Scaling the Service Provider NGN with unified MPLS

Transcription

1 Scaling the Service Provider NGN with unified MPLS BRKSPG-3305 Istvan Kakonyi Vertical Solutions Architect

2 Housekeeping We value your feedback- don't forget to complete your online session evaluations after each session & the Overall Conference Evaluation which will be available online from Thursday Visit the World of Solutions and Meet the Engineer Visit the Cisco Store to purchase your recommended readings Please switch off your mobile phones After the event don t forget to visit Cisco Live Virtual: 2

3 Agenda Introduction Problem Statement MPLS Network Architecture Evolution Unified MPLS Architecture - Using BGP+Labels for E2E LSP Creation - Architecture, control and data plane operations - Configuration Examples Further Developments Q & A 3

4 Introduction

5 Strategic Business Value New Packet Based Services Entering the Market Application Services On-Demand, Cloud Services Network Services 1997~ IP Communications Services 2002~ Collaborate 2007~ Compute Unified Communications B2B TelePresence Hosted UC Network Based Security XaaS Managed WebEx Cloud Services Mobile Collaboration Services Connectivity Managed Router SIP Trunking Managed Firewall Managed VPN WAN Optimization WAAS Application Fluent Services Time 5

6 Bandwidth Explosion: Dominance of Video, Mobile, and Cloud Fundamentally Different Traffic Mix IP traffic will increase 4X (767 exabytes by 2014) Video & Multimedia 90% Consumer Traffic Mobile Internet 39X Traffic Increase IT Services via Cloud 2009 Global IP Traffic 2014 Source: Cisco Visual Networking Index Forecast, % Market Growth Exponential Growth + Evolving Traffic Mix = Complexity 6

7 SP networks today and going forward Evolution to MPLS Ethernet Policy and Service Control Plane (per subscriber) Subscriber Access Edge Core Business L2SE Ethernet Aggregation L2SE L3SE Corporate DSLAM L2SE Residential OLT L2SE IP/MPLS Ethernet Aggregation L2SE L2SE L3SE L2SW ATM Aggregation ATM/FR networks L2SW L2SW capped and to be closed BNG Mobile L2SW L2SW SONET/SDH evolving to L2SW MPLS Ethernet and OTN SDH L1SW Access L1SW and Edge optimized for MPLS Ethernet SW L1SW L0 W Optical L0SW L0SW L0SW Access Aggregation Boundary Aggregation Edge Boundary 7

8 MPLS is the Right Choice MPLS Provides a robust solution -Benefits of flexibility and statistical multiplexing of connectionless -Deterministic path of Traffic Engineering if needed -Benefits of static and dynamic protection schemes that meet transport requirements -Virtualization: L2VPN, L3VPN -Multipoint capabilities MPLS Platforms are becoming smaller and lower cost - Moving closer to end user Unified MPLS from access to core provides the following features - Unify the network to a single control plane - Unify service delivery and transport infrastructure - Reduce operational costs MPLS-TP Adds -Key OAM functions to MPLS -Offers dynamic or static provisioning option for MPLS -Advances in MPLS-TP will merge into IP/MPLS 8

9 Problem Statement

10 IP NGN High-level Architecture Mobile Access PW Creation between Agg. Carrier Ethernet Aggregation And Distr. Node Service Creation Edge on Edge Node Multiservice Core 2G/3G/4G Node Business Corporate RAN Access Network MPLS/IP Provisioning in the Access Ethernet Node PW Creation between Agg. And Distr. Node Aggregation Node Aggregation Node BSC/RNC Distribution Node Edge Node VoD TV SIP Content Network Residential STB Aggregation Network MPLS/IP Distribution Node Core Nodes Core Network IP / MPLS Business Corporate DSLNode Aggregation Node Edge Node Content Network VoD TV SIP Residential PON Node Aggregation Node BSC/RNC STB Business Corporate IPoDWDM Optical Network 10

11 IPNGN Scaling and Performance Business and Residential Services for SP scale and performance Access And Aggregation Core and Edge Access And Aggregation Platform Scale: 400Gbps backplane 180Gbps fabric 400G 40G/80G linecards Nx100G Up to 3 Million queues per system Up to 2 Million policers per system Hierarchical scheduling support Service Scale: 4M v4 / 2M v6 FIB 5k+ BGP NSR 1250 BFD sessions per LC 16K mroutes 4k+ VRF 64 K Pseudowires 2 M MAC address entries 11

12 IP NGN Scaling Number of Nodes Core Nodes few 10s IP Edge Nodes s Distribution Nodes 100s 1,000s Aggregation Nodes 1,000s 10,000s Access Nodes 10,00s 100,000s Transport CPE / NT 0000s 00000s As MPLS moves into aggregation and access number of nodes increases sharply 12

13 Requirements for Simplified Operation Reduction of number of operational points Possibility of easy provisioning of services: L3VPN, VPWS, VPLS, without PW-stitching or Inter-AS mechanisms Providing end-to-end MPLS transport Keep IGP areas separated and routing tables small Fast convergence Easy configuration and troubleshooting Ability to integrate with any access technology IPv6 Readiness 13

14 MPLS Network Architecture Evolution

15 Evolving MPLS Networks Classical Model Access Edge Core Edge Agg Access Agg L2 L2 L2VPN L3VPN IP L2VPN L3VPN MPLS Segmented networks (IGP) MPLS core L2 & L3VPN services on Core-Edge Traditional/Native Access and Aggregation Cisco ME 3.0, 3.1, etc architectures in the past 15

16 Evolving MPLS Networks Next Gen Model I Access Agg Edge Core Edge Agg Access L2 + IP L2 + IP L2VPN L3VPN IP L3VPN L2VPN MPLS Segmented Networks MPLS core IP+MPLS extended to Aggregation, simplifying Aggregation-to-Core-Edge networks L3VPN services on Core-Edge Most of today s network, like Cisco IP NGN Release

17 Evolving MPLS Networks Next Gen Model II Access Agg Edge Core Edge Agg Access IP IP L2VPN L2VPN IP L2VPN L2VPN MPLS Segmented Networks MPLS core L3VPN services on Core-Edge IP+MPLS extended to Access to provide seamless end-end L2VPN service Requires L2VPN on mid-point devices This model is possible in most of today s networks, but rarely utilized L2VPN PW 17

18 Evolving MPLS Networks Next Gen Model II Access Agg Edge Core Edge Agg Access IP IP L2VPN L2VPN IP L2VPN L2VPN MPLS L2VPN PW Supporting multiple L2VPN Pseudowire (PW) segments requires provisioning and storing L2VPN information on mid-pint devices Increases design, deployment and management complexities to achieve scalable fast converging network 18

19 Evolving MPLS Networks : Target Architecture A single PW segment over a multi-segmented transport without multi-domain complexities would be an ideal solution Access Agg Edge Core Edge Agg Access IP IP L2VPN L3VPN IP L3VPN L2VPN MPLS L2VPN PW 19

20 Unified MPLS Design Goals & Reference Model Scale - Interconnect 100k Access nodes through an MPLS domain Resilience - < 50msec convergence as often as possible Simplicity - Operation of big MPLS networks is often considered difficult DSLAM1 PE11 Distribution / Aggregation ABR11 Core and Edge ABR21 Distribution / Aggregation PE21 DSLAM2 PE12 PE22 ABR12 ABR22 IGP2 IGP1 IGP3 1k Nodes / Core 10k Nodes / Aggregation 100k Nodes / Access 20

21 Unified MPLS Architecture

22 Unified MPLS Architecture Access, Aggregation and Core are in different IGP areas No or very limited IGP route redistribution from Core towards Aggregation areas Every nodes are in the same BGP AS RFC 3107 for label distribution (prefix+label through BGP): - PE loopbacks - Central Infrastructure: Edge Nodes, etc ABRs between IGP areas also act as BGP RRs - Next-hop self for inserting ABRs into the Data Path - Loop avoidance via Cluster-id BGP Additional-path + existing mechanisms for Fast convergence 22

23 Unified MPLS High-level View IGP area 3 IGP area 1 IGP area 2 ABR-RR1 ABR-RR2 PE1 Aggregati on Domain 1 Core Aggregati on Domain 2 PE2 ABR-RR3 ABR-RR4 BGP AS 23

24 ibgp Peering between Access and Aggregation PE1 ibgp peers ABR-RR1 ibgp peers ABR-RR2 ibgp peers IGP 1 IGP 2 IGP 3 PE2 Next-Hop-Self Next-Hop-Self ABRs are also Route Reflectors PEs in the same segment peer with ABR-RRs RRs are inserted in data path by setting next-hop-self 24

25 Exchange of Route and Label information between Domains ibgp + Labels (RFC 3107) ibgp IPv4 update: PE1 Label=(L2) NH=ABR-RR1 ibgp IPv4 update: PE1 Label=(L1) NH=ABR-RR2 PE1 Aggregation ibgp IPv4 update: PE2 Label=(L4) NH=ABR-RR1 ABR-RR1 Core ibgp IPv4 update: PE2 Label=(L3) NH=ABR-RR2 ABR-RR2 Aggregation PE2 BGP updates include labels for IPv4 prefixes Only share PE loopbacks with other segments 25

26 L2VPN Circuit Establishment PW VCID:X Label:Z GE0/1 VCID:X PE1 ABR-RR1 ABR-RR2 PE2 GE0/1 VCID:X VCID:X Label:Y PE1 and PE2 exchange PW Virtual Circuit labels as usual 26

27 Traffic Forwarding & Label Stacks IGP Label BGP Label PW VC Label Payload Aggregation Core Aggregation PE1 ABR-RR1 ABR-RR2 PE2 21 L4 Z 22 L3 Z 23 Z Egress PE pops VC label IGP and BGP labels are exchanged 27

28 High Availability & Fast Convergence Today XR Soon XE/IOS RR PE1 ABR-RR1 ABR-RR2 PE2 Aggregation Core Aggregation PE4 PE3 ABR-RR3 ABR-RR4 BGP Additional-path: (cluster-id 1) (cluster-id 2) RR sends all paths for ABRs, and they perform path selection RR performs path selection, sends path + additional path 28

29 Unified MPLS Configuration examples

30 Unified MPLS Deployment Scenario Topology used for this exercise Segmented Networks with three separate IGP OSPF) Processes: Process 1, Process 2, Process 3 All network segments run MPLS ABRs with Route Reflector functionality End-to-End LSP transport between PEs PEs are RR clients of immediate ABR RRs 30

31 Unified MPLS Test Topology Segmented Networks IGP (OSPF) Each Segment is a separate OSPF process PE1 and PE5 can reach only ABR2RR and ABR6RR in segment 2 through IGP PE4 and PE8 can reach only ABR3RR and ABR7RR in segment 2 through IGP PE1#sh ip ro % Subnet not in table PE1#sh ip ro % Subnet not in table PE1#sh ip ro % Subnet not in table PE5# sh ip ro % Subnet not in table PE5# sh ip ro % Subnet not in table PE5# sh ip ro % Subnet not in table PE4#sh ip ro % Subnet not in table PE4#sh ip ro % Subnet not in table PE4#sh ip ro % Subnet not in table PE8#sh ip route % Subnet not in table PE8#sh ip route % Subnet not in table PE8#sh ip route % Subnet not in table 31

32 Unified MPLS Test Topology Segmented Networks Limited Loopback propagation via IGP to allow ibgp peering ABR2RR#! router ospf 1! Redistribute only ABR2RR & ABR6RR loopbacks into OSPF Process 1 redistribute ospf 2 subnets match internal external 1 external 2 route-map OSPF2INTo1 access-list 1 permit ! route-map OSPF2INTo1 permit 10! match ip address 1 ABR3RR#! router ospf 3! redistribute ospf 2 subnets match internal external 1 external 2 route-map OSPF2Into3 access-list 1 permit ! route-map OSPF2Into3 permit 10! match ip address 1 32

33 Unified MPLS Test Topology Segmented Networks Segment 1 IGP routes PE5#sh ip route /16 is variably subnetted, 8 subnets, 2 masks O /32 [110/21] via , 19:32:07, Ethernet0/0 C /32 is directly connected, Loopback0 O /32 [110/11] via , 19:32:07, Ethernet0/0 O /24 [110/20] via , 19:31:52, Ethernet0/0 O /24 [110/20] via , 19:32:07, Ethernet0/0 O /24 [110/20] via , 19:32:07, Ethernet0/0 C /24 is directly connected, Ethernet0/0 L /32 is directly connected, Ethernet0/ /32 is subnetted, 2 subnets O E [110/1] via , 19:31:52, Ethernet0/0 O E [110/1] via , 19:32:07, Ethernet0/0 33

34 Unified MPLS Test Topology Segmented Networks IGP MPLS PE5#sh mpls forwarding-table Local Outgoing Prefix Bytes Label Outgoing Next Hop Label Label or Tunnel Id Switched interface 17 Pop Label /32 0 Et0/ /32 0 Et0/ Pop Label /24 0 Et0/ Pop Label /24 0 Et0/ Pop Label /24 0 Et0/ /32 0 Et0/ /32 0 Et0/ PE1#sh mpls forwarding-table Local Outgoing Prefix Bytes Label Outgoing Next Hop Label Label or Tunnel Id Switched interface 18 Pop Label /32 0 Et0/ Pop Label /24 0 Et0/ Pop Label /24 0 Et0/ Pop Label /24 0 Et0/ /32 0 Et0/ /32 0 Et0/ /32 0 Et0/ P9#sh mpls forwarding-table Local Outgoing Prefix Bytes Label Outgoing Next Hop Label Label or Tunnel Id Switched interface 21 Pop Label / Et3/ Pop Label / Et0/ Pop Label / Et1/ Pop Label / Et2/

35 Unified MPLS Test Topology Segmented Networks IGP MPLS ABR2RR#sh mpls forwarding-table Local Outgoing Prefix Bytes Label Outgoing Next Hop Label Label or Tunnel Id Switched interface 19 Pop Label /32 0 Et0/ /32 0 Et0/ /32 0 Et0/ Pop Label /24 0 Et0/ Pop Label /24 0 Et0/ Pop Label /24 0 Et0/ /32 0 Et1/ Pop Label /32 0 Et1/ Pop Label /24 0 Et1/ Pop Label /24 0 Et1/ Pop Label /24 0 Et1/ /32 0 Et1/ /32 0 Et1/ / Et1/ /32 0 Et1/ /32 0 Et1/ / Et1/ ABR3RR#sh mpls forwarding-table Local Outgoing Prefix Bytes Label Outgoing Next Hop Label Label or Tunnel Id Switched interface 18 Pop Label /32 0 Et0/ Pop Label /24 0 Et0/ Pop Label /24 0 Et0/ Pop Label /24 0 Et0/ Pop Label /32 0 Et1/ /32 0 Et1/ Pop Label /24 0 Et1/ Pop Label /24 0 Et1/ Pop Label /24 0 Et1/ /32 0 Et1/ /32 0 Et1/ /32 0 Et0/ / Et0/ /32 0 Et1/ / Et1/ /32 0 Et1/ /32 0 Et1/

36 Unified MPLS Test Topology Segmented Networks ibgp peering All segments are in the same BGP domain PEs are clients of ABR RRs in the same segment -PE1 and PE5 are ibgp peers with ABR2RR and ABR6RR -PE4 and PE8 are ibgp peers with ABR3RR and ABR7RR 36

37 Unified MPLS Test Topology Segmented Networks seup ibgp Sessions ABR2RR# & ABR6RR! router bgp 100!! neighbor remote-as 100 neighbor update-source Loopback0 neighbor remote-as 100 neighbor update-source Loopback0 address-family ipv4 neighbor activate neighbor route-reflector-client neighbor activate neighbor route-reflector-client ABR3RR# & ABR7RR 37! router bgp 100!! neighbor remote-as 100 neighbor update-source Loopback0 neighbor remote-as 100 neighbor update-source Loopback0 address-family ipv4 neighbor activate neighbor route-reflector-client neighbor activate neighbor route-reflector-client

38 Unified MPLS Test Topology Segmented Networks ibgp Sessions up ABR2RR#sh ip bgp neighbors BGP neighbor is , remote AS 100, internal link BGP version 4, remote router ID BGP state = Established, up for 22:00:41.snip. BGP neighbor is , remote AS 100, internal link BGP version 4, remote router ID BGP state = Established, up for 22:05:34.snip. ABR2RR#sh ip bgp neighbors BGP neighbor is , remote AS 100, internal link BGP version 4, remote router ID BGP state = Established, up for 22:06:46.snip. BGP neighbor is , remote AS 100, internal link BGP version 4, remote router ID BGP state = Established, up for 22:14:39 BGP neighbor is , remote AS 100, internal link BGP version 4, remote router ID BGP state = Established, up for 22:12:12 38

39 Unified MPLS Test Topology Segmented Networks Establishing LSPs End-to-End ibgp Next-Hop-Self is enabled on ABR RRs to insert ABR RRs in the data path ABR2RR & ABR6RR pair is in the different cluster id than ABR3RR & ABR7RR pair for loop avoidance Use ibgp + Label to announce PEs loopbacks and labels through ibgp to build end-to-end LSPs Redistribute IGP into BGP, to get PE loopbacks into BGP table Use filters (IPv4+Labels) to allow only desired PE loopbacks + labels to be propagated Each ABR RRs allocate labels for PEs loopbacks and propagate PE NLRIs through ibgp only ibgp multipath allows to store multipaths in the FIB table for load balancing - installing paths to both RRs on PEs PEs in segment 1 can reach PEs in segment 3 allowing to build PWs directly between Access PEs located in different segments 39

40 Unified MPLS Test Topology Segmented Networks ibgp Multipath Traffic should be able to utilize any path in the network ibgp multipath allows to store multiple paths in the FIB table for load balancing - installing paths to both RRs on PEs. PE1 should be able to forward traffic to ABR2RR and ABR6RR. Similarly, ABR2RR should be able to forward traffic to both ABR3RR and ABR7RR Enable ibgp multipath functionality on all devices! address-family ipv4 maximum-paths ibgp 2! exit-address-family 40

41 Unified MPLS Test Topology Segmented Networks ibgp Multipath PE1# & PE5# sh ip bgp nexthops # Paths Nexthop Address PE4 & PE8# sh ip bgp nexthops # Paths Nexthop Address ABR2RR#sh ip bgp nexthops # Paths Nexthop Address ABR6RR#sh ip bgp nexthops # Paths Nexthop Address ABR3RR#sh ip bgp nexthops # Paths Nexthop Address ABR7RR#sh ip bgp nexthops # Paths Nexthop Address

42 Unified MPLS Test Topology Segmented Networks ibgp Multiple paths reflected For example: to reach PE4, two paths are reflected on PE1 PE1#sh bgp BGP routing table entry for /32, version 17 Paths: (2 available, best #2, table default) Multipath: ibgp Not advertised to any peer Refresh Epoch 1 Local (metric 1) from ( ) Origin incomplete, metric 21, localpref 100, valid, internal, multipath(oldest) Originator: , Cluster list: mpls labels in/out nolabel/31 Refresh Epoch 1 Local (metric 1) from ( ) Origin incomplete, metric 21, localpref 100, valid, internal, multipath, best Originator: , Cluster list: mpls labels Presentation_ID in/out nolabel/ Cisco and/or its affiliates. All rights reserved. Cisco Public 42

43 Unified MPLS Test Topology Segmented Networks ibgp Next Hop Self ibgp Next-Hop-Self is enabled on ABR RRs towards each BGP peer to insert ABR RRs in the data path Sample configuration on ABR3RR! address-family ipv4 neighbor next-hop-self all neighbor next-hop-self all neighbor next-hop-self all neighbor next-hop-self all neighbor next-hop-self all exit-address-family! 43

44 Unified MPLS Test Topology Segmented Networks ibgp Next Hop Self Redistribute IGP into BGP, to get PE loopbacks into BGP table on ABR RRs (only) Use filters (IPv4+Labels) to allow only the desired PE loopbacks + labels to be propagated to other segments (ABR RR to ABR RR)!Sample configuration on ABR3RR access-list 2 permit log access-list 2 permit log! route-map OUT permit 11 match ip address 2 set mpls-label! Sample configuration on ABR3RR address-family ipv4 redistribute ospf 3 neighbor route-map OUT out neighbor route-map OUT out! *Filter is not needed towards PEs 44

45 Unified MPLS Test Topology Segmented Networks ibgp + Label Use ibgp + Label to announce PEs loopbacks and labels through ibgp to build end-to-end LSPs Enable RFC 3107 on all PEs and ABR RRs as send-label capability needs to be negotiated between peers! Sample configuration on ABR3RR address-family ipv4 neighbor send-label neighbor send-label neighbor send-label neighbor send-label neighbor send-label exit-address-family! 45

46 Unified MPLS Test Topology Segmented Networks BGP Lables ABR2RR#sh bgp ipv4 unicast labels Network Next Hop In label/out label / / /nolabel / / /nolabel / / / / / /29 ABR3RR#sh bgp ipv4 unicast labels Network Next Hop In label/out label / / / / / / / / /nolabel / / /nolabel 46

47 Unified MPLS Test Topology Segmented Networks BGP Lables PE1#sh bgp ipv4 unicast labels Network Next Hop In label/out label / nolabel/ nolabel/ / nolabel/ nolabel/ / nolabel/ nolabel/ / nolabel/ nolabel/33 PE4# sh bgp ipv4 unicast labels Network Next Hop In label/out label / nolabel/ nolabel/ / nolabel/ nolabel/ / nolabel/ nolabel/ / nolabel/ nolabel/29 47

48 Unified MPLS Test Topology Segmented Networks BGP Routes PE5#sh ip route ABR2RR#sh ip route /32 is subnetted, 2 subnets /32 is subnetted, 2 subnets B [200/21] via , 1d00h B [200/21] via , 1d00h [200/21] via , 1d00h [200/21] via , 1d00h B [200/21] via , 1d00h [200/21] via , 1d00h B [200/21] via , 1d00h [200/21] via , 1d00h PE1#sh ip route /32 is subnetted, 2 subnets ABR6RR#sh ip route /32 is subnetted, 2 subnets B [200/21] via , 1d00h [200/21] via , 1d00h B [200/21] via , 1d00h [200/21] via , 1d00h B [200/21] via , 1d00h B [200/21] via , 1d00h [200/21] via , 1d00h [200/21] via , 1d00h 48

49 Unified MPLS Test Topology Segmented Networks Enabling L2VPN service PEs in segment 1 can reach PEs in segment 3 allowing to build PWs directly between Access PEs located in different segments!pe1 interface Ethernet0/1 no ip address xconnect encapsulation mpls!!pe4! interface Ethernet0/1 no ip address xconnect encapsulation mpls 49

50 Unified MPLS Test Topology Segmented Networks Enabling L2VPN service PEs in segment 1 can reach PEs in segment 3 allowing to build PWs directly between Access PEs located in different segments!pe5 interface Ethernet0/1 no ip address xconnect encapsulation mpls!!pe8! interface Ethernet0/1 no ip address xconnect encapsulation mpls 50

51 Unified MPLS Test Topology L2VPN Pseudowire Circuit Verification PE1#sh mpls l2transport vc 14 detail Local interface: Et0/1 up, line protocol up, Ethernet up Destination address: , VC ID: 14, VC status: up Output interface: Et0/0, imposed label stack { } Preferred path: not configured Default path: active Next hop: Create time: 00:20:47, last status change time: 00:18:53 Signaling protocol: LDP, peer :0 up Targeted Hello: (LDP Id) -> , LDP is UP MPLS VC labels: local 16, remote 16 PWID: snip : PW VC Label 32: BGP (PE) Label 23: IGP label 51

52 Unified MPLS Test Topology L2VPN Pseudowire Circuit Verification PE4#sh mpls l2transport vc 14 detail Local interface: Et0/1 up, line protocol up, Ethernet up Destination address: , VC ID: 14, VC status: up Output interface: Et0/0, imposed label stack { } Preferred path: not configured Default path: active Next hop: Create time: 00:24:29, last status change time: 00:23:30 Signaling protocol: LDP, peer :0 up Targeted Hello: (LDP Id) -> , LDP is UP MPLS VC labels: local 16, remote 16 PWID: snip : PW VC Label 32: BGP (PE) Label 22: IGP label 52

53 Unified MPLS Test Topology L2VPN Pseudowire forwarding Verification PE1# ping mpls pseudowire source Sending 5, 100-byte MPLS Echos to , timeout is 2 seconds, send interval is 0 msec: snip Type escape sequence to abort.!!!!! Success rate is 100 percent (5/5), round-trip min/avg/max = 4/4/5 ms 53

54 Further Developments for Scale and Convergence

55 Access Domain Integration via LDP DoD Service Provisioning Service Provisioning Port P xconnect PE D1 PE12 IP/MPLS control plane Simple access devices only need to ask aggregation routers for the label to use when configured to deliver service to a remote end point. Reduces provisioning to only end points of service 55

56 Loop Free Alternate (LFA) Link Protection The link between R P and R B fails. Today XR & IOS R P reroutes all traffic originally for link R P -R B to R C. This is done by precomputing available paths that do not create loops Gives benefits of TE-FRR, but no configuration or design required Route X: NH: R B, LFA: R C R P R B packet addr X packet addr X packet addr X Network X R C Primary Path Repair Path 56

57 BGP Prefix-Independent Protection (PIC) 50msec protection Prefix-Independent PE3 Cust1 Default behavior, entirely automated computation PE1 No operator involvement Algorithm uses a pointer to move all prefixes to new next hop, not a hop by hop calculation as in the past. Enables 3107 BGP+labels operation to scale via hierarchy while maintaining fast convergence characteristics PE2 Cust2 57

58 Scalability through Divide & Conquer Disconnect & Isolate IGP domains - No more end-to-end IGP view Leverage BGP for infrastructure (i.e. PE) routes - Also for infrastructure (i.e. PE) labels Access BGP for Services BGP for Infrastructure Isolated IGP & LDP Isolated IGP & LDP Isolated IGP & LDP Aggregation Region1. Backbone Region 2 Aggregation Access ISIS Level 1 Or OSPF Area Y ISIS Level 2 Or OSPF Area 0.. ISIS Level 1 Or OSPF Area X R. 58

59 References draft-ietf-mpls-seamless-mpls-00 draft-ietf-idr-aigp-06 draft-ietf-idr-add-paths-06 draft-shand-remote-lfa-00 draft-ietf-rtgwg-lfa-applicability-03 Cisco Unified MPLS Mobile Transport 1.0 (will be on CCO soon) 59

60 Recommended Reading BRKSPG- 3305

61 61

62 Thank you. 62