespecialpurposenetworks airlinereservationnetwork bankingnetwork creditcardnetwork telephonygeneralpurposenetwork Internet TypesofNetworks Righttouseprivate enterprisenetworkspublic telephonynetwork InternetOwnershipofprotocolsproprietary SNAopen IPTechnologiesterrestrialvs satellitewiredvs wirelessProtocolsIP AppleTalk SNA 计算机网络发展历史回顾 七十年代的计算机网络X 25分组交换网 各国的电信部门建设运行各种专用的网络体系结构 SNA DNAInternet的前身ARPANET进行实验运行八十年代的计算机网络标准化计算机网络体系结构 OSI局域网络LAN技术空前发展建成NSFNET Internet初具规模九十年代的计算机网络Internet空前发展Web技术在Internet Intranet得到广泛应用 主要内容 网络概述Internet的发展和成功经验计算机网络技术的历史回顾国际高速计算机网络研究计划中国高速计算机网络研究计划 TheInternet Globalscale generalpurpose heterogeneous technologies public computernetworkInternetProtocolopenstandard InternetEngineeringTaskForce IETF asstandardbodytechnicalbasisforothertypesofnetworksIntranet enterpriseIPnetworkDevelopedbytheresearchcommunity HistoryoftheInternet 70 s startedasaresearchproject 56kbps 100computers80 83 ARPANETandMILNETsplit 85 86 NSFbuildsNSFNETasbackbone links6Supercomputercenters 1 5Mbps 10 000computers87 90 linkregionalnetworks NSI NASA ESNet DOE DARTnet TWBNet DARPA 100 000computers90 92 NSFNETmovesto45Mbps 16mid levelnetworks94 NSFbackbonedismantled multipleprivatebackbonesToday backbonesrunat2 5 10Gbps 10smillionscomputersin150countries GrowthoftheInternet NumberofHostsontheInternet Aug 1981213Oct 19841 024Dec 198728 174Oct 1990313 000Oct 19932 056 000Apr 19955 706 000Jul 199719 540 000Jul 200093 047 785 RecentGrowth 1991 2000 Internet发展规模和趋势 Internet的发展速度是历史上发展最快的一种技术以商业化后达到5000万用户为例电视用了13年 收音机用了38年 电话更长Internet从商业化后达到5000万用户用了4年时间Internet正在以超过摩尔定理的速度发展 网络时代的三大基本定律 摩尔定律 CPU性能18个月翻番 10年100倍 所有电子系统 包括电子通信系统 计算机 都适用 光纤定律 超摩尔定律 骨干网带宽9个月翻番 10年10000倍 带宽需求呈超高速增长的趋势 迈特卡尔夫定律 联网定律 网络价值随用户数平方成正比 未联网设备增加N倍 效率增加N倍 联网设备增加N倍 效率增加N2倍 网络带宽与CPU性能 光纤容量 高水平大容量光纤传输试验系统 Data Still OvertakingVoice 0 RelativeCapacity Voice Data Source MCI VintCerf InternationaldatatrafficalreadyexceedsinternationalvoicefromAustraliaandScandinavia 2000 1998 1996 IP Still ConqueringData AllOther IPX TCP IP TrafficRatios Source Gartner1997 100 90 80 70 60 50 40 30 20 10 0 RelativeUserPopulation 8 17 39 27 7 2 13 18 23 23 16 7 14 17 12 15 28 14 Internet Still GoingInteractive ToTransactionalPages Red andAudio VideoContent Purple 100 80 60 40 20 0 2000 1998 1996 Source TheYankeeGroup 1996 Apps Always DrivingCapacity SimpleVideo Multimedia Browsing PCMVoice IP PCS E Mail FileTransfer Paging VideoConferencing MPEG1NTSCVideo Telnet VoIP ISDN FrameRelay ATM POS T3 E3 T1 E1 NewModem WirelessWAN OldModem 004 0192 0288 128 1 5 3 155 Mb s MinimumBandwidthforApplicationperUser VirtualReality MedicalImaging Data Voice VideoTransportConvergence 高速信息网络的发展方向 通信与计算聚合 通信和计算技术的聚合改变了各自的原有特征高速信息网络体系结构的发展趋势分层结构 分布控制 管理和安全机制分层结构比特路层服务层应用层 高速信息网络的体系结构 比特路层主干网传输技术 SDH SONET 光纤主干网交换技术 IPoverSDH或光纤 GbE 支持IPv6端系统接入技术 LAN ADSL FTTH HFC服务层 支撑技术 全球统一的地址 域名 安全的系统管理和访问控制Browser Server计算模式 支持Data Voice Video以Java为代表的网络编程语言应用层 用户功能 用户 用户 立即响应 可适当延迟 用户 服务器 立即响应 可适当延迟 ATM SDH SONET IP Optical IPTransportAlternatives B ISDN IPoverATM IPoverSDH SONET IPoverOptical Long TermWinners IP ATM Optical IP SDH SONET Optical IP Optical Multiplexing Protection andManagementatEveryLayer EliminatingLayersLowersCosts WhoisWhoontheInternet InternetEngineeringTaskForce IETF TheIETFistheprotocolengineeringanddevelopmentarmoftheInternet Subdividedintomanyworkinggroups whichspecifyRequestForCommentsorRFCs IRTF InternetResearchTaskForce TheInternetResearchTaskForceisacomposedofanumberoffocused long termandsmallResearchGroups InternetArchitectureBoard IAB TheIABisresponsiblefordefiningtheoverallarchitectureoftheInternet providingguidanceandbroaddirectiontotheIETF TheInternetEngineeringSteeringGroup IESG TheIESGisresponsiblefortechnicalmanagementofIETFactivitiesandtheInternetstandardsprocess Standards ComposedoftheAreaDirectorsoftheIETFworkinggroups InternetStandardizationProcess AllstandardsoftheInternetarepublishedasRFC RequestforComments ButnotallRFCsareInternetStandards available http www ietf orgAtypical butnotonly wayofstandardizationis InternetDraftsRFCProposedStandardDraftStandard requires2workingimplementation InternetStandard declaredbyIAB DavidClark MIT 1992 Wereject kings presidents andvoting Webelievein roughconsensusandrunningcode ServicesProvidedbytheInternet Sharedaccesstocomputingresourcestelnet 1970 s Sharedaccesstodata filesFTP NFS AFS 1980 s Communicationmediumoverwhichpeopleinteractemail 1980 s on linechatrooms instantmessaging 1990 s audio video 1990 s replacingtelephonenetwork AmediumforinformationdisseminationUSENET 1980 s WWW 1990 s replacingnewspaper magazine audio video 1990 s replacingradio CD TV Today sVision Everythingisdigital voice video music picturesEverythingison line bankstatement medicalrecord books airlineschedule weather highwaytraffic toaster refrigerator Everyoneisconnected doctor teacher broker mother son friends enemies WhatisNext ElectroniccommercevirtualenterpriseInternetentertainmentinteractivesitcomWorldasasmallvillagecommunityorganizedaccordingtointerestsenhancedunderstandingamongdiversegroupsElectronicdemocracylittlepeoplecanvoicetheiropinionstothewholeworldbridgethegapbetweeninformationhavesandhaveno sElectronicterrorismhackercanbringthewholeworldtoitsknee IndustrialPlayers Telephonecompaniesownlong haulandaccesscommunicationlinks customersCablecompaniesownaccesslinksWireless SatellitecompaniesalternativecommunicationlinksUtilitycompanies power water railwayownrightofwaytolaydownmorewiresMediumcompaniesowncontentInternetServiceProvidersEquipmentcompaniesswitches routers chips optics computersSoftwarecompanies Backbone ISP ISP InternetPhysicalInfrastructure ResidentialAccessModemDSLCablemodemSatelliteLAN Enterprise ISPaccess BackbonetransmissionT1 T3 DS 1DS 3OC 3 OC 12ATMvs SONET vs WDM CampusnetworkEthernet ATMInternetServiceProvidersaccess regional backbonePointofPresence POP NetworkAccessPoint NAP LinksforLongHaulTransmission TypesoflinksT1 DS1 1 544MbpsT3 DS3 44 736MbpsSTS 1 OC 1 51 850MbpsSTS 3 OC 3 155 2MbpsSTS 12 OC 12 622 080MbpsSTS 48 OC 48 2 488GbpsSTS 192 OC 192 9 953GbpsHigherlevelsofservicesofferedcommerciallyFrameRelayATM PossibilitiesIPoverSDH SONETIPoverATMIPoverFrameRelayIPoverWDM Internet的成功经验 有远见的政府不断支持 1969 有风险的企业参与和投入 NFS MCI IBMvBNS MCI Abilene Qwest CISCO联合协作的开放式研究 IETF RFC教育和科研的示范网络为起点具有实验物理学的研究特点ARPAnet NSF ANS vBNS简单实用的技术路线 TCP IP Researchand Development Commercialization Partnerships Privatization NSFNET Internet2 Abilene vBNS AdvancedUSGovtNetworks ARPAnet gigabit testbeds Active Nets wireless WDM SprintLink InternetMCI USGovt Networks ANS Interoperable HighPerformance Research Education Networks 21stCentury Networking QualityofService QoS 主要内容 网络概述Internet的发展和成功经验计算机网络技术的历史回顾国际高速计算机网络研究计划中国高速计算机网络研究计划 Outline Thefirstdesignforapacket switchednetworkbyPaulBaraninearly60 sTCP IPprotocol onerealizationofthebasicprinciplesinBaran sdesignHowtheInternetlooksliketodayWhereitmaybemovingto PaulBaran sclassicalwork OnDistributedCommunications Time 1960 1964Goal buildingarobustcommunicationsystemthatcouldsurvivenuclearattacksOutcome apacket switchednetwork WhatisinBaran sDesign Aself adaptivesystem hot potatoroutingIfdon tknowbetter forwardingpacketstoallneighborsUpdateroutingtablebyobservingpacketspassingby oldroutingentriestimeoutanddeletedForwardpacketsASAPnotnecessarilyalongshortestpathsallthetime Learning adaptingtothechangingenvironment WhatisinBaran sDesign cont DatagramdeliveryEachswitchmakespacketforwardingdecisionbasedonitsownroutingtableEachpacketisforwardedindependentlyfromanyothersSwitcheskeepnostateaboutendnodesNotamostefficientnetworkDeliverywillnotbeperfect Endsystemsmusttolerate recoverfromtransmissionerrors WhatisinBaran sDesign cont Adistributedsystemallswitchnodesareequaleliminatinganysinglepointoffailurecomponentsmayfail thesystemmustnotsystemrobustnessthroughadequatephysicalredundancyadaptiverouting Simulationexperimentdemonstratedthat extremelysurvivablenetworkscanbebuiltusingamoderatelylowredundancyofconnectivitylevel PaulBaran 1964 somelessknownsideofthestory atthetimeanumberofpeopleintelecommunicationindustryconsideredBaran sproposal totallypreposterous theykicked screamed grumbled andworse Theirresponsetendedtobeemotional oftenwithanger rarelywithhumor outsiderscouldnotpossiblyunderstandthecomplexityoflargesystemslikethetelephonenetwork Differentapproachestosystemreliability phonesystematthetime stilltruetoday Dumbendnodes smartnetworkmakingeachandeverynetworkcomponentreliablesystemreliability componentreliabilityhighcomponentreliabilitybylocalredundancyeverythingexpectedtowork failuresexpectedtobeextremeexceptionsahumanconfigured tightlycontrolledsystemBaranproposalareliablesystembuiltoutofsimple unreliablepartsanadaptivesystemthatadjustsitselftochangesautomaticallySmartendstofixtransmissionerrors OneRealizationofBaran sDesing theInternet interconnectingheterogeneoussubnetstwobasicfunctionalities globallyuniqueaddressesdatagramdeliveryfromsourcestodestinationsviadynamicroutingClaim simple flexible scalable androbust IP sviewoftheworld IP Allkindsofsubnettechnologies allkindsofapplications allkindsoftransportprotocols Datagramsasthebasicbuildingblock SystemSimplicity EachpacketcarriesitsownaddressOneroutingtableservesalltrafficAnenablertotheexplosivegrowththesimpler thefewerchancestogowrongthesimpler theeasiertogrowtherequirementofleastcommonnetworkfunctionalitymaximizesthenumberofusablenetworks Datagramsasthebasicbuildingblock Flexibility runsoveranything IPstartedwithARPANET PRNET SATNET alllonggonetodayIPrunsover100MborGb sEthernet FDDI FrameRelay ATM SONET DWDM Supportsalldifferentapplicationsusedtobetelnet ftp emailonlynowaudio video web e commerce online education Whatwillthefuturebring Thebestbetwouldbetostayflexible Datagramsasthebasicbuildingblock Scalability ToscalethesystemmustbeabletogracefullyhandlethegrowthinthetotalnumberofendsystemsthegrowthintrafficvolumethegrowthinnetworksizelargerroutingtablesMorefrequentchanges WithIP thenetworkknowsnothingaboutindividualendapplications endapplicationsknownothingaboutnetworkinternals VanJacobsonmadeitpossibletoaggregateroutingtableentriesaccordingtoscalingneed Datagramsasthebasicbuildingblock Robustness self adaptivenatureofdynamicroutingfacilitatedthegrowthdynamicroutinganddatagramdeliverygohand in handperiodicroutingupdates areyoustillthere silentassumption existingpartsmayfail newpartsmayappearanytime changesconsideredthenormratherthanexceptionstradesoffoptimallinkusage e g headeroverhead updateoverhead forsystemrobustness Unintendedconsequence InacasualconversationNoelChiappasaid withoutDoD sInternetresearchinitiative Perhapswe dhavediscoveredcomputernetworkseventually butIhavenoideawhen It salsounclearifwe dhavethekindofnetworkwedoifithadn tbeenintendedforsuchause Aspiritofintenserobustness andtheabilitytokeepgoingnomatterwhat hasbeenpartofthe Internetethic sincedayone anditisdueinparttotherathersevereoperationenvironmentforwhichitwasoriginallyintended ThisextremelymonolithicnatureoftheInternetisoneofthethingswhichismakingithostiletocontrol andperhapsthatisinsomepartduetotheoriginalgoalsofthenetwork Whyatoughersystem thingsdogowrongfromtimetotime despiteallthehonesteffortstopreventthemHaveaguessonhowoftennetworkoutagehappens Getworseasthesystemgrowslarger Whatsortofchangeswouldbehelpfulinnextgenerationnetworks Ofcoursehigherdatarateandlowercostaredesirable butofallparametersIwouldoptforgreateremphasisonrobustness PaulBaran 1977 HowtheInternetlooksliketoday Nolongerthesameas20yearsagoAlotbiggerAlotmoreusersAlotmoreuseful alotmorevaluablelessrobust lessadaptive andlessconnected EffectivelywewereoutofIPv4addressspacesomeyearsagomostusersareconnectedviaNetworkAddressTranslator NAT now Latercomer NAT afeatureoraproblem NATsolvestheaddressexhaustionproblem andevenbringsabenefitofmorecontrol moresecurity NATsbrokeanumberofexistingprotocolsandapplicationswhichwerebuiltontheassumptionofIPaddressbeinggloballyuniqueLostabilitytosupportnewpeer to peerapplicationsend to endpacketdeliverypathbecomesaconcatenationofNATboxes effectivelyavirtualcircuit Internet Factsversusfairystories Overtimetheoriginalarchitecturehaseroded Entropy delay thathappenstoalllargesystems TorestoreIParchitecture movingtoIPv6fixestheaddressexhaustionproblemWhyIPv6hasnotbeenwidelydeployed a wedonotneedit groupgoingaroundtellingfairystoriesNATlovers cheaper easierthanmovingtov6bettingagainstaddressexhaustion IPv4addresseswouldlastustill2010 oreven2020 Howsoon really HowSoon WhendowerunoutIPv4addresses NATassuresyounever runout butendupwithafragmentedInternetIfIPv6needstobedeployed bettersoonerthanlaterWithexponentialgrowthofthenet aproblemignoredistwiceashardnextyear fixingitistwiceascostly Wait watchfornextcoupleyears 4timesashardandcostly Towrapup Thedatagram basedIParchitecturehasbeenanenablerforthenet sexplosivegrowthThenetneedstomovetoIPv6quicklytostopthestructuralerosionandenablefuturegrowththefactthattheInternethasnotcollapsedyetdoesnotmeanIPv4isOKandwilllastIP6fixestheaddressexhaustionproblem butthedeploymentwontbecheaporeasy 主要内容 网络概述Internet的发展和成功经验计算机网络技术的历史回顾国际高速计算机网络研究计划中国高速计算机网络研究计划 国际高速信息网络技术研究计划 1992年美国政府的 国家信息基础设施NII 1993年西方七国的 全球信息基础设施GII 美国NII组成部分 高性能计算和通信HPCC NGI和vBNSInternet2和AbileneTransPAC APAN STARTAPCANARIE和CA net3 NGI 美国下一代Internet研究计划 美国总统和副总统宣布启动NGI 1996 10 10 保证美国的重要利益保持技术领先 创造更多的就业和市场机会NGI的三个主要目标 先进网络技术的实验研究下一代网络测试床革命性的应用政府协调 计算机 信息和通信协调办公室CCIC可访问的Web站点 www ccic gov www ngi govNGI的进展 NGI实现计划 1998年2月 NGI概念文章 1997年7月 NGI目标1 先进网络技术的实验研究 网络工程规划和模拟 监视 集成 数据传递 网络管理 动态和自适应的网络服务质量 端到端 服务质量体系结构 允许控制 计费和优先权 可观察和控制的API DrillDown技术安全用户用安全和公平的方法服务网络资源 优越的网络管理 网络内部的监视 游动 远程访问 公钥基础设施 NGI目标2 下一代网络测试床 开发下一代网络测试床 用比现在Internet快100倍以上的速度连接至少100个大学和国家研究实验室以1997年1 54Mbps计 10个连接点速度达到比现在Internet快1000倍端到端连接速度达到100Mbps 1GbpsNGI目标2包括高性能连接 开发广域网结构 用100 Mbps速度连接100个广域点下一代网络技术和超高性能连接 开发超高速交换和传输技术 用1 Gbps速度连接10个以上的局域点主要策略 协调建立一个高性能的协作网络利用现有的网络试验床 vBNS ESnet NREN评价标准 连接点的数量 端到端的性能支持目标1的研究 支持目标3的应用 NGI目标3 革命性的网络应用 开发今天Internet没有 对国家重要的网络应用健康保健 远程医疗 紧急医疗响应支持教育 远程教育 数字图书馆科学研究 能源 地理系统 气象 生物国家安全 高性能全球通信 先进的信息传播环境 监测 预测 警告 响应政府 传递政府服务和信息给公民和企业突发事件 灾难响应 危机管理设计和制造 制造工程主要策略 重点研究基础性应用分布式计算应用 协同性应用 vBNSCooperativeAgreement CompetitivelyawardedinApril1995EstablishedbytheNSFinordertoensuretheavailabilityofhighperformancenetworkingresourcesfortheUSResearch EducationcommunityfostertheadvancementofnetworkingtechnologyNSFcontributes programmanagementandfundingMCIcontributes bandwidth equipment andengineering SanFrancisco NationalCenterforAtmosphericResearch SanDiegoSupercomputerCenter Houston Denver AmeritechNAP Chicago NationalCenterforSupercomputingApplications Cleveland Perryman MD SprintNAP MFSNAP PittsburghSupercomputingCenter LosAngeles A Atlanta A NewYorkCity vBNSBackboneNetworkMap Boston Washington DC Seattle A A C C C C C C C C C C C C C C C C C C J J AscendGRF400Cisco7507JuniperM40FOREASX 1000NAP A C DS 3OC 3COC 12COC 48 J SanFrancisco NationalCenterforAtmosphericResearch SanDiegoSupercomputerCenter Houston Denver Chicago NationalCenterforSupercomputingApplications Cleveland Perryman MD PittsburghSupercomputingCenter LosAngeles Atlanta NewYorkCity vBNSBackbone3Q99 Boston Washington DC Seattle Internet2 UCAID 120多个大学会员 的一项研究计划UniversityCorporationforAdvancedInternetDevelopment形成大学试验网 开发下一代Internet技术和应用IPv6 Multicasting QOS以竞争方式得到NGI计划的经费支持NGI是政府计划 Internet2是大学合作计划相互补充 相互依靠Internet2和NGI的合作范围NSF支持的vBNSInternet2将建立用于地区连接的gigaPoPInternet2的许多网络应用开发由NGI支持 AdvancedInternetBenefits RichercontentthroughhigherbandwidthVideo audioVirtualrealityDynamicnotstaticMoreinteractivityviaminimaldelayReliablecontentdeliverythroughqualityofservicemodel Internet2Applications Deliverqualitativeandquantitativeimprovementsintheconductof ResearchTeachingLearningRequireadvancednetworking VirtualLaboratories InteractiveresearchandinstructionReal timeaccesstoremotescientificinstruments ImagescourtesyoftheUniversityofMichigan VirtualLaboratories Real timeaccesstoremoteinstrumentsUniversityofPittsburgh PittsburghSupercomputingCenter3 DBrainMapping DigitalLibraries VideoandaudioIndianaUniversityVariationsProject DistributedComputation Multi sitedatabasesOldDominionUniversityChesapeakeBaySimulation ImagecourtesyofOldDominionUniversity Teleimmersion SharedvirtualrealityUniversityofIllinoisatChicagoVirtualTemporalBone ImagescourtesyUnivofIllinois Chicago Abilene Projectannounced14April1998byVPGoreMostadvancedandfarreachingresearchandeducationnetworkintheworldsupportadvancedresearchapplicationsintegratedadvancednetworkservicesDevelopedbyUCAIDQwest NortelandCiscocorporatepartnersAdvancednativeIPbackbonenetworkavailabletouniversitiesparticipatinginUCAID sInternet2project AbileneCharacteristics 2 4Gbps OC48 amonggigaPoPs increasingto9 6Gbps OC192 Connectionsat622Mbps OC12 or155Mbps OC3 IPoverSonettechnologyAccessPoPsveryclosetoalmostalloftheanticipateduniversitygigaPoPs AbileneSchedule Fall1998 Demonstratednetworkatmembermeeting inpre productionatseveraluniversi