1、配置PfRv2有静态路由和策略基于路由的数据流控制控制机制 目录简介先决条件要求使用的组件配置网络图配置验证案件1:Parent路由通过在边界路由器的静态路由学习案件2:Parent路由通过OSPF学习相关的思科支持社区讨论简介本文如何描述PfRv2 (性能路由)根据PfRv2政策决策的控制数据流。本文在PfRv2讨论使用静态路由和策略基于路由。 先决条件要求思科建议您有基础知识性能路由(PfR)。使用的组件配置PfRv2允许网络管理员配置策略和根据PfRv2策略结果相应地路由流量。有PfRv2控制数据流和取决于协议目的地前缀的parent路由学习的多种模式。PfRv2能够更改路由信息库(RI
2、B)由操作的路由协议，注入静态路由或通过动态策略基于路由。如果parent路由通过BGP学习，使用属性类似本地首选， PfRv2能动态地操作路由。a71如果parent路由通过EIGRP学习， PfRv2在EIGRP拓扑表里能注入一个新的路由。a71如果parent路由通过静态路由学习， PfR2注入在PfR的更多specific(better)路由选择边界路由器(BR)。a71如果parent路由通过不在上述三机制学习， PfRv2使用基于策略的路由(PBR)推送在选定BR的a71流量。 此条款讨论PfRv2使用静态路由(当parent路由是通过静态路由)时和PBR (当在RIB的paren
3、t路由是通过RIP、OSPF， ISIS等)对控制流量。网络图本文将参考跟随的镜像作为topolgy的示例为本文的其余。R1- R3- PfRR4 & R5- PfRR9 & R10R1配置在应用程序的(APPLICATION-LEARN-LIST)此scenatio两请学习列表将配置，一和数据(DATA-LEARN-LIST)流量。此方案使用一张前缀列表定义流量。access-list能也使用匹配流量类型类似TCP，UDP、ICMP等DSCP和TOS能也使用定义您的流量。key chain pfrkey 0key-string ciscopfr masterpolicy-rules PFR!
4、border 10.4.4.4 key-chain pfrinterface Tunnel0 internalinterface Ethernet1/0 externalinterface Ethernet1/2 internallink-group MPLS!border 10.5.5.5 key-chain pfrinterface Tunnel0 internalinterface Ethernet1/3 internalinterface Ethernet1/0 externallink-group INET!learntraffic-class filter access-list
5、DENY-ALLlist seq 10 refname APPLICATION-LEARN-LIST /Learn-list for application traffictraffic-class prefix-list APPLICATIONthroughputlist seq 20 refname DATA-LEARN-LIST /Learn-list for data traffictraffic-class prefix-list DATAthroughput!pfr-map PFR 10match pfr learn list APPLICATION-LEARN-LISTset p
6、eriodic 90set delay threshold 25set mode monitor activeset active-probe echo 10.20.21.1set probe frequency 5set link-group MPLS fallback INET!pfr-map PFR 20match pfr learn list DATA-LEARN-LIST set periodic 90set delay threshold 25set mode monitor activeset resolve delay priority 1 variance 10set act
7、ive-probe echo 10.30.31.1set probe frequency 5set link-group INET fallback MPLSip prefix-list DATAseq 5 permit 10.30.0.0/24ip prefix-list APPLICATIONseq 5 permit 10.20.0.0/24验证案件1:Parent路由通过在边界路由器的静态路由学习在此方案中，流量为目的地10.20.20.1和10.30.30.1流。下面是parent路由如何看起来象在R4和R5。R4#show ip route-output suppressed-S 1
8、0.20.0.0/16 1/0 via 10.0.68.8S 10.30.0.0/16 1/0 via 10.0.68.8R5#show ip route-output suppressed-S 10.20.0.0/16 1/0 via 10.0.57.7S 10.30.0.0/16 1/0 via 10.0.57.7当通信流， PfRv2学习时流量前缀和流量在输出中落入INPOLICY状态如下所示。R3#show pfr master traffic-classOER Prefix Statistics:-output suppressed-DstPrefix Appl_ID Dscp Pr
9、ot SrcPort DstPort SrcPrefix Flags State Time CurrBR CurrI/F ProtocolPasSDly PasLDly PasSUn PasLUn PasSLos PasLLos EBw IBwActSDly ActLDly ActSUn ActLUn ActSJit ActPMOS ActSLos ActLLos-10.20.20.0/24 N N N N N N INPOLICY 31 10.4.4.4 Et1/0 STATIC N N N N N N N N1 2 0 0 N N N N10.30.30.0/24 N N N N N N
10、INPOLICY 30 10.5.5.5 Et1/0 STATIC N N N N N N N N4 2 0 0 N N N N如下所示该R4 (10.4.4.4)路由器注入了具体的路由10.20.20.0/24。此自动生成的路由用标记值为5000自动地标记。此更加特定的佳路由做R4作为离开为10.20.20.0/24的流量的更加好的BR。R4#show pfr border routes staticFlags: C - Controlled by oer, X - Path is excluded from control,E - The control is exact, N - The
11、 control is non-exactFlags Network Parent Tag CE 10.20.20.0/24 10.20.0.0/16 5000 XN 10.30.30.0/24 R4#show ip route 10.20.20.0 255.255.255.0Routing entry for 10.20.20.0/24Known via “static“, distance 1, metric 0Tag 5000Redistributing via ospf 100 Routing Descriptor Blocks: * 10.0.46.6, via Ethernet1/
12、0Route metric is 0, traffic share count is 1Route tag 5000 同样相似的行为在R5能被看到，并且有标记5000的注入具体的路由10.30.30.0/24。这做R5一适当的候选对10.30.30.0/24的路由流量。这是PfRv2如上所述路由的prefer流量在“如何显示pfr重要的数据流类别”。R5#show pfr border routes staticFlags: C - Controlled by oer, X - Path is excluded from control,E - The control is exact, N
13、- The control is non-exactFlags Network Parent Tag XN 10.20.20.0/24 CE 10.30.30.0/24 10.30.0.0/16 5000 R5#show ip route 10.30.30.0 255.255.255.0Routing entry for 10.30.30.0/24Known via “static“, distance 1, metric 0Tag 5000Redistributing via ospf 100Routing Descriptor Blocks:* 10.0.57.7, via Etherne
14、t1/0Route metric is 0, traffic share count is 1Route tag 5000在事件有广泛边界路由器(类似在这种情况下)，这些自动生成的静态路由必须手工再分布到IGP至于它可能到达其他边界路由器，并且他们可能根据具体的路由的路由流量生成由选定BR。案件2:Parent路由通过OSPF学习没有通过BGP、EIGRP或者静态路由学习使用策略基于routing(PBR)的所有parent路由被控制。PfRv2注入动态路由路线图和access-list对控制流量。下面是OSPF parent路由如何看起来象在R4和R5。R4#show ip route-ou
15、tput suppressed-O E2 10.20.0.0/16 110/20 via 10.0.46.6, 02:16:35, Ethernet1/0O E2 10.30.0.0/16 110/20 via 10.0.46.6, 02:16:35, Ethernet1/0R5#show ip route-output suppressed-O E2 10.20.0.0/16 110/20 via 10.0.57.7, 02:18:20, Ethernet1/0O E2 10.30.0.0/16 110/20 via 10.0.57.7, 02:18:20, Ethernet1/0当PfRv
16、2必须操作通信流通过策略基于路由时，直接地要求BR之间的一个连接的接口。此连链路可能直接地是物理连接或它可能是GRE隧道。此通道必须手工创建和配置，在PfRv2边界定义的内部接口。R4#show ip route-output suppressed-O E2 10.20.0.0/16 110/20 via 10.0.46.6, 02:16:35, Ethernet1/0O E2 10.30.0.0/16 110/20 via 10.0.46.6, 02:16:35, Ethernet1/0R5#show ip route-output suppressed-O E2 10.20.0.0/16
17、110/20 via 10.0.57.7, 02:18:20, Ethernet1/0O E2 10.30.0.0/16 110/20 via 10.0.57.7, 02:18:20, Ethernet1/0根据PfRv2定义的策略，它用最好的退出路由器(BR)出来10.20.20.0/24和10.30.30.0/24的。例如在事件，当为10.20.20.0/24注定的流量来到不是选定BR的R5时(10.5.5.5)，动态路由路线图和access-list自动地被注入到策略路由流量选定BR R4 (10.4.4.4)。在定义前的隧道接口是策略路由的数据包。R4#show ip route-output suppressed-O E2 10.20.0.0/16 110/20 via 10.0.46.6, 02:16:35, Ethernet1/0O E2 10.30.0.0/16 110/20 via 10.0.46.6, 02:16:35, Ethernet1/0R5#show ip route-output suppressed-O E2 10.20.0.0/16 110/20 via 10.0.57.7, 02:18:20, Ethernet1/0O E2 10.30.0.0/16 110/20 via 10.0.57.7, 02:18:20, Ethernet1/0
