《數(shù)據(jù)鏈路層》課件

Chapter5TheNetworkLayer01Introduction 6H02PhysicalLayer 6H03DataLinkLayer 6H04MACSublayer 8H05NetworkLayer 6H06TransportLayer

6H07ApplicationLayer 6H08NetworkSecurity 4HTotal 48HContentsIPLayer〔End-to-endPackets)應(yīng)用層運(yùn)輸層網(wǎng)絡(luò)層表示層會(huì)話層數(shù)據(jù)鏈路層物理層7654321OSI應(yīng)用層網(wǎng)絡(luò)接口層網(wǎng)際層IP(各種應(yīng)用層協(xié)議如TELNET,FTP,SMTP等)運(yùn)輸層(TCP

或

UDP)TCP/IP應(yīng)用層物理層網(wǎng)絡(luò)層IP(各種應(yīng)用層協(xié)議如TELNET,FTP,SMTP等)運(yùn)輸層(TCP

或

UDP)數(shù)據(jù)鏈路層HybridModel(textbookmodel)Thenetworklayeristhelowestlayerthatdealswithend-to-endtransmission.Itmust:

-Knowaboutthetopologyofthecommunicationsubnet;

-Takecaretochooseroutes,and

-Dealwithdataexchangebetweendifferentnetworks.5.1NetworkLayerDesignIssuesStore-and-ForwardPacketSwitchingServicesProvidedtotheTransportLayerImplementationofConnectionlessServiceImplementationofConnection-OrientedServiceComparisonofVirtual-CircuitandDatagramSubnetsStore-and-ForwardPacketSwitchingTheenvironmentofthenetworklayerprotocols.Ahostwithapackettosendtransmitsittothenearestrouter,eitheronitsownLANoroverapoint-to-pointlinktothecarrier.Thepacketisstoredthereuntilithasfullyarrivedsothechecksumcanbeverified.Thenitisforwardedtothenextrouteralongthepathuntilitreachesthedestinationhost.

fig5-1ServicesProvidedtotheTransportLayerThenetworkservicestothetransportlayergoals:Theservicesshouldbeindependentoftheroutertechnology.Thetransportlayershouldbeshieldedfromtherouterspresent.TheuniformaddressesplanacrossLANsandWANs.Connection-orientedorconnectionless?-therouters'jobismovingpacketsaround.Thesubnetisinherentlyunreliable.Therefore,thehostsshoulddoerrorcontrolandflowcontrolthemselves.Theconclusionisconnectionless,withprimitivesSENDPACKETandRECEIVEPACKETandlittleelse.-thesubnetserviceshouldbereliable,connection-oriented.ThesetwocampsareexemplifiedbytheInternetandATM.ImplementationofConnectionlessServiceRoutingwithinadiagramsubnet.ImplementationofConnection-OrientedServiceRoutingwithinavirtual-circuitsubnet.ComparisonofVirtual-CircuitandDatagramSubnets5-4trade-offs1.routermemoryspaceandbandwidth:VCallowpacketstocontaincircuitnumbersinsteadoffulldestinationaddresses.Itsavesbandwidth.Thepricepaidisthetablespacewithintherouters.2.setuptimeversusaddressparsingtime:UsingVCsrequiresasetupphase,needtimeandresources.However,routinginaVCsubnetiseasy.Inadatagramsubnet,amorecomplicatedlookupprocedureisrequired.3.amountoftablespacerequiredinroutermemory:Adatagramsubnetneedsanentryforeverypossibledestination,whereasavirtual-circuitsubnetjustneedsanentryforeachvirtualcircuit.4.qualityofserviceandavoidingcongestion:VCshavesomeadvantageswhentheconnectionisestablished.Withadatagramsubnet,congestionavoidanceismoredifficult.5.overheadrequiredtosetupandclearaVC:Ifthemajorityofthetrafficisexpectedtobetransaction,theuseofVCisnotwise.(permanentVCmaybeusefulhere.6.lossofacommunicationlineisfataltovirtualcircuitsusingitbutcanbeeasilycompensatedforifdatagramsareused.Datagramsalsoallowtherouterstobalancethetrafficthroughoutthesubnet,sinceroutescanbechangedpartwaythroughalongsequenceofpackettransmissions.中間設(shè)備又稱為中間系統(tǒng)或中繼(relay)系統(tǒng)。物理層中繼系統(tǒng)：轉(zhuǎn)發(fā)器(repeater)。數(shù)據(jù)鏈路層中繼系統(tǒng)：網(wǎng)橋或橋接器(bridge)。網(wǎng)絡(luò)層中繼系統(tǒng)：路由器(router)。網(wǎng)橋和路由器的混合物：橋路器(brouter)。網(wǎng)絡(luò)層以上的中繼系統(tǒng)：網(wǎng)關(guān)(gateway)。

網(wǎng)絡(luò)互相連接起來(lái)要使用一些中間設(shè)備當(dāng)中繼系統(tǒng)是轉(zhuǎn)發(fā)器或網(wǎng)橋時(shí)，一般并不稱之為網(wǎng)絡(luò)互連，因?yàn)檫@僅僅是把一個(gè)網(wǎng)絡(luò)擴(kuò)大了，而這仍然是一個(gè)網(wǎng)絡(luò)。

網(wǎng)關(guān)由于比較復(fù)雜，目前使用得較少。

互聯(lián)網(wǎng)都是指用路由器進(jìn)行互連的網(wǎng)絡(luò)。

由于歷史的原因，許多有關(guān)TCP/IP的文獻(xiàn)將網(wǎng)絡(luò)層使用的路由器稱為網(wǎng)關(guān)。

網(wǎng)絡(luò)互連使用路由器直接交付和間接交付間接交付間接交付間接交付ABC直接交付直接交付直接交付不需要使用路由器但間接交付就必須使用路由器TypicalRouter

路由選擇路由選擇處理機(jī)路由選擇協(xié)議路由表3輸入端口3交換結(jié)構(gòu)輸入端口輸出端口分組轉(zhuǎn)發(fā)轉(zhuǎn)發(fā)表分組處理輸出端口……11133122223——網(wǎng)絡(luò)層2——數(shù)據(jù)鏈路層1——物理層“轉(zhuǎn)發(fā)〞和“路由選擇〞的區(qū)別“轉(zhuǎn)發(fā)〞(forwarding)就是路由器根據(jù)轉(zhuǎn)發(fā)表將用戶的IP數(shù)據(jù)報(bào)從適宜的端口轉(zhuǎn)發(fā)出去?！奥酚蛇x擇〞(routing)那么是按照分布式算法，根據(jù)從各相鄰路由器得到的關(guān)于網(wǎng)絡(luò)拓?fù)涞淖兓闆r，動(dòng)態(tài)地改變所選擇的路由。路由表是根據(jù)路由選擇算法得出的。而轉(zhuǎn)發(fā)表是從路由表得出的。在討論路由選擇的原理時(shí)，往往不去區(qū)分轉(zhuǎn)發(fā)表和路由表的區(qū)別，輸入端口對(duì)線路上

收到的分組的處理數(shù)據(jù)鏈路層剝?nèi)撞亢臀膊亢螅瑢⒎纸M送到網(wǎng)絡(luò)層的隊(duì)列中排隊(duì)等待處理。這會(huì)產(chǎn)生一定的時(shí)延。物理層處理數(shù)據(jù)鏈路層處理網(wǎng)絡(luò)層處理分組排隊(duì)

交換結(jié)構(gòu)輸入端口的處理從線路接收分組查表和轉(zhuǎn)發(fā)輸出端口將交換結(jié)構(gòu)傳送來(lái)的分組發(fā)送到線路當(dāng)交換結(jié)構(gòu)傳送過(guò)來(lái)的分組先進(jìn)行緩存。數(shù)據(jù)鏈路層處理模塊將分組加上鏈路層的首部和尾部，交給物理層后發(fā)送到外部線路。物理層處理數(shù)據(jù)鏈路層處理網(wǎng)絡(luò)層處理分組排隊(duì)輸出端口的處理向線路發(fā)送分組緩存管理交換結(jié)構(gòu)分組丟棄假設(shè)路由器處理分組的速率趕不上分組進(jìn)入隊(duì)列的速率，那么隊(duì)列的存儲(chǔ)空間最終必定減少到零，這就使后面再進(jìn)入隊(duì)列的分組由于沒(méi)有存儲(chǔ)空間而只能被丟棄。路由器中的輸入或輸出隊(duì)列產(chǎn)生溢出是造成分組喪失的重要原因。5.2RoutingAlgorithmsTheOptimalityPrincipleShortestPathRoutingFloodingDistanceVectorRoutingLinkStateRoutingHierarchicalRoutingBroadcastRoutingMulticastRoutingRoutingforMobileHostsRoutinginAdHocNetworksConflictbetweenfairnessandoptimalityCorrectness,simplicity,robustness,stability,fairness,andoptimalityarepropertiesdesirableinaroutingalgorithm.However,someofthemareoftencontradictorygoals.Reducingthenumberofhopstendstoimprovethedelayandalsoreducetheamountofbandwidthconsumed,whichtendstoimprovethethroughputaswell.Routingalgorithmscanbegroupedintotwomajorclasses:

1.Innonadaptivealgorithms,thechoiceoftherouteiscomputedinadvance,off-line,anddownloadedtotherouterswhenthenetworkisbooted.Alsocalledstaticrouting.

2.Adaptivealgorithmschangetheirroutingdecisionstoreflectchangesinthenetwork(topologyandtraffic).Adaptivealgorithmsdifferinwheretheygettheirinformation(e.g.fromadjacentrouters,orfromallrouters),whentheychangetheroutes,andwhatmetricisusedforoptimization(e.g.,distance,numberofhops,orestimatedtransittime).Alsocalleddynamicrouting.TheOptimalityPrinciple(a)Asubnet.(b)AsinktreeforrouterB.“IfrouterJisontheoptimalpathfromrouterItorouterK,thentheoptimalpathfromJtoKalsofallsalongthesameroute〞.Consequentlythesetofoptimalroutesfromallsourcestoagivendestinationformatreerootedatthedestination.Suchatreeiscalledasinktree.Thegoalofallroutingalgorithmsistodiscoverandusethesinktreesforallrouters.i最正確路由不存在一種絕對(duì)的最正確路由算法。所謂“最正確〞只能是相對(duì)于某一種特定要求下得出的較為合理的選擇而已。實(shí)際的路由選擇算法，應(yīng)盡可能接近于理想的算法。路由選擇是個(gè)非常復(fù)雜的問(wèn)題它是網(wǎng)絡(luò)中的所有結(jié)點(diǎn)共同協(xié)調(diào)工作的結(jié)果。路由選擇的環(huán)境往往是不斷變化的，而這種變化有時(shí)無(wú)法事先知道。從路由算法的自適應(yīng)性考慮靜態(tài)路由選擇策略——即非自適應(yīng)路由選擇，其特點(diǎn)是簡(jiǎn)單和開(kāi)銷較小，但不能及時(shí)適應(yīng)網(wǎng)絡(luò)狀態(tài)的變化。動(dòng)態(tài)路由選擇策略——即自適應(yīng)路由選擇，其特點(diǎn)是能較好地適應(yīng)網(wǎng)絡(luò)狀態(tài)的變化，但實(shí)現(xiàn)起來(lái)較為復(fù)雜，開(kāi)銷也比較大。ShortestPathRoutingThefirst5stepsusedincomputingtheshortestpathfromAtoD.

Thearrowsindicatetheworkingnode.1425361112223355StepND(2)D(3)D(4)D(5)D(6)Initialization{1}251∞∞1{1,4}24①2∞2{1,4,5}231②43{1,2,4,5}②31244{1,2,3,4,5}2③1245{1,2,3,4,5,6}2312estinationNexthopDistance1--222343441542644結(jié)點(diǎn)1上路由表Q:計(jì)算以結(jié)點(diǎn)2為起點(diǎn)，到其他結(jié)點(diǎn)的最短路由以及相應(yīng)的轉(zhuǎn)發(fā)表。Anotherstaticalgorithmisflooding,inwhicheveryincomingpacketissentoutoneveryoutgoingline.Floodingcangeneratesvastnumbersofduplicatepackets.Onemeasuretocontroltheduplicationishopcounter.

Analternativetechniqueistokeeptrackofpacketsflooded,toavoidsendingthemoutasecondtime.Thenneedsalistpersourceroutertellingwhichsequencenumbershasalreadybeenseen.Anyincomingpacketonthelistisnotflooded.

Avariationisselectiveflooding.Therouterssendonlyonthoselinesthataregoingapproximatelyintherightdirection.

Floodingisnotpractical,butthetremendousrobustnessoffloodingdoeshavesomeuses.Inmilitary,indistributeddatabaseapplications,inwirelessnetworksandinuseasametricagainstwhichotherroutingalgorithmscanbecompared.Becausefloodingalwayschoosestheshortest.floodingModerncomputernetworksgenerallyusedynamicroutingalgorithmsratherthanthestaticonesbecausestaticalgorithmsdonottakethecurrentnetworkloadintoaccount.

Twodynamicalgorithmsinparticular,distancevectorroutingandlinkstaterouting,arethemostpopular.

Distancevectorroutingalgorithmsoperatebyhavingeachroutermaintainatable(i.e,avector)givingthebestknowndistancetoeachdestinationandwhichlinetousetogetthere.Thesetablesareupdatedbyexchanginginformationwiththeneighbors.

ThealgorithmwastheoriginalARPANETroutingalgorithmandwasalsousedintheInternetunderthenameRIP.1〕DistanceVectorRoutingAroutingtableindexedbyoneentryforeachrouterinthesubnet.Thisentrycontainstwoparts:thelinetouseforthatdestinationandanestimateofthetime/distancetothatdestination.Themetricusedmightbehops,timedelay,totalnumberofpacketsqueuedetc.

Assumethatdelayisusedasametric.OnceeveryTmseceachroutersendstoeachneighboralistofitsestimateddelaystoeachdestination.Italsoreceivesasimilarlistfromeachneighbor.IftherouterknowsthatthedelaytoXismmsec,italsoknowsthatitcanreachrouteriviaXinXi+mmsec.Byperformingthiscalculationforeachneighbor,aroutercanfindoutwhichestimateseemsthebestandusethatestimateandthecorrespondinglineinitsnewroutingtable.Notethattheoldroutingtableisnotusedinthecalculation.距離向量算法收到相鄰路由器〔其地址為X〕的一個(gè)RIP報(bào)文：(1)先修改此RIP報(bào)文中的所有工程：將“下一跳〞字段中的地址都改為X，并將所有的“距離〞字段的值加1。(2)對(duì)修改后的RIP報(bào)文中的每一個(gè)工程，重復(fù)以下步驟：假設(shè)工程中的目的網(wǎng)絡(luò)不在路由表中，那么將該工程加到路由表中。否那么假設(shè)下一跳字段給出的路由器地址是同樣的，那么將收到的項(xiàng) 目 替換原路由表中的工程。否那么假設(shè)收到工程中的距離小于路由表中的距離，那么進(jìn)行更新， 否那么，什么也不做。(3)假設(shè)3分鐘還沒(méi)有收到相鄰路由器的更新路由表，那么將此相鄰路 由器記為不可達(dá)的路由器，即將距離置為16〔距離為16表 示不可達(dá)〕。(4)返回。RIP〔RoutingInformationProtocol〕RIP基于Distance-Vector，使用UDP報(bào)文進(jìn)行路由信息的交換。RIP每隔30秒鐘發(fā)送一次路由刷新報(bào)文，如果在180秒內(nèi)收不到從某一網(wǎng)絡(luò)鄰居發(fā)來(lái)的路由刷新報(bào)文，那么將該網(wǎng)絡(luò)鄰居的所有路由標(biāo)記為不可達(dá)。如果在300秒之內(nèi)收不到從某一鄰居發(fā)來(lái)的路由刷新報(bào)文，那么將該網(wǎng)上鄰居的路由從路由表中去除。RIP-1不具備報(bào)文加密驗(yàn)證功能，而在RIP-2中實(shí)現(xiàn)了該功能。RIP使用跳數(shù)〔HopCount〕來(lái)衡量到達(dá)信宿機(jī)的距離，稱為路由權(quán)〔RoutingMetric〕。在RIP中，路由器到與它直接相連網(wǎng)絡(luò)的跳數(shù)為1，通過(guò)n個(gè)路由器可達(dá)的網(wǎng)絡(luò)的跳數(shù)為n+1，其余依此類推。為限制收斂時(shí)間，RIP規(guī)定metric取值0~15之間的整數(shù)，大于或等于16的跳數(shù)被定義為無(wú)窮大，即目的網(wǎng)絡(luò)或主機(jī)不可達(dá)。11

2131FEDCBA51612151314141611151一開(kāi)始，各路由表只有到相鄰路由器的信息網(wǎng)

3網(wǎng)

2網(wǎng)

4網(wǎng)

6網(wǎng)

5網(wǎng)

1“4〞表示“從本路由器到網(wǎng)4〞“1〞表示“距離是1〞“〞表示“直接交付〞112131FEDCBA51612151314141611151路由器B收到相鄰路由器A和C的路由表網(wǎng)

3網(wǎng)

2網(wǎng)

4網(wǎng)

6網(wǎng)

5網(wǎng)

1112131416112A22A314162C更新后A說(shuō)：“我到網(wǎng)1的距離是1。〞因此B現(xiàn)在也可以到網(wǎng)1，距離是2，經(jīng)過(guò)A。〞112131FEDCBA51612151314141611151路由器B收到相鄰路由器A和C的路由表網(wǎng)

3網(wǎng)

2網(wǎng)

4網(wǎng)

6網(wǎng)

5網(wǎng)

1112131416112A22A314162C更新后A說(shuō)：“我到網(wǎng)2的距離是1。〞因此B現(xiàn)在也可以到網(wǎng)2，距離是2，經(jīng)過(guò)A。〞最終所有的路由器的路由表都更新了FEDCBA11213142B52E63B1122A32A43A5162F12E22D33C42C516113B23B32B4152F61網(wǎng)2網(wǎng)6網(wǎng)5網(wǎng)1網(wǎng)3網(wǎng)412A2132A43A5162F12A22A314153C62CRIP協(xié)議的優(yōu)缺點(diǎn)RIP存在的一個(gè)問(wèn)題是當(dāng)網(wǎng)絡(luò)出現(xiàn)故障時(shí)，要經(jīng)過(guò)比較長(zhǎng)的時(shí)間才能將此信息傳送到所有的路由器。RIP協(xié)議最大的優(yōu)點(diǎn)就是實(shí)現(xiàn)簡(jiǎn)單，開(kāi)銷較小。RIP限制了網(wǎng)絡(luò)的規(guī)模，它能使用的最大距離為15〔16表示不可達(dá)〕。路由器之間交換的路由信息是路由器中的完整路由表，因而隨著網(wǎng)絡(luò)規(guī)模的擴(kuò)大，開(kāi)銷也就增加。R2R1網(wǎng)1網(wǎng)3網(wǎng)2正常情況11

12R1R1說(shuō)：“我到網(wǎng)1的距離是1，是直接交付。〞“1〞表示“從本路由器到網(wǎng)1〞“1〞表示“距離是1〞“〞表示“直接交付〞R2R1網(wǎng)1網(wǎng)3網(wǎng)2正常情況1112R1R2說(shuō)：“我到網(wǎng)1的距離是2，是經(jīng)過(guò)R1。〞“1〞表示“從本路由器到網(wǎng)1〞“2〞表示“距離是2〞“R1〞表示經(jīng)過(guò)R1R2R1網(wǎng)1網(wǎng)3網(wǎng)2R2R1網(wǎng)1網(wǎng)3網(wǎng)2網(wǎng)

1出了故障正常情況1111612R112R1R1說(shuō)：“我到網(wǎng)1的距離是16〔表示無(wú)法到達(dá)〕，是直接交付。〞但R2在收到R1的更新報(bào)文之前，還發(fā)送原來(lái)的報(bào)文，因?yàn)檫@時(shí)R2并不知道R1出了故障。R2R1網(wǎng)1網(wǎng)3網(wǎng)2R2R1網(wǎng)1網(wǎng)3網(wǎng)2網(wǎng)

1出了故障正常情況1111612R112R1R1收到R2的更新報(bào)文后，誤認(rèn)為可經(jīng)過(guò)R2到達(dá)網(wǎng)1，于是更新自己的路由表，說(shuō)：“我到網(wǎng)1的距離是3，下一跳經(jīng)過(guò)R2〞。然后將此更新信息發(fā)送給R2。13R2R2R1網(wǎng)1網(wǎng)3網(wǎng)2R2R1網(wǎng)1網(wǎng)3網(wǎng)2網(wǎng)

1出了故障正常情況1111612R112R1R2以后又更新自己的路由表為“1,4,R1〞，說(shuō)明“我到網(wǎng)1距離是4，下一跳經(jīng)過(guò)R1〞。13R214R1R2R1網(wǎng)1網(wǎng)3網(wǎng)2R2R1網(wǎng)1網(wǎng)3網(wǎng)2網(wǎng)

1出了故障正常情況11…11613R215R2116R212R112R114R1116R1…這樣不斷更新下去，直到R1和R2到網(wǎng)1的距離都增大到16時(shí)，R1和R2才知道網(wǎng)1是不可達(dá)的。這就是好消息傳播得快，而壞消息傳播得慢。網(wǎng)絡(luò)出故障的傳播時(shí)間往往需要較長(zhǎng)的時(shí)間(例如數(shù)分鐘)。這是RIP的一個(gè)主要缺點(diǎn)。Ref:IGRP協(xié)議Cisco公司在1980年代中期開(kāi)發(fā)了IGRP〔InteriorGatewayRoutingProtocol，內(nèi)部網(wǎng)關(guān)路由協(xié)議〕，是一個(gè)距離矢量路由協(xié)議，與RIP相比，IGRP具有以下特點(diǎn)：與RIP一樣，從所有配置接口上定期發(fā)出路由更新；IGRP使用的復(fù)合度量值來(lái)建立路由最正確路徑；支持單個(gè)目的地的非同等本錢負(fù)載均衡：最多6條；缺省情況下，IGRP每90秒發(fā)送一次路由更新播送；默認(rèn)最大跳數(shù)為100，但可以設(shè)置為255；快速收斂：觸發(fā)更新。IGRP復(fù)合度量值IGRP協(xié)議使用的復(fù)合度量值使它可以比RIP協(xié)議更加精確的評(píng)價(jià)路由。度量值包括的指標(biāo)有：Bandwidth—帶寬，單位是kbpsDelay—累積的接口延遲，單位是10微秒Reliability—數(shù)據(jù)包傳輸?shù)某晒β?，取值?到255.Loading—有效帶寬MTU—最大傳輸單元.2〕LinkStateRoutingTheideabehindlinkstateroutingcanbestatedasfiveparts.Eachroutermustdothefollowing:Discoveritsneighbors,learntheirnetworkaddress.Measurethedelayorcosttoeachofitsneighbors.Constructapackettellingallithasjustlearned.Sendthispackettoallotherrouters.Computetheshortestpathtoeveryotherrouter.Ineffect,thecompletetopologyandalldelaysareexperimentallymeasuredanddistributedtoeveryrouter.ThenDijkstra'salgorithmcanberuntofindtheshortestpathtoeveryotherrouter.LearningabouttheNeighbors(a)NineroutersandaLAN.(b)Agraphmodelof(a).Whenarouterisbooted,itsfirsttaskistolearnneighbors.ItsendsaspecialHELLOpacketoneachline.Therouterontheotherendisexpectedtosendbackareplytellingwhoitis.BuildingLinkStatePackets(a)Asubnet.(b)Thelinkstatepacketsforthissubnet.DistributingtheLinkStatePacketsThepacketbufferforrouterBinthepreviousslide.Theideaistousefloodingtodistributethelinkstatepackets.Tokeepincheck,eachpacketcontainsasequencenumberthatisincrementedforeachnewpacketsent.Oncearouterhasaccumulatedafullsetoflinkstatepackets,itconstructtheentiresubnetgraphusingDijkstra'salgorithm.鏈路狀態(tài)數(shù)據(jù)庫(kù)

(link-statedatabase)

由于各路由器之間頻繁地交換鏈路狀態(tài)信息，因此所有的路由器最終都能建立一個(gè)鏈路狀態(tài)數(shù)據(jù)庫(kù)。這個(gè)數(shù)據(jù)庫(kù)實(shí)際上就是全網(wǎng)的拓?fù)浣Y(jié)構(gòu)圖，它在全網(wǎng)范圍內(nèi)是一致的〔這稱為鏈路狀態(tài)數(shù)據(jù)庫(kù)的同步〕。OSPF的鏈路狀態(tài)數(shù)據(jù)庫(kù)能較快地進(jìn)行更新，使各個(gè)路由器能及時(shí)更新其路由表。OSPF的更新過(guò)程收斂得快是其重要優(yōu)點(diǎn)。ThefivetypesofOSPFmesseges5-66OSPF(OpenShortestPathFirst)OSPF的特點(diǎn)OSPF規(guī)定每隔一段時(shí)間〔30分鐘〕，要刷新一次數(shù)據(jù)庫(kù)中的鏈路狀態(tài)。由于一個(gè)路由器的鏈路狀態(tài)只涉及到與相鄰路由器的連通狀態(tài)，因而與整個(gè)互聯(lián)網(wǎng)的規(guī)模并無(wú)直接關(guān)系。因此當(dāng)互聯(lián)網(wǎng)規(guī)模很大時(shí)，OSPF協(xié)議要比距離向量協(xié)議RIP好得多。OSPF沒(méi)有“壞消息傳播得慢〞的問(wèn)題，據(jù)統(tǒng)計(jì)，其響應(yīng)網(wǎng)絡(luò)變化的時(shí)間小于100ms。OSPF不用UDP而是直接用IP數(shù)據(jù)報(bào)傳送，可見(jiàn)OSPF的位置在網(wǎng)絡(luò)層。數(shù)據(jù)報(bào)很短的另一好處是可以不必將長(zhǎng)的數(shù)據(jù)報(bào)分片傳送。分片傳送的數(shù)據(jù)報(bào)只要喪失一個(gè)，就無(wú)法組裝成原來(lái)的數(shù)據(jù)報(bào)，而整個(gè)數(shù)據(jù)報(bào)就必須重傳。OSPF的其他特點(diǎn)OSPF對(duì)不同的鏈路可根據(jù)IP分組的不同效勞類型TOS而設(shè)置成不同的代價(jià)。因此，OSPF對(duì)于不同類型的業(yè)務(wù)可計(jì)算出不同的路由。如果到同一個(gè)目的網(wǎng)絡(luò)有多條相同代價(jià)的路徑，那么可以將通信量分配給這幾條路徑。這叫作多路徑間的負(fù)載平衡。所有在OSPF路由器之間交換的分組都具有鑒別的功能。支持可變長(zhǎng)度的子網(wǎng)劃分和無(wú)分類編址CIDR。每一個(gè)鏈路狀態(tài)都帶上一個(gè)32bit的序號(hào)，序號(hào)越大狀態(tài)就越新。Distancevectorversuslink-statePacketsRoutingtableinAthesubnetThelinkstatepacketsDistancevectorFromA

A-B4C6D9E5F10A--BB4CB6DB9EE5FB10ForwardingtableD-V和L-S的比較D-V算法規(guī)定每隔一段時(shí)間〔30秒〕，與相鄰路由器交換一次D-V；OSPF規(guī)定環(huán)境改變才通知所有路由器發(fā)送一次L-S。不過(guò)隔一段時(shí)間〔30分鐘〕，要刷新發(fā)送L-S。D-V算法只考慮下一跳，沒(méi)有全局的概念，交給下一跳就完成任務(wù)，所以容易產(chǎn)生環(huán)路；L-S算法每個(gè)路由器可以根據(jù)網(wǎng)絡(luò)整體結(jié)構(gòu)決定路徑，所以不會(huì)產(chǎn)生環(huán)，OSPF也沒(méi)有“壞消息傳得慢〞的問(wèn)題。RIP用UDP數(shù)據(jù)報(bào)傳送；而OSPF直接用IP數(shù)據(jù)報(bào)傳送。HierarchicalRouting

(分層次的路由選擇協(xié)議)Asnetworkgrowinsizes,theroutersaredividedintowhatwewillcallregions,witheachrouterknowingitsownregion,butnothingaboutotherregions.Forhugenetworks,itmaybenecessarytogrouptheregionsintoclusters,theclustersintozones,thezonesintogroups,andsoon.Nextisanexampleofroutinginatwo-levelhierarchywithfiveregions.Thefullroutingtableforrouter1Ahas17entries.Thereareentriesforallthelocalrouters,andallotherregionscondensedintoasinglerouter,soalltrafficforregion2goesviathe1B-2Aline,buttherestoftheremotetrafficgoesviathe1C-3Bline.Hierarchicalroutinghasreducedthetablefrom17to7entries.Hierarchicalrouting因特網(wǎng)采用分層次的路由選擇協(xié)議。因特網(wǎng)的規(guī)模非常大。如果讓所有的路由器知道所有的網(wǎng)絡(luò)應(yīng)怎樣到達(dá)，那么這種路由表將非常大，處理起來(lái)也太花時(shí)間。而所有這些路由器之間交換路由信息所需的帶寬就會(huì)使因特網(wǎng)的通信鏈路飽和。許多單位不愿意外界了解自己?jiǎn)挝痪W(wǎng)絡(luò)的布局細(xì)節(jié)和本部門所采用的路由選擇協(xié)議〔這屬于本部門內(nèi)部的事情〕，但同時(shí)還希望連接到因特網(wǎng)上。3〕自治系統(tǒng)(autonomoussystem)Internet是以路由器為根底的網(wǎng)絡(luò)〔routerbasednetwork〕Internet將整個(gè)互聯(lián)網(wǎng)劃分為許多較小的自治系統(tǒng)AS。一個(gè)自治系統(tǒng)是一個(gè)互聯(lián)網(wǎng)，其最重要的特點(diǎn)就是自治系統(tǒng)有權(quán)自主地決定在本系統(tǒng)內(nèi)應(yīng)采用何種路由選擇協(xié)議。一個(gè)自治系統(tǒng)內(nèi)的所有網(wǎng)絡(luò)都屬于一個(gè)行政單位(例如，一個(gè)公司，一所大學(xué)，政府的一個(gè)部門，等等)來(lái)管轄。一個(gè)自治系統(tǒng)的所有路由器在本自治系統(tǒng)內(nèi)都必須是連通的。內(nèi)部網(wǎng)關(guān)協(xié)議IGP：具體的協(xié)議有多種，如RIP和OSPF等。外部網(wǎng)關(guān)協(xié)議EGP：目前使用的協(xié)議就是BGP。R1H1H2內(nèi)部網(wǎng)關(guān)協(xié)議IGP〔例如，RIP〕自治系統(tǒng)A自治系統(tǒng)B自治系統(tǒng)CIGPIGPIGPIGPIGPIGPIGPIGPIGPIGPIGPIGPEGPEGPEGP內(nèi)部網(wǎng)關(guān)協(xié)議IGP〔例如，OSPF〕外部網(wǎng)關(guān)協(xié)議EGP〔例如，BGP-4〕IGPR3R2自治系統(tǒng)和

內(nèi)部網(wǎng)關(guān)協(xié)議、外部網(wǎng)關(guān)協(xié)議因特網(wǎng)有兩大類路由選擇協(xié)議內(nèi)部網(wǎng)關(guān)協(xié)議IGP(InteriorGatewayProtocol)即在一個(gè)自治系統(tǒng)內(nèi)部使用的路由選擇協(xié)議。目前這類路由選擇協(xié)議使用得最多，如RIP和OSPF協(xié)議。外部網(wǎng)關(guān)協(xié)議EGP(ExternalGatewayProtocol)假設(shè)源站和目的站處在不同的自治系統(tǒng)中，當(dāng)數(shù)據(jù)報(bào)傳到一個(gè)自治系統(tǒng)的邊界時(shí)，就需要使用一種協(xié)議將路由選擇信息傳遞到另一個(gè)自治系統(tǒng)中。這樣的協(xié)議就是外部網(wǎng)關(guān)協(xié)議EGP。在外部網(wǎng)關(guān)協(xié)議中目前使用最多的是BGP-4。4〕BGP(BorderGatewayProtocol)

–TheExteriorGatewayRoutingProtocol(a)AsetofBGProuters.(b)InformationsenttoF.BGP使用的環(huán)境卻不同因特網(wǎng)的規(guī)模太大，使得自治系統(tǒng)之間路由選擇非常困難。對(duì)于自治系統(tǒng)之間的路由選擇，要尋找最正確路由是很不現(xiàn)實(shí)的。自治系統(tǒng)之間的路由選擇必須考慮有關(guān)策略。因此，邊界網(wǎng)關(guān)協(xié)議BGP只能是力求尋找一條能夠到達(dá)目的網(wǎng)絡(luò)且比較好的路由〔不能兜圈子〕，而并非要尋找一條最正確路由。BGP發(fā)言人每一個(gè)自治系統(tǒng)的管理員要選擇至少一個(gè)路由器作為該自治系統(tǒng)的“BGP發(fā)言人〞。一般說(shuō)來(lái)，兩個(gè)BGP發(fā)言人都是通過(guò)一個(gè)共享網(wǎng)絡(luò)連接在一起的，而B(niǎo)GP發(fā)言人往往就是BGP邊界路由器，但也可以不是BGP邊界路由器。一個(gè)BGP發(fā)言人與其他自治系統(tǒng)中的BGP發(fā)言人要交換路由信息，就要先建立TCP連接，然后在此連接上交換BGP報(bào)文以建立BGP會(huì)話(session)，利用BGP會(huì)話交換路由信息。使用TCP連接能提供可靠的效勞，也簡(jiǎn)化了路由選擇協(xié)議。使用TCP連接交換路由信息的兩個(gè)BGP發(fā)言人，彼此成為對(duì)方的鄰站或?qū)Φ日?。路由routingStatic動(dòng)態(tài)Dynamic/AdaptiveEGP(BGP-4)

外部網(wǎng)關(guān)協(xié)議IGP內(nèi)部網(wǎng)關(guān)協(xié)議

RIP/IGRP(距離向量法〕OSPF〔鏈路狀態(tài)法〕RoutingSummary(flooding〕5〕RoutingforMobileHostsAWANtowhichLANs,MANs,andwirelesscellsareattached.Eachareahasoneormoreforeignagents,keepingtrackofallmobilehostsvisitingthearea.Eachareahasahomeagent,keepingtrackofhostswhosehomeinthearea,butvisitinganotherarea.Hostshaveapermanenthomelocationandaddress.Amobilehostenteringanareamustfirstregisteritselfwiththeforeignagentthere,likethis:

--Periodically,eachforeignagentbroadcastsitsexistenceandaddress.Anewly-arrivedmobilehostmaybroadcastaskingforanyforeignagents.

--Themobilehostregisterswiththeforeignagent,givingitshomeaddress,currentdatalinklayeraddress,andsomesecurityinformation.

Theforeignagentcontactsthemobilehost'shomeagent,tellingownaddressandalsothesecurityinformation.

--Thehomeagentexaminesthesecurityinformationandatimestamp.Ifitisfresh,ittellstheforeignagenttoproceed.

--Thentheforeignagentmakesanentryinitstablesandinformsthemobilehostthatitisnowregistered.

--Ideally,leavinganareawithaderegistration.

Whenapacketissenttoamobilehost,itisroutedtothehost'shomeLAN.Thehomeagentthenfindsthemobilehost'snewlocationviatheforeignagenthandlinghim.Packetroutingformobileusers.5.3CongestionControlAlgorithmsGeneralPrinciplesofCongestionControlCongestionPreventionPoliciesCongestionControlinVirtual-CircuitSubnetsCongestionControlinDatagramSubnetsLoadSheddingJitterControlCongestionWhentoomuchtrafficisoffered,congestionsetsinandperformancedegradessharply.GeneralPrinciplesofCongestionControlTwogroupsofsystemscontrolsolutions:openloopandclosedloop.Open-loopcontrolmakedecisionswithoutregardtothecurrentstateofthenetwork.Closedloopsolutionsarebasedontheconceptofafeedbackloopwiththreeparts:Monitorthesystem.detectwhenandwherecongestionoccurs.(suchas:thepercentageofallpacketsdiscarded,thequeuelengths,thenumberoftimeoutpacketst,theaveragepacketdelay,andthestandarddeviationofpacketdelay)Passinformationtowhereactioncanbetaken.Adjustsystemoperationtocorrecttheproblem.CongestionPreventionPoliciesPoliciesthataffectcongestion.5-26CongestionControlinVirtual-CircuitSubnets(a)Acongestedsubnet.(b)Aredrawnsubnet,eliminatescongestionandavirtualcircuitfromAtoB.CongestionControlinDatagramSubnetsTheWarningBitassociatewitheachlineavariableutoreflecttherecentutilization.Wheneverumovesabovethethreshold,theoutputlineentersa''warning''state.(bysettingaspecialbitinthepacket'sheader).Whenthepacketarrivedatitsdestination,thetransportentitycopiedthebitintothenextacknowledgementsentbacktothesource.Thesourcethencutbackontraffic.Asrouterinthewarningstate,itcontinuedtosetthewarningbit.Thesourcemonitoredthefractionofacknowledgementswiththebitsetandadjusteditstransmissionrateaccordingly.Aslongasthewarningbitscontinuedtoflowin,thesourcecontinuedtodecreaseitstransmissionrate.Hop-by-HopChokePackets(a)Achokepacketthataffectsonlythesource.(b)Achokepacketthataffectseachhopitpassesthrough.LoadsheddingArouterjustpickpacketsatrandomtodrop.anoldpacketisworthmorethananewoneand,anewpacketismoreimportantthananoldone.Forfiletransfer,oldisbetterthannewandformultimedia,newisbetterthanold.Toimplementanintelligentdiscardpolicy,applicationsmustmarktheirpacketsinpriorityclasses.Anotheroptionistoallowhoststoexceedthelimitsandallexcesstrafficbemarkedaslowpriority.RandomEarlyDetectionRED(RandomEarlyDetection).routersmaintainarunningaverageoftheirqueuelengths.therouterpickingapacketatrandomanddiscardit.Thisonlyworkswhensourcesrespondtolostpacketsbyslowingdowntheirtransmissionrate.JitterControl(a)Highjitter.(b)Lowjitter.5.4QualityofServiceRequirementsTechniquesforAchievingGoodQualityofServiceIntegratedServicesDifferentiatedServicesLabelSwitchingandMPLSRequirementsHowstringentthequality-of-servicerequirementsare.5-30TechniquesforAchievingGoodQualityofService1.Overprovisioning

2.Buffering-Flowsbuffereddoesnotaffectthereliabilityorbandwidth,andincreasesthedelay,butitsmoothsoutthejitter.

3.TrafficShaping-aboutregulatingtheaveragerate(andburstiness)ofdatatransmission

4.TheLeakyBucketAlgorithm

BufferingSmoothingtheoutputstreambybufferingpackets.TheLeakyBucketAlgorithm(a)Aleakybucketwithwater.(b)aleakybucketwithpackets.PacketScheduling(a)ArouterwithfivepacketsqueuedforlineO.(b)Finishingtimesforthefivepackets.LabelSwitchingandMPLS

(MultiProtocolLabelSwitching)TransmittingaTCPsegmentusingIP,MPLS,andPPP.5.6TheNetworkLayerintheInternetTheIPProtocolIPAddressesInternetControlProtocolsOSPF–TheInteriorGatewayRoutingProtocolBGP–TheExteriorGatewayRoutingProtocolInternetMulticastingMobileIPIPv6NetworkDesign:top10principlesMakesureitworks.

Keepitsimple.Whenindoubt,usethesimplestsolution.Makeclearchoices.Havingtwoormorewaystodothesamethingislookingfortrouble.Exploitmodularity.Expectheterogeneity.Thenetworkdesignmustbesimple,general,andflexible.Avoidstaticoptionsandparameters.Ifparametersareunavoidable,itisbesttohavethesenderandreceivernegotiateavaluethandefiningfixedchoices.Lookforagooddesign;itneednotbeperfect.Thedesignersshouldgowiththegooddesignandputtheextraburdenonthepeoplewiththeweirdspecialrequirements.Bestrictwhensendingandtolerantwhenreceiving.Thinkaboutscalability.Considerperformanceandcost.Ifanetworkhaspoorperformanceoroutrageouscosts,nobodywilluseit.CollectionofSubnetworksTheInternetisaninterconnectedcollectionofmanynetworks.TheIPProtocolTheIPv4(InternetProtocol)header.(TypeofService:Priority(3bit)+Delay〔D〕+Throughput〔T〕+Reliability〔R〕SomeoftheIPoptions5-54IPAddressesIPaddressformats.classA,128networkswith16millionhostseach,classB,16,384networkswithupto64Khosts,classC,2millionnetworkswithupto256hosts.SpecialIPaddresses點(diǎn)分十進(jìn)制記法10000000000010110000001100011111機(jī)器中存放的IP地址是32bit二進(jìn)制代碼10000000000010110000001100011111每隔8bit插入一個(gè)空格能夠提高可讀性采用點(diǎn)分十進(jìn)制記法那么進(jìn)一步提高可讀性1128

11331將每8bit的二進(jìn)制數(shù)轉(zhuǎn)換為十進(jìn)制數(shù)常用的三種類別的IP地址IP地址的使用范圍

網(wǎng)絡(luò)最大第一個(gè)最后一個(gè)每個(gè)網(wǎng)絡(luò)類別網(wǎng)絡(luò)數(shù)可用的可用的中最大的網(wǎng)絡(luò)號(hào)網(wǎng)絡(luò)號(hào)主機(jī)數(shù)A126(27

–2)112616,777,214B16,384(214)128.0191.25565,534C2,097,152(221)192.0.0223.255.255254IP地址的一些重要特點(diǎn)

(1)IP地址是一種分等級(jí)的地址結(jié)構(gòu)。分兩個(gè)等級(jí)的好處是：第一，IP地址管理機(jī)構(gòu)在分配IP地址時(shí)只分配網(wǎng)絡(luò)號(hào)，而剩下的主機(jī)號(hào)那么由得到該網(wǎng)絡(luò)號(hào)的單位自行分配。這樣就方便了IP地址的管理。第二，路由器僅根據(jù)目的主機(jī)所連接的網(wǎng)絡(luò)號(hào)來(lái)轉(zhuǎn)發(fā)分組〔而不考慮目的主機(jī)號(hào)〕，這樣就可以使路由表中的工程數(shù)大幅度減少，從而減小了路由表所占的存儲(chǔ)空間。(2)實(shí)際上IP地址是標(biāo)志一個(gè)主機(jī)〔或路由器〕和一條鏈路的接口。當(dāng)一個(gè)主機(jī)同時(shí)連接到兩個(gè)網(wǎng)絡(luò)上時(shí)，該主機(jī)就必須同時(shí)具有兩個(gè)相應(yīng)的IP地址，其網(wǎng)絡(luò)號(hào)net-id必須是不同的。這種主機(jī)稱為多接口主機(jī)(multihomedhost)。由于一個(gè)路由器至少應(yīng)當(dāng)連接到兩個(gè)網(wǎng)絡(luò)〔這樣它才能將IP數(shù)據(jù)報(bào)從一個(gè)網(wǎng)絡(luò)轉(zhuǎn)發(fā)到另一個(gè)網(wǎng)絡(luò)〕，因此一個(gè)路由器至少應(yīng)當(dāng)有兩個(gè)不同的IP地址。(3)用轉(zhuǎn)發(fā)器或網(wǎng)橋連接起來(lái)的假設(shè)干個(gè)局域網(wǎng)仍為一個(gè)網(wǎng)絡(luò)，因此這些局域網(wǎng)都具有同樣的網(wǎng)絡(luò)號(hào)net-id。(4)所有分配到網(wǎng)絡(luò)號(hào)net-id的網(wǎng)絡(luò)，范圍很小的局域網(wǎng)，還是可能覆蓋很大地理范圍的廣域網(wǎng)，都是平等的。IP地址與硬件地址TCP報(bào)文IP數(shù)據(jù)報(bào)MAC幀應(yīng)用層數(shù)據(jù)首部首部尾部首部鏈路層及以下使用硬件地址硬件地址網(wǎng)絡(luò)層及以上使用IP地址IP地址SubnetsAcampusnetworkconsistingofLANsforvariousdepartments.AclassBnetworksubnettedinto64subnetsthesubnetmaskcanbewrittenas.Alternativenotationis/22(indicatethatthesubnetmaskis22bitslong)IP地址::={<網(wǎng)絡(luò)號(hào)>,<子網(wǎng)號(hào)>,<主機(jī)號(hào)>

劃分子網(wǎng)的根本思路劃分子網(wǎng)純屬一個(gè)單位內(nèi)部的事情。單位對(duì)外仍然表現(xiàn)為沒(méi)有劃分子網(wǎng)的網(wǎng)絡(luò)。從主機(jī)號(hào)借用假設(shè)干個(gè)比特作為子網(wǎng)號(hào)subnet-id，而主機(jī)號(hào)host-id也就相應(yīng)減少了假設(shè)干個(gè)比特。但凡從其他網(wǎng)絡(luò)發(fā)送給本單位某個(gè)主機(jī)的IP數(shù)據(jù)報(bào)，仍然是根據(jù)IP數(shù)據(jù)報(bào)的目的網(wǎng)絡(luò)號(hào)net-id，先找到連接在本單位網(wǎng)絡(luò)上的路由器。然后此路由器在收到IP數(shù)據(jù)報(bào)后，再按目的網(wǎng)絡(luò)號(hào)net-id和子網(wǎng)號(hào)subnet-id找到目的子網(wǎng)。最后就將IP數(shù)據(jù)報(bào)直接交付給目的主機(jī)。劃分子網(wǎng)后變成了三級(jí)結(jié)構(gòu)當(dāng)沒(méi)有劃分子網(wǎng)時(shí)，IP地址是兩級(jí)結(jié)構(gòu)，地址的網(wǎng)絡(luò)號(hào)字段也就是IP地址的“因特網(wǎng)局部〞，而主機(jī)號(hào)字段是IP地址的“本地局部〞。

劃分子網(wǎng)后IP地址就變成了三級(jí)結(jié)構(gòu)。劃分子網(wǎng)只是將IP地址的本地局部進(jìn)行再劃分，而不改變IP地址的因特網(wǎng)局部。

從一個(gè)IP數(shù)據(jù)報(bào)的首部并無(wú)法判斷源主機(jī)或目的主機(jī)所連接的網(wǎng)絡(luò)是否進(jìn)行了子網(wǎng)的劃分。

使用子網(wǎng)掩碼(subnetmask)可以找出IP地址中的子網(wǎng)局部。IP地址的各字段和子網(wǎng)掩碼網(wǎng)絡(luò)號(hào)net-id主機(jī)號(hào)host-id兩級(jí)IP地址網(wǎng)絡(luò)號(hào)net-idhost-id三級(jí)IP地址主機(jī)號(hào)subnet-id子網(wǎng)號(hào)子網(wǎng)掩碼因特網(wǎng)局部本地局部因特網(wǎng)局部本地局部劃分子網(wǎng)時(shí)的網(wǎng)絡(luò)地址1111111111111111

1111111100000000net-idsubnet-idhost-id為全0(IP

地址)AND(子網(wǎng)掩碼)=

網(wǎng)絡(luò)地址網(wǎng)絡(luò)號(hào)net-id主機(jī)號(hào)host-id兩級(jí)IP地址網(wǎng)絡(luò)號(hào)三級(jí)IP地址主機(jī)號(hào)net-idhost-idsubnet-id子網(wǎng)號(hào)子網(wǎng)掩碼因特網(wǎng)局部本地局部因特網(wǎng)局部本