idnits 2.17.1 draft-ietf-rtgwg-bgp-pic-09.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- -- The document has examples using IPv4 documentation addresses according to RFC6890, but does not use any IPv6 documentation addresses. Maybe there should be IPv6 examples, too? Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == The document seems to contain a disclaimer for pre-RFC5378 work, but was first submitted on or after 10 November 2008. The disclaimer is usually necessary only for documents that revise or obsolete older RFCs, and that take significant amounts of text from those RFCs. If you can contact all authors of the source material and they are willing to grant the BCP78 rights to the IETF Trust, you can and should remove the disclaimer. Otherwise, the disclaimer is needed and you can ignore this comment. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (April 1, 2019) is 1845 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Outdated reference: A later version (-22) exists of draft-ietf-spring-segment-routing-mpls-12 == Outdated reference: A later version (-05) exists of draft-bashandy-rtgwg-segment-routing-ti-lfa-02 Summary: 0 errors (**), 0 flaws (~~), 4 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 1 Network Working Group A. Bashandy, Ed. 2 Internet Draft Arrcus 3 Intended status: Informational C. Filsfils 4 Expires: October 2019 Cisco Systems 5 P. Mohapatra 6 Sproute Networks 7 April 1, 2019 9 BGP Prefix Independent Convergence 10 draft-ietf-rtgwg-bgp-pic-09.txt 12 Abstract 14 In the network comprising thousands of iBGP peers exchanging millions 15 of routes, many routes are reachable via more than one next-hop. 16 Given the large scaling targets, it is desirable to restore traffic 17 after failure in a time period that does not depend on the number of 18 BGP prefixes. In this document we proposed an architecture by which 19 traffic can be re-routed to ECMP or pre-calculated backup paths in a 20 timeframe that does not depend on the number of BGP prefixes. The 21 objective is achieved through organizing the forwarding data 22 structures in a hierarchical manner and sharing forwarding elements 23 among the maximum possible number of routes. The proposed technique 24 achieves prefix independent convergence while ensuring incremental 25 deployment, complete automation, and zero management and provisioning 26 effort. It is noteworthy to mention that the benefits of BGP-PIC are 27 hinged on the existence of more than one path whether as ECMP or 28 primary-backup. 30 Status of this Memo 32 This Internet-Draft is submitted in full conformance with the 33 provisions of BCP 78 and BCP 79. 35 This document may contain material from IETF Documents or IETF 36 Contributions published or made publicly available before November 37 10, 2008. The person(s) controlling the copyright in some of this 38 material may not have granted the IETF Trust the right to allow 39 modifications of such material outside the IETF Standards Process. 40 Without obtaining an adequate license from the person(s) 41 controlling the copyright in such materials, this document may not 42 be modified outside the IETF Standards Process, and derivative 43 works of it may not be created outside the IETF Standards Process, 44 except to format it for publication as an RFC or to translate it 45 into languages other than English. 47 Internet-Drafts are working documents of the Internet Engineering 48 Task Force (IETF), its areas, and its working groups. Note that 49 other groups may also distribute working documents as Internet- 50 Drafts. 52 Internet-Drafts are draft documents valid for a maximum of six 53 months and may be updated, replaced, or obsoleted by other 54 documents at any time. It is inappropriate to use Internet-Drafts 55 as reference material or to cite them other than as "work in 56 progress." 58 The list of current Internet-Drafts can be accessed at 59 http://www.ietf.org/ietf/1id-abstracts.txt 61 The list of Internet-Draft Shadow Directories can be accessed at 62 http://www.ietf.org/shadow.html 64 This Internet-Draft will expire on October 1, 2019. 66 Copyright Notice 68 Copyright (c) 2019 IETF Trust and the persons identified as the 69 document authors. All rights reserved. 71 This document is subject to BCP 78 and the IETF Trust's Legal 72 Provisions Relating to IETF Documents 73 (http://trustee.ietf.org/license-info) in effect on the date of 74 publication of this document. Please review these documents 75 carefully, as they describe your rights and restrictions with 76 respect to this document. Code Components extracted from this 77 document must include Simplified BSD License text as described in 78 Section 4.e of the Trust Legal Provisions and are provided without 79 warranty as described in the Simplified BSD License. 81 Table of Contents 83 1. Introduction...................................................3 84 1.1. Terminology...............................................4 85 2. Overview.......................................................5 86 2.1. Dependency................................................6 87 2.1.1. Hierarchical Hardware FIB............................6 88 2.1.2. Availability of more than one primary or secondary BGP 89 next-hops...................................................7 90 2.2. BGP-PIC Illustration......................................7 91 3. Constructing the Shared Hierarchical Forwarding Chain..........9 92 3.1. Constructing the BGP-PIC forwarding Chain.................9 93 3.2. Example: Primary-Backup Path Scenario....................10 94 4. Forwarding Behavior...........................................11 95 5. Handling Platforms with Limited Levels of Hierarchy...........12 96 5.1. Flattening the Forwarding Chain..........................12 97 5.2. Example: Flattening a forwarding chain...................14 98 6. Forwarding Chain Adjustment at a Failure......................21 99 6.1. BGP-PIC core.............................................22 100 6.2. BGP-PIC edge.............................................23 101 6.2.1. Adjusting forwarding Chain in egress node failure...23 102 6.2.2. Adjusting Forwarding Chain on PE-CE link Failure....23 103 6.3. Handling Failures for Flattened Forwarding Chains........24 104 7. Properties....................................................25 105 7.1. Coverage.................................................25 106 7.1.1. A remote failure on the path to a BGP next-hop......25 107 7.1.2. A local failure on the path to a BGP next-hop.......25 108 7.1.3. A remote iBGP next-hop fails........................26 109 7.1.4. A local eBGP next-hop fails.........................26 110 7.2. Performance..............................................26 111 7.3. Automated................................................27 112 7.4. Incremental Deployment...................................27 113 8. Security Considerations.......................................27 114 9. IANA Considerations...........................................27 115 10. Conclusions..................................................27 116 11. References...................................................28 117 11.1. Normative References....................................28 118 11.2. Informative References..................................28 119 12. Acknowledgments..............................................29 120 Appendix A. Perspective..........................................30 122 1. Introduction 124 As a path vector protocol, BGP propagates reachability serially. 125 Hence BGP convergence speed is limited by the time taken to 126 serially propagate reachability information from the point of 127 failure to the device that must re-converge. BGP speakers exchange 128 reachability information about prefixes[1][2] and, for labeled 129 address families, namely AFI/SAFI 1/4, 2/4, 1/128, and 2/128, an 130 edge router assigns local labels to prefixes and associates the 131 local label with each advertised prefix such as L3VPN [7], 6PE 132 [8], and Softwire [6] using BGP label unicast technique[3]. A BGP 133 speaker then applies the path selection steps to choose the best 134 path. In modern networks, it is not uncommon to have a prefix 135 reachable via multiple edge routers. In addition to proprietary 136 techniques, multiple techniques have been proposed to allow for 137 BGP to advertise more than one path for a given prefix 138 [5][10][11], whether in the form of equal cost multipath or 139 primary-backup. Another common and widely deployed scenario is 140 L3VPN with multi-homed VPN sites with unique Route Distinguisher. 141 It is advantageous to utilize the commonality among paths used by 142 NLRIs to significantly improve convergence in case of topology 143 modifications. 145 This document proposes a hierarchical and shared forwarding chain 146 organization that allows traffic to be restored to pre-calculated 147 alternative equal cost primary path or backup path in a time 148 period that does not depend on the number of BGP prefixes. The 149 technique relies on internal router behavior that is completely 150 transparent to the operator and can be incrementally deployed and 151 enabled with zero operator intervention. 153 1.1. Terminology 155 This section defines the terms used in this document. For ease of 156 use, we will use terms similar to those used by L3VPN [7] 158 o BGP prefix: A prefix P/m (of any AFI/SAFI) that a BGP speaker 159 has a path for. 161 o IGP prefix: A prefix P/m (of any AFI/SAFI) that is learnt via 162 an Interior Gateway Protocol, such as OSPF and ISIS, has a path 163 for. The prefix may be learnt directly through the IGP or 164 redistributed from other protocol(s) 166 o CE: An external router through which an egress PE can reach a 167 prefix P/m. 169 o Ingress PE, "iPE": A BGP speaker that learns about a prefix 170 through a IBGP peer and chooses an egress PE as the next-hop for 171 the prefix. 173 o Path: The next-hop in a sequence of nodes starting from the 174 current node and ending with the destination node or network 175 identified by the prefix. The nodes may not be directly 176 connected. 178 o Recursive path: A path consisting only of the IP address of the 179 next-hop without the outgoing interface. Subsequent lookups are 180 necessary to determine the outgoing interface and a directly 181 connected next-hop 183 o Non-recursive path: A path consisting of the IP address of a 184 directly connected next-hop and outgoing interface 186 o Primary path: A recursive or non-recursive path that can be 187 used all the time as long as a walk starting from this path can 188 end to an adjacency. A prefix can have more than one primary 189 path 191 o Backup path: A recursive or non-recursive path that can be used 192 only after some or all primary paths become unreachable 194 o Leaf: A container data structure for a prefix or local label. 195 Alternatively, it is the data structure that contains prefix 196 specific information. 198 o IP leaf: The leaf corresponding to an IPv4 or IPv6 prefix 200 o Label leaf. The leaf corresponding to a locally allocated label 201 such as the VPN label on an egress PE [7]. 203 o Pathlist: An array of paths used by one or more prefix to forward 204 traffic to destination(s) covered by a IP prefix. Each path in 205 the pathlist carries its "path-index" that identifies its 206 position in the array of paths. "). In general, the value of the 207 "path-index" stored in path may not necessarily has the same 208 value of the location of the path in the pathlist. For example 209 the 3rd path may carry path-index value of 1 211 o A pathlist may contain a mix of primary and backup paths 213 o OutLabel-List: Each labeled prefix is associated with an 214 OutLabel-List. The OutLabel-List is an array of one or more 215 outgoing labels and/or label actions where each label or label 216 action has 1-to-1 correspondence to a path in the pathlist. 217 Label actions are: push the label, pop the label, swap the 218 incoming label with the label in the Outlabel-Array entry, or 219 don't push anything at all in case of "unlabeled". The prefix 220 may be an IGP or BGP prefix 222 o Adjacency: The layer 2 encapsulation leading to the layer 3 223 directly connected next-hop 225 o Dependency: An object X is said to be a dependent or child of 226 object Y if there is at least one forwarding chain where the 227 forwarding engine must visits the object X before visiting the 228 object Y in order to forward a packet. Note that if object X is 229 a child of object Y, then Y cannot be deleted unless object X 230 is no longer a dependent/child of object Y 232 o Route: A prefix with one or more paths associated with it. 233 Hence the minimum set of objects needed to construct a route is 234 a leaf and a pathlist. 236 2. Overview 238 The idea of BGP-PIC is based on two pillars 239 o A shared hierarchical forwarding Chain: It is not uncommon to see 240 multiple destinations are reachable via the same list of next- 241 hops. Instead of having a separate list of next-hops for each 242 destination, all destinations sharing the same list of next-hops 243 can point to a single copy of this list thereby allowing fast 244 convergence by making changes to a single shared list of next- 245 hops rather than possibly a large number of destinations. Because 246 paths in a pathlist may be recursive, a hierarchy is formed 247 between pathlist and the resolving prefix whereby the pathlist 248 depends on the resolving prefix. 250 o A forwarding plane that supports multiple levels of indirection: 251 A forwarding that starts with a destination and ends with an 252 outgoing interface is not a simple flat structure. Instead a 253 forwarding entry is constructed via multiple levels of 254 dependency. A BGP NLRI uses a recursive next-hop, which in turn 255 resolves via an IGP next-hop, which in turn resolves via an 256 adjacency consisting of one or more outgoing interface(s) and 257 next-hop(s). 259 Designing a forwarding plane that constructs multi-level forwarding 260 chains with maximal sharing of forwarding objects allows rerouting a 261 large number of destinations by modifying a small number of objects 262 thereby achieving convergence in a time frame that does not depend 263 on the number of destinations. For example, if the IGP prefix that 264 resolves a recursive next-hop is updated there is no need to update 265 the possibly large number of BGP NLRIs that use this recursive next- 266 hop. 268 2.1. Dependency 270 This section describes the required functionality in the forwarding 271 and control planes to support BGP-PIC described in this document 273 2.1.1. Hierarchical Hardware FIB 275 BGP PIC requires a hierarchical hardware FIB support: for each BGP 276 forwarded packet, a BGP leaf is looked up, then a BGP Pathlist is 277 consulted, then an IGP Pathlist, then an Adjacency. 279 An alternative method consists in "flattening" the dependencies when 280 programming the BGP destinations into HW FIB resulting in 281 potentially eliminating both the BGP Path-List and IGP Path-List 282 consultation. Such an approach decreases the number of memory 283 lookup's per forwarding operation at the expense of HW FIB memory 284 increase (flattening means less sharing hence duplication), loss of 285 ECMP properties (flattening means less pathlist entropy) and loss of 286 BGP PIC properties. 288 2.1.2. Availability of more than one primary or secondary BGP next-hops 290 When the primary BGP next-hop fails, BGP PIC depends on the 291 availability of a pre-computed and pre-installed secondary BGP next- 292 hop in the BGP Pathlist. 294 The existence of a secondary next-hop is clear for the following 295 reason: a service caring for network availability will require two 296 disjoint network connections hence two BGP next-hops. 298 The BGP distribution of the secondary next-hop is available thanks 299 to the following BGP mechanisms: Add-Path [10], BGP Best-External 300 [5], diverse path [11], and the frequent use in VPN deployments of 301 different VPN RD's per PE. It is noteworthy to mention that the 302 availability of another BGP path does not mean that all failure 303 scenarios can be covered by simply forwarding traffic to the 304 available secondary path. The discussion of how to cover various 305 failure scenarios is beyond the scope of this document 307 2.2. BGP-PIC Illustration 309 To illustrate the two pillars above as well as the platform 310 dependency, we will use an example of a simple multihomed L3VPN [7] 311 prefix in a BGP-free core running LDP [4] or segment routing over 312 MPLS forwarding plane [13]. 314 +--------------------------------+ 315 | | 316 | ePE2 (IGP-IP1 192.0.2.1, Loopback) 317 | | \ 318 | | \ 319 | | \ 320 iPE | CE....VRF "Blue", ASnum 65000 321 | | / (VPN-IP1 198.51.100.0/24) 322 | | / (VPN-IP2 203.0.113.0/24) 323 | LDP/Segment-Routing Core | / 324 | ePE1 (IGP-IP2 192.0.2.2, Loopback) 325 | | 326 +--------------------------------+ 327 Figure 1 VPN prefix reachable via multiple PEs 329 Referring to Figure 1, suppose the iPE (the ingress PE) receives 330 NLRIs for the VPN prefixes VPN-IP1 and VPN-IP2 from two egress PEs, 331 ePE1 and ePE2 with next-hop BGP-NH1 and BGP-NH2, respectively. 332 Assume that ePE1 advertise the VPN labels VPN-L11 and VPN-L12 while 333 ePE2 advertise the VPN labels VPN-L21 and VPN-L22 for VPN-IP1 and 334 VPN-IP2, respectively. Suppose that BGP-NH1 and BGP-NH2 are resolved 335 via the IGP prefixes IGP-IP1 and IGP-P2, where each happen to have 2 336 ECMP paths with IGP-NH1 and IGP-NH2 reachable via the interfaces I1 337 and I2, respectively. Suppose that local labels (whether LDP [4] or 338 segment routing [13]) on the downstream LSRs for IGP-IP1 are IGP-L11 339 and IGP-L12 while for IGP-P2 are IGP-L21 and IGP-L22. As such, the 340 routing table at iPE is as follows: 342 65000: 198.51.100.0/24 343 via ePE1 (192.0.2.1), VPN Label: VPN-L11 344 via ePE2 (192.0.2.2), VPN Label: VPN-L21 346 65000: 203.0.113.0/24 347 via ePE1 (192.0.2.1), VPN Label: VPN-L12 348 via ePE2 (192.0.2.2), VPN Label: VPN-L22 350 192.0.2.1/32 351 via Core, Label: IGP-L11 352 via Core, Label: IGP-L12 354 192.0.2.2/32 355 via Core, Label: IGP-L21 356 via Core, Label: IGP-L22 358 Based on the above routing table, a hierarchical forwarding chain 359 can be constructed as shown in Figure 2. 361 IP Leaf: Pathlist: IP Leaf: Pathlist: 362 -------- +-------+ -------- +----------+ 363 VPN-IP1-->|BGP-NH1|-->IGP-IP1(BGP NH1)--->|IGP NH1,I1|--->Adjacency1 364 | |BGP-NH2|-->.... | |IGP NH2,I2|--->Adjacency2 365 | +-------+ | +----------+ 366 | | 367 | | 368 v v 369 OutLabel-List: OutLabel-List: 370 +----------------------+ +----------------------+ 371 |VPN-L11 (VPN-IP1, NH1)| |IGP-L11 (IGP-IP1, NH1)| 372 |VPN-L21 (VPN-IP1, NH2)| |IGP-L12 (IGP-IP1, NH2)| 373 +----------------------+ +----------------------+ 375 Figure 2 Shared Hierarchical Forwarding Chain at iPE 377 The forwarding chain depicted in Figure 2 illustrates the first 378 pillar, which is sharing and hierarchy. We can see that the BGP 379 pathlist consisting of BGP-NH1 and BGP-NH2 is shared by all NLRIs 380 reachable via ePE1 and ePE2. As such, it is possible to make changes 381 to the pathlist without having to make changes to the NLRIs. For 382 example, if BGP-NH2 becomes unreachable, there is no need to modify 383 any of the possibly large number of NLRIs. Instead only the shared 384 pathlist needs to be modified. Likewise, due to the hierarchical 385 structure of the forwarding chain, it is possible to make 386 modifications to the IGP routes without having to make any changes 387 to the BGP NLRIs. For example, if the interface "I2" goes down, only 388 the shared IGP pathlist needs to be updated, but none of the IGP 389 prefixes sharing the IGP pathlist nor the BGP NLRIs using the IGP 390 prefixes for resolution need to be modified. 392 Figure 2 can also be used to illustrate the second BGP-PIC pillar. 393 Having a deep forwarding chain such as the one illustrated in Figure 394 2 requires a forwarding plane that is capable of accessing multiple 395 levels of indirection in order to calculate the outgoing 396 interface(s) and next-hops(s). While a deeper forwarding chain 397 minimizes the re-convergence time on topology change, there will 398 always exist platforms with limited capabilities and hence imposing 399 a limit on the depth of the forwarding chain. Section 5 describes 400 how to gracefully trade off convergence speed with the number of 401 hierarchical levels to support platforms with different 402 capabilities. 404 3. Constructing the Shared Hierarchical Forwarding Chain 406 Constructing the forwarding chain is an application of the two 407 pillars described in Section 2. This section describes how to 408 construct the forwarding chain in hierarchical shared manner 410 3.1. Constructing the BGP-PIC forwarding Chain 412 The whole process starts when BGP downloads a prefix to FIB. The 413 prefix contains one or more outgoing paths. For certain labeled 414 prefixes, such as VPN [7] prefixes, each path may be associated with 415 an outgoing label and the prefix itself may be assigned a local 416 label. The list of outgoing paths defines a pathlist. If such 417 pathlist does not already exist, then FIB creates a new pathlist, 418 otherwise the existing pathlist is used. The BGP prefix is added as 419 a dependent of the pathlist. 421 The previous step constructs the upper part of the hierarchical 422 forwarding chain. The forwarding chain is completed by resolving the 423 paths of the pathlist. A BGP path usually consists of a next-hop. 424 The next-hop is resolved by finding a matching IGP prefix. 426 The end result is a hierarchical shared forwarding chain where the 427 BGP pathlist is shared by all BGP prefixes that use the same list of 428 paths and the IGP prefix is shared by all pathlists that have a path 429 resolving via that IGP prefix. It is noteworthy to mention that the 430 forwarding chain is constructed without any operator intervention at 431 all. 433 The remainder of this section goes over an example to illustrate the 434 applicability of BGP-PIC in a primary-backup path scenario. 436 3.2. Example: Primary-Backup Path Scenario 438 Consider the egress PE ePE1 in the case of the multi-homed VPN 439 prefixes in the BGP-free core depicted in Figure 1. Suppose ePE1 440 determines that the primary path is the external path but the backup 441 path is the iBGP path to the other PE ePE2 with next-hop BGP-NH2. 442 ePE2 constructs the forwarding chain depicted in Figure 3. We are 443 only showing a single VPN prefix for simplicity. But all prefixes 444 that are multihomed to ePE1 and ePE2 share the BGP pathlist. 446 BGP OutLabel Array 447 VPN-L11 +---------+ 448 (Label-leaf)---+---->|Unlabeled| 449 | +---------+ 450 | | VPN-L21 | 451 | | (swap) | 452 | +---------+ 453 | 454 | BGP Pathlist 455 | +------------+ Connected route 456 | | CE-NH |------>(to the CE) 457 | |path-index=0| 458 | +------------+ 459 | | VPN-NH2 | 460 VPN-IP1 -----+------------------>| (backup) |------>IGP Leaf 461 (IP prefix leaf) |path-index=1| (Towards ePE2) 462 | +------------+ 463 | 464 | BGP OutLabel Array 465 | +---------+ 466 +------------->|Unlabeled| 467 +---------+ 468 | VPN-L21 | 469 | (push) | 470 +---------+ 472 Figure 3 : VPN Prefix Forwarding Chain with eiBGP paths on egress PE 474 The example depicted in Figure 3 differs from the example in Figure 475 2 in two main aspects. First, as long as the primary path towards 476 the CE (external path) is useable, it will be the only path used for 477 forwarding while the OutLabel-List contains both the unlabeled label 478 (primary path) and the VPN label (backup path) advertised by the 479 backup path ePE2. The second aspect is presence of the label leaf 480 corresponding to the VPN prefix. This label leaf is used to match 481 VPN traffic arriving from the core. Note that the label leaf shares 482 the pathlist with the IP prefix. 484 4. Forwarding Behavior 486 This section explains how the forwarding plane uses the hierarchical 487 shared forwarding chain to forward a packet. 489 When a packet arrives at a router, it matches a leaf. A labeled 490 packet matches a label leaf while an IP packet matches an IP prefix 491 leaf. The forwarding engines walks the forwarding chain starting 492 from the leaf until the walk terminates on an adjacency. Thus when a 493 packet arrives, the chain is walked as follows: 495 1. Lookup the leaf based on the destination address or the label at 496 the top of the packet 498 2. Retrieve the parent pathlist of the leaf 500 3. Pick the outgoing path "Pi" from the list of resolved paths in 501 the pathlist. The method by which the outgoing path is picked is 502 beyond the scope of this document (e.g. flow-preserving hash 503 exploiting entropy within the MPLS stack and IP header). Let the 504 "path-index" of the outgoing path "Pi" be "j". 506 4. If the prefix is labeled, use the "path-index" "j" to retrieve 507 the jth label "Lj" stored the jth entry in the OutLabel-List and 508 apply the label action of the label on the packet (e.g. for VPN 509 label on the ingress PE, the label action is "push"). As 510 mentioned in Section 1.1 the value of the "path-index" stored 511 in path may not necessarily be the same value of the location of 512 the path in the pathlist. 514 5. Move to the parent of the chosen path "Pi" 516 6. If the chosen path "Pi" is recursive, move to its parent prefix 517 and go to step 2 519 7. If the chosen path is non-recursive move to its parent adjacency. 520 Otherwise go to the next step. 522 8. Encapsulate the packet in the layer string specified by the 523 adjacency and send the packet out. 525 Let's apply the above forwarding steps to the forwarding chain 526 depicted in Figure 2 in Section 2. Suppose a packet arrives at 527 ingress PE iPE from an external neighbor. Assume the packet matches 528 the VPN prefix VPN-IP1. While walking the forwarding chain, the 529 forwarding engine applies a hashing algorithm to choose the path and 530 the hashing at the BGP level yields path 0 while the hashing at the 531 IGP level yields path 1. In that case, the packet will be sent out 532 of interface I2 with the label stack "IGP-L12,VPN-L11". 534 5. Handling Platforms with Limited Levels of Hierarchy 536 This section describes the construction of the forwarding chain if a 537 platform does not support the number of recursion levels required to 538 resolve the NLRIs. There are two main design objectives 540 o Being able to reduce the number of hierarchical levels from any 541 arbitrary value to a smaller arbitrary value that can be 542 supported by the forwarding engine 544 o Minimal modifications to the forwarding algorithm due to such 545 reduction. 547 5.1. Flattening the Forwarding Chain 549 Let's consider a pathlist associated with the leaf "R1" consisting 550 of the list of paths . Assume that the leaf "R1" has 551 an Outlabel-list . Suppose the path Pi is a 552 recursive path that resolves via a prefix represented by the leaf 553 "R2". The leaf "R2" itself is pointing to a pathlist consisting of 554 the paths 556 If the platform supports the number of hierarchy levels of the 557 forwarding chain, then a packet that uses the path "Pi" will be 558 forwarded as follows: 560 1. The forwarding engine is now at leaf "R1" 562 2. So it moves to its parent pathlist, which contains the list . 565 3. The forwarding engine applies a hashing algorithm and picks the 566 path "Pi". So now the forwarding engine is at the path "Pi" 568 4. The forwarding engine retrieves the label "Li" from the outlabel- 569 list attached to the leaf "R1" and applies the label action 571 5. The path "Pi" uses the leaf "R2" 573 6. The forwarding engine walks forward to the leaf "R2" for 574 resolution 576 7. The forwarding plane performs a hash to pick a path among the 577 pathlist of the leaf "R2", which is 579 8. Suppose the forwarding engine picks the path "Qj" 581 9. Now the forwarding engine continues the walk to the parent of 582 "Qj" 584 Suppose the platform cannot support the number of hierarchy levels 585 in the forwarding chain. FIB needs to reduce the number of hierarchy 586 levels. The idea of reducing the number of hierarchy levels is to 587 "flatten" two chain levels into a single level. The "flattening" 588 steps are as follows 590 1. FIB wants to reduce the number of levels used by "Pi" by 1 592 2. FIB walks to the parent of "Pi", which is the leaf "R2" 594 3. FIB extracts the parent pathlist of the leaf "R2", which is 597 4. FIB also extracts the OutLabel-list(R2) associated with the leaf 598 "R2". Remember that OutLabel-list(R2) = 600 5. FIB replaces the path "Pi", with the list of paths 603 6. Hence the path list now becomes " 606 7. The path index stored inside the locations "Q1", "Q2", ..., "Qm" 607 must all be "i" because the index "i" refers to the label "Li" 608 associated with leaf "R1" 610 8. FIB attaches an OutLabel-list with the new pathlist as follows: 611 . The size of the label list associated with the 613 flattened pathlist equals the size of the pathlist. Hence there 614 is a 1-1 mapping between every path in the "flattened" pathlist 615 and the OutLabel-list associated with it. 617 It is noteworthy to mention that the labels in the outlabel-list 618 associated with the "flattened" pathlist may be stored in the same 619 memory location as the path itself to avoid additional memory 620 access. But that is an implementation detail that is beyond the 621 scope of this document. 623 The same steps can be applied to all paths in the pathlist so that all paths are "flattened" thereby reducing the 625 number of hierarchical levels by one. Note that that "flattening" a 626 pathlist pulls in all paths of the parent paths, a desired feature 627 to utilize all ECMP/UCMP paths at all levels. A platform that has a 628 limit on the number of paths in a pathlist for any given leaf may 629 choose to reduce the number paths using methods that are beyond the 630 scope of this document. 632 The steps can be recursively applied to other paths at the same 633 levels or other levels to recursively reduce the number of 634 hierarchical levels to an arbitrary value so as to accommodate the 635 capability of the forwarding engine. 637 Because a flattened pathlist may have an associated OutLabel-list 638 the forwarding behavior has to be slightly modified. The 639 modification is done by adding the following step right after step 4 640 in Section 4. 642 5. If there is an OutLabel-list associated with the pathlist, then 643 if the path "Pi" is chosen by the hashing algorithm, retrieve the 644 label at location "i" in that OutLabel-list and apply the label 645 action of that label on the packet 647 In the next subsection, we apply the steps in this subsection to a 648 sample scenario. 650 5.2. Example: Flattening a forwarding chain 652 This example uses a case of inter-AS option C [7] where there are 3 653 levels of hierarchy. Figure 4 illustrates the sample topology. To 654 force 3 levels of hierarchy, the ASBRs on the ingress domain (domain 655 1) advertise the core routers of the egress domain (domain 2) to the 656 ingress PE (iPE) via BGP-LU [3] instead of redistributing them into 657 the IGP of domain 1. The end result is that the ingress PE (iPE) has 658 2 levels of recursion for the VPN prefix VPN-IP1 and VPN2-IP2. 660 Domain 1 Domain 2 661 +-------------+ +-------------+ 662 | | | | 663 | LDP/SR Core | | LDP/SR core | 664 | | | | 665 | (192.0.2.4) | | 666 | ASBR11-------ASBR21........ePE1(192.0.2.1) 667 | | \ / | . . |\ 668 | | \ / | . . | \ 669 | | \ / | . . | \ 670 | | \/ | .. | \VPN-IP1(198.51.100.0/24) 671 | | /\ | . . | /VRF "Blue" ASn: 65000 672 | | / \ | . . | / 673 | | / \ | . . | / 674 | | / \ | . . |/ 675 iPE ASBR12-------ASBR22........ePE2 (192.0.2.2) 676 | (192.0.2.5) | |\ 677 | | | | \ 678 | | | | \ 679 | | | | \VRF "Blue" ASn: 65000 680 | | | | /VPN-IP2(203.0.113.0/24) 681 | | | | / 682 | | | | / 683 | | | |/ 684 | ASBR13-------ASBR23........ePE3(192.0.2.3) 685 | (192.0.2.6) | | 686 | | | | 687 | | | | 688 +-------------+ +-------------+ 689 <=========== <========= <============ 690 Advertise ePEx Advertise Redistribute 691 Using iBGP-LU ePEx Using IGP into 692 eBGP-LU BGP 694 Figure 4 : Sample 3-level hierarchy topology 696 We will make the following assumptions about connectivity 698 o In "domain 2", both ASBR21 and ASBR22 can reach both ePE1 and 699 ePE2 using the same distance 701 o In "domain 2", only ASBR23 can reach ePE3 703 o In "domain 1", iPE (the ingress PE) can reach ASBR11, ASBR12, and 704 ASBR13 via IGP using the same distance. 706 We will make the following assumptions about the labels 707 o The VPN labels advertised by ePE1 and ePE2 for prefix VPN-IP1 are 708 VPN-L11 and VPN-L21, respectively 710 o The VPN labels advertised by ePE2 and ePE3 for prefix VPN-IP2 are 711 VPN-L22 and VPN-L32, respectively 713 o The labels advertised by ASBR11 to iPE using BGP-LU [3] for the 714 egress PEs ePE1 and ePE2 are LASBR111(ePE1) and LASBR112(ePE2), 715 respectively. 717 o The labels advertised by ASBR12 to iPE using BGP-LU [3] for the 718 egress PEs ePE1 and ePE2 are LASBR121(ePE1) and LASBR122(ePE2), 719 respectively 721 o The label advertised by ASBR13 to iPE using BGP-LU [3] for the 722 egress PE ePE3 is LASBR13(ePE3) 724 o The IGP labels advertised by the next hops directly connected to 725 iPE towards ASBR11, ASBR12, and ASBR13 in the core of domain 1 726 are IGP-L11, IGP-L12, and IGP-L13, respectively. 728 o Both the routers ASBR21 and ASBR22 of Domain 2 advertise the same 729 label LASBR21 and LASBR22 to the egress PEs ePE1 and ePE2, 730 respectively, to the routers ASBR11 and ASBR22 of Domain 1 732 o The router ASBR23 of Domain 2 advertises the label LASBR23 for 733 the egress PE ePE3 to the router ASBR13 of Domain 1 735 Based on these connectivity assumptions and the topology in Figure 736 4, the routing table on iPE is 737 65000: 198.51.100.0/24 738 via ePE1 (192.0.2.1), VPN Label: VPN-L11 739 via ePE2 (192.0.2.2), VPN Label: VPN-L21 740 65000: 203.0.113.0/24 741 via ePE1 (192.0.2.2), VPN Label: VPN-L22 742 via ePE2 (192.0.2.3), VPN Label: VPN-L32 744 192.0.2.1/32 (ePE1) 745 Via ASBR11, BGP-LU Label: LASBR111(ePE1) 746 Via ASBR12, BGP-LU Label: LASBR121(ePE1) 747 192.0.2.2/32 (ePE2) 748 Via ASBR11, BGP-LU Label: LASBR112(ePE2) 749 Via ASBR12, BGP-LU Label: LASBR122(ePE2) 750 192.0.2.3/32 (ePE3) 751 Via ASBR13, BGP-LU Label: LASBR13(ePE3) 753 192.0.2.4/32 (ASBR11) 754 via Core, Label: IGP-L11 755 192.0.2.5/32 (ASBR12) 756 via Core, Label: IGP-L12 757 192.0.2.6/32 (ASBR13) 758 via Core, Label: IGP-L13 760 The diagram in Figure 5 illustrates the forwarding chain in iPE 761 assuming that the forwarding hardware in iPE supports 3 levels of 762 hierarchy. The leaves corresponding to the ABSRs on domain 1 763 (ASBR11, ASBR12, and ASBR13) are at the bottom of the hierarchy. 764 There are few important points: 766 o Because the hardware supports the required depth of hierarchy, 767 the sizes of a pathlist equal the size of the label list 768 associated with the leaves using this pathlist 770 o The index inside the pathlist entry indicates the label that will 771 be picked from the Outlabel-List associated with the child leaf 772 if that path is chosen by the forwarding engine hashing function. 774 Outlabel-List Outlabel-List 775 For VPN-IP1 For VPN-IP2 776 +------------+ +--------+ +-------+ +------------+ 777 | VPN-L11 |<---| VPN-IP1| |VPN-IP2|-->| VPN-L22 | 778 +------------+ +---+----+ +---+---+ +------------+ 779 | VPN-L21 | | | | VPN-L32 | 780 +------------+ | | +------------+ 781 | | 782 V V 783 +---+---+ +---+---+ 784 | 0 | 1 | | 0 | 1 | 785 +-|-+-\-+ +-/-+-\-+ 786 | \ / \ 787 | \ / \ 788 | \ / \ 789 | \ / \ 790 v \ / \ 791 +-----+ +-----+ +-----+ 792 +----+ ePE1| |ePE2 +-----+ | ePE3+-----+ 793 | +--+--+ +-----+ | +--+--+ | 794 v | / v | v 795 +--------------+ | / +--------------+ | +-------------+ 796 |LASBR111(ePE1)| | / |LASBR112(ePE2)| | |LASBR13(ePE3)| 797 +--------------+ | / +--------------+ | +-------------+ 798 |LASBR121(ePE1)| | / |LASBR122(ePE2)| | Outlabel-List 799 +--------------+ | / +--------------+ | For ePE3 800 Outlabel-List | / Outlabel-List | 801 For ePE1 | / For ePE2 | 802 | / | 803 | / | 804 | / | 805 v / v 806 +---+---+ Shared Pathlist +---+ Pathlist 807 | 0 | 1 | For ePE1 and ePE2 | 0 | For ePE3 808 +-|-+-\-+ +-|-+ 809 | \ | 810 | \ | 811 | \ | 812 | \ | 813 v \ v 814 +------+ +------+ +------+ 815 +---+ASBR11| |ASBR12+--+ |ASBR13+---+ 816 | +------+ +------+ | +------+ | 817 v v v 818 +-------+ +-------+ +-------+ 819 |IGP-L11| |IGP-L12| |IGP-L13| 820 +-------+ +-------+ +-------+ 822 Figure 5 : Forwarding Chain for hardware supporting 3 Levels 824 Now suppose the hardware on iPE (the ingress PE) supports 2 levels 825 of hierarchy only. In that case, the 3-levels forwarding chain in 826 Figure 5 needs to be "flattened" into 2 levels only. 828 Outlabel-List Outlabel-List 829 For VPN-IP1 For VPN-IP2 830 +------------+ +-------+ +-------+ +------------+ 831 | VPN-L11 |<---|VPN-IP1| | VPN-IP2|--->| VPN-L22 | 832 +------------+ +---+---+ +---+---+ +------------+ 833 | VPN-L21 | | | | VPN-L32 | 834 +------------+ | | +------------+ 835 | | 836 | | 837 | | 838 Flattened | | Flattened 839 pathlist V V pathlist 840 +===+===+ +===+===+===+ +==============+ 841 +--------+ 0 | 1 | | 0 | 0 | 1 +---->|LASBR112(ePE2)| 842 | +=|=+=\=+ +=/=+=/=+=\=+ +==============+ 843 v | \ / / \ |LASBR122(ePE2)| 844 +==============+ | \ +-----+ / \ +==============+ 845 |LASBR111(ePE1)| | \/ / \ |LASBR13(ePE3) | 846 +==============+ | /\ / \ +==============+ 847 |LASBR121(ePE1)| | / \ / \ 848 +==============+ | / \ / \ 849 | / \ / \ 850 | / + + \ 851 | + | | \ 852 | | | | \ 853 v v v v \ 854 +------+ +------+ +------+ 855 +----|ASBR11| |ASBR12+---+ |ASBR13+---+ 856 | +------+ +------+ | +------+ | 857 v v v 858 +-------+ +-------+ +-------+ 859 |IGP-L11| |IGP-L12| |IGP-L13| 860 +-------+ +-------+ +-------+ 862 Figure 6 : Flattening 3 levels to 2 levels of Hierarchy on iPE 864 Figure 6 represents one way to "flatten" a 3 levels hierarchy into 865 two levels. There are few important points: 867 o As mentioned in Section 5.1 a flattened pathlist may have label 868 lists associated with them. The size of the label list associated 869 with a flattened pathlist equals the size of the pathlist. Hence 870 it is possible that an implementation includes these label lists 871 in the flattened pathlist itself 873 o Again as mentioned in Section 5.1, the size of a flattened 874 pathlist may not be equal to the size of the OutLabel-lists of 875 leaves using the flattened pathlist. So the indices inside a 876 flattened pathlist still indicate the label index in the 877 Outlabel-Lists of the leaves using that pathlist. Because the 878 size of the flattened pathlist may be different from the size of 879 the OutLabel-lists of the leaves, the indices may be repeated. 881 o Let's take a look at the flattened pathlist used by the prefix 882 "VPN-IP2", The pathlist associated with the prefix "VPN-IP2" has 883 three entries. 885 o The first and second entry have index "0". This is because 886 both entries correspond to ePE2. Hence when hashing performed 887 by the forwarding engine results in using first or the second 888 entry in the pathlist, the forwarding engine will pick the 889 correct VPN label "VPN-L22", which is the label advertised by 890 ePE2 for the prefix "VPN-IP2" 892 o The third entry has the index "1". This is because the third 893 entry corresponds to ePE3. Hence when the hashing is performed 894 by the forwarding engine results in using the third entry in 895 the flattened pathlist, the forwarding engine will pick the 896 correct VPN label "VPN-L32", which is the label advertised by 897 "ePE3" for the prefix "VPN-IP2" 899 Now let's try and apply the forwarding steps in Section 4 together 900 with the additional step in Section 5.1 to the flattened forwarding 901 chain illustrated in Figure 6. 903 o Suppose a packet arrives at "iPE" and matches the VPN prefix 904 "VPN-IP2" 906 o The forwarding engine walks to the parent of the "VPN-IP2", which 907 is the flattened pathlist and applies a hashing algorithm to pick 908 a path 910 o Suppose the hashing by the forwarding engine picks the second 911 path in the flattened pathlist associated with the leaf "VPN- 912 IP2". 914 o Because the second path has the index "0", the label "VPN-L22" is 915 pushed on the packet 917 o Next the forwarding engine picks the second label from the 918 Outlabel-Array associated with the flattened pathlist. Hence the 919 next label that is pushed is "LASBR122(ePE2)" 921 o The forwarding engine now moves to the parent of the flattened 922 pathlist corresponding to the second path. The parent is the IGP 923 label leaf corresponding to "ASBR12" 925 o So the packet is forwarded towards the ASBR "ASBR12" and the IGP 926 label at the top will be "L12" 928 Based on the above steps, a packet arriving at iPE and destined to 929 the prefix VPN-L22 reaches its destination as follows 931 o iPE sends the packet along the shortest path towards ASBR12 with 932 the following label stack starting from the top: {L12, 933 LASBR122(ePE2), VPN-L22}. 935 o The penultimate hop of ASBR12 pops the top label "L12". Hence the 936 packet arrives at ASBR12 with the label stack {LASBR122(ePE2), 937 VPN-L22} where "LASBR12(ePE2)" is the top label. 939 o ASBR12 swaps "LASBR122(ePE2)" with the label "LASBR22(ePE2)", 940 which is the label advertised by ASBR22 for the ePE2 (the egress 941 PE). 943 o ASBR22 receives the packet with "LASBR22(ePE2)" at the top. 945 o Hence ASBR22 swaps "LASBR22(ePE2)" with the IGP label for ePE2 946 advertised by the next-hop towards ePE2 in domain 2, and sends 947 the packet along the shortest path towards ePE2. 949 o The penultimate hop of ePE2 pops the top label. Hence ePE2 950 receives the packet with the top label VPN-L22 at the top 952 o ePE2 pops "VPN-L22" and sends the packet as a pure IP packet 953 towards the destination VPN-IP2. 955 6. Forwarding Chain Adjustment at a Failure 957 The hierarchical and shared structure of the forwarding chain 958 explained in the previous section allows modifying a small number of 959 forwarding chain objects to re-route traffic to a pre-calculated 960 equal-cost or backup path without the need to modify the possibly 961 very large number of BGP prefixes. In this section, we go over 962 various core and edge failure scenarios to illustrate how FIB 963 manager can utilize the forwarding chain structure to achieve BGP 964 prefix independent convergence. 966 6.1. BGP-PIC core 968 This section describes the adjustments to the forwarding chain when 969 a core link or node fails but the BGP next-hop remains reachable. 971 There are two case: remote link failure and attached link failure. 972 Node failures are treated as link failures. 974 When a remote link or node fails, IGP on the ingress PE receives 975 advertisement indicating a topology change so IGP re-converges to 976 either find a new next-hop and/or outgoing interface or remove the 977 path completely from the IGP prefix used to resolve BGP next-hops. 978 IGP and/or LDP download the modified IGP leaves with modified 979 outgoing labels for labeled core. 981 When a local link fails, FIB manager detects the failure almost 982 immediately. The FIB manager marks the impacted path(s) as unusable 983 so that only useable paths are used to forward packets. Hence only 984 IGP pathlists with paths using the failed local link need to be 985 modified. All other pathlists are not impacted. Note that in this 986 particular case there is actually no need even to backwalk to IGP 987 leaves to adjust the OutLabel-Lists because FIB can rely on the 988 path-index stored in the useable paths in the pathlist to pick the 989 right label. 991 It is noteworthy to mention that because FIB manager modifies the 992 forwarding chain starting from the IGP leaves only. BGP pathlists 993 and leaves are not modified. Hence traffic restoration occurs within 994 the time frame of IGP convergence, and, for local link failure, 995 assuming a backup path has been precomputed, within the timeframe of 996 local detection (e.g. 50ms). Examples of solutions that pre- 997 computing backup paths are IP FRR [15] remote LFA [16], Ti-LFA [14] 998 and MRT [17] or eBGP path having a backup path [9]. 1000 Let's apply the procedure mentioned in this subsection to the 1001 forwarding chain depicted in Figure 2. Suppose a remote link failure 1002 occurs and impacts the first ECMP IGP path to the remote BGP next- 1003 hop. Upon IGP convergence, the IGP pathlist used by the BGP next-hop 1004 is updated to reflect the new topology (one path instead of two). As 1005 soon as the IGP convergence is effective for the BGP next-hop entry, 1006 the new forwarding state is immediately available to all dependent 1007 BGP prefixes. The same behavior would occur if the failure was local 1008 such as an interface going down. As soon as the IGP convergence is 1009 complete for the BGP next-hop IGP route, all its BGP depending 1010 routes benefit from the new path. In fact, upon local failure, if 1011 LFA protection is enabled for the IGP route to the BGP next-hop and 1012 a backup path was pre-computed and installed in the pathlist, upon 1013 the local interface failure, the LFA backup path is immediately 1014 activated (e.g. sub-50msec) and thus protection benefits all the 1015 depending BGP traffic through the hierarchical forwarding dependency 1016 between the routes. 1018 6.2. BGP-PIC edge 1020 This section describes the adjustments to the forwarding chains as a 1021 result of edge node or edge link failure. 1023 6.2.1. Adjusting forwarding Chain in egress node failure 1025 When an edge node fails, IGP on neighboring core nodes send route 1026 updates indicating that the edge node is no longer reachable. IGP 1027 running on the iBGP peers instructs FIB to remove the IP and label 1028 leaves corresponding to the failed edge node from FIB. So FIB 1029 manager performs the following steps: 1031 o FIB manager deletes the IGP leaf corresponding to the failed edge 1032 node 1034 o FIB manager backwalks to all dependent BGP pathlists and marks 1035 that path using the deleted IGP leaf as unresolved 1037 o Note that there is no need to modify the possibly large number of 1038 BGP leaves because each path in the pathlist carries its path 1039 index and hence the correct outgoing label will be picked. 1040 Consider for example the forwarding chain depicted in Figure 2. 1041 If the 1st BGP path becomes unresolved, then the forwarding 1042 engine will only use the second path for forwarding. Yet the path 1043 index of that single resolved path will still be 1 and hence the 1044 label VPN-L12 will be pushed. 1046 6.2.2. Adjusting Forwarding Chain on PE-CE link Failure 1048 Suppose the link between an edge router and its external peer fails. 1049 There are two scenarios (1) the edge node attached to the failed 1050 link performs next-hop self and (2) the edge node attached to the 1051 failure advertises the IP address of the failed link as the next-hop 1052 attribute to its iBGP peers. 1054 In the first case, the rest of iBGP peers will remain unaware of the 1055 link failure and will continue to forward traffic to the edge node 1056 until the edge node attached to the failed link withdraws the BGP 1057 prefixes. If the destination prefixes are multi-homed to another 1058 iBGP peer, say ePE2, then FIB manager on the edge router detecting 1059 the link failure applies the following steps: 1061 o FIB manager backwalks to the BGP pathlists marks the path through 1062 the failed link to the external peer as unresolved 1064 o Hence traffic will be forwarded used the backup path towards ePE2 1066 o For labeled traffic 1068 o The Outlabel-List attached to the BGP leaf already contains 1069 an entry corresponding to the backup path. 1071 o The label entry in OutLabel-List corresponding to the 1072 internal path to backup egress PE has swap action to the 1073 label advertised by backup egress PE 1075 o For an arriving label packet (e.g. VPN), the top label is 1076 swapped with the label advertised by backup egress PE and the 1077 packet is sent towards that backup egress PE 1079 o For unlabeled traffic, packets are simply redirected towards 1080 backup egress PE. 1082 In the second case where the edge router uses the IP address of the 1083 failed link as the BGP next-hop, the edge router will still perform 1084 the previous steps. But, unlike the case of next-hop self, IGP on 1085 failed edge node informs the rest of the iBGP peers that IP address 1086 of the failed link is no longer reachable. Hence the FIB manager on 1087 iBGP peers will delete the IGP leaf corresponding to the IP prefix 1088 of the failed link. The behavior of the iBGP peers will be identical 1089 to the case of edge node failure outlined in Section 6.2.1. 1091 It is noteworthy to mention that because the edge link failure is 1092 local to the edge router, sub-50 msec convergence can be achieved as 1093 described in [9]. 1095 Let's try to apply the case of next-hop self to the forwarding chain 1096 depicted in Figure 3. After failure of the link between ePE1 and CE, 1097 the forwarding engine will route traffic arriving from the core 1098 towards VPN-NH2 with path-index=1. A packet arriving from the core 1099 will contain the label VPN-L11 at top. The label VPN-L11 is swapped 1100 with the label VPN-L21 and the packet is forwarded towards ePE2. 1102 6.3. Handling Failures for Flattened Forwarding Chains 1104 As explained in the in Section 5 if the number of hierarchy levels 1105 of a platform cannot support the native number of hierarchy levels 1106 of a recursive forwarding chain, the instantiated forwarding chain 1107 is constructed by flattening two or more levels. Hence a 3 levels 1108 chain in Figure 5 is flattened into the 2 levels chain in Figure 6. 1110 While reducing the benefits of BGP-PIC, flattening one hierarchy 1111 into a shallower hierarchy does not always result in a complete loss 1112 of the benefits of the BGP-PIC. To illustrate this fact suppose 1113 ASBR12 is no longer reachable in domain 1. If the platform supports 1114 the full hierarchy depth, the forwarding chain is the one depicted 1115 in Figure 5 and hence the FIB manager needs to backwalk one level to 1116 the pathlist shared by "ePE1" and "ePE2" and adjust it. If the 1117 platform supports 2 levels of hierarchy, then a useable forwarding 1118 chain is the one depicted in Figure 6. In that case, if ASBR12 is no 1119 longer reachable, the FIB manager has to backwalk to the two 1120 flattened pathlists and updates both of them. 1122 The main observation is that the loss of convergence speed due to 1123 the loss of hierarchy depth depends on the structure of the 1124 forwarding chain itself. To illustrate this fact, let's take two 1125 extremes. Suppose the forwarding objects in level i+1 depend on the 1126 forwarding objects in level i. If every object on level i+1 depends 1127 on a separate object in level i, then flattening level i into level 1128 i+1 will not result in loss of convergence speed. Now let's take the 1129 other extreme. Suppose "n" objects in level i+1 depend on 1 object 1130 in level i. Now suppose FIB flattens level i into level i+1. If a 1131 topology change results in modifying the single object in level i, 1132 then FIB has to backwalk and modify "n" objects in the flattened 1133 level, thereby losing all the benefit of BGP-PIC. Experience shows 1134 that flattening forwarding chains usually results in moderate loss 1135 of BGP-PIC benefits. Further analysis is needed to corroborate and 1136 quantify this statement. 1138 7. Properties 1140 7.1. Coverage 1142 All the possible failures, except CE node failure, are covered, 1143 whether they impact a local or remote IGP path or a local or remote 1144 BGP next-hop as described in Section 6. This section provides 1145 details for each failure and now the hierarchical and shared FIB 1146 structure proposed in this document allows recovery that does not 1147 depend on number of BGP prefixes. 1149 7.1.1. A remote failure on the path to a BGP next-hop 1151 Upon IGP convergence, the IGP leaf for the BGP next-hop is updated 1152 upon IGP convergence and all the BGP depending routes leverage the 1153 new IGP forwarding state immediately. Details of this behavior can 1154 be found in Section 6.1. 1156 This BGP resiliency property only depends on IGP convergence and is 1157 independent of the number of BGP prefixes impacted. 1159 7.1.2. A local failure on the path to a BGP next-hop 1161 Upon LFA protection, the IGP leaf for the BGP next-hop is updated to 1162 use the precomputed LFA backup path and all the BGP depending routes 1163 leverage this LFA protection. Details of this behavior can be found 1164 in Section 6.1. 1166 This BGP resiliency property only depends on LFA protection and is 1167 independent of the number of BGP prefixes impacted. 1169 7.1.3. A remote iBGP next-hop fails 1171 Upon IGP convergence, the IGP leaf for the BGP next-hop is deleted 1172 and all the depending BGP Path-Lists are updated to either use the 1173 remaining ECMP BGP best-paths or if none remains available to 1174 activate precomputed backups. Details about this behavior can be 1175 found in Section 6.2.1. 1177 This BGP resiliency property only depends on IGP convergence and is 1178 independent of the number of BGP prefixes impacted. 1180 7.1.4. A local eBGP next-hop fails 1182 Upon local link failure detection, the adjacency to the BGP next-hop 1183 is deleted and all the depending BGP pathlists are updated to either 1184 use the remaining ECMP BGP best-paths or if none remains available 1185 to activate precomputed backups. Details about this behavior can be 1186 found in Section 6.2.2. 1188 This BGP resiliency property only depends on local link failure 1189 detection and is independent of the number of BGP prefixes impacted. 1191 7.2. Performance 1193 When the failure is local (a local IGP next-hop failure or a local 1194 eBGP next-hop failure), a pre-computed and pre-installed backup is 1195 activated by a local-protection mechanism that does not depend on 1196 the number of BGP destinations impacted by the failure. Sub-50msec 1197 is thus possible even if millions of BGP routes are impacted. 1199 When the failure is remote (a remote IGP failure not impacting the 1200 BGP next-hop or a remote BGP next-hop failure), an alternate path is 1201 activated upon IGP convergence. All the impacted BGP destinations 1202 benefit from a working alternate path as soon as the IGP convergence 1203 occurs for their impacted BGP next-hop even if millions of BGP 1204 routes are impacted. 1206 Appendix A puts the BGP PIC benefits in perspective by providing 1207 some results using actual numbers. 1209 7.3. Automated 1211 The BGP PIC solution does not require any operator involvement. The 1212 process is entirely automated as part of the FIB implementation. 1214 The salient points enabling this automation are: 1216 o Extension of the BGP Best Path to compute more than one primary 1217 ([10]and [11]) or backup BGP next-hop ([5] and [12]). 1219 o Sharing of BGP Path-list across BGP destinations with same 1220 primary and backup BGP next-hop 1222 o Hierarchical indirection and dependency between BGP pathlist and 1223 IGP pathlist 1225 7.4. Incremental Deployment 1227 As soon as one router supports BGP PIC solution, it benefits from 1228 all its benefits without any requirement for other routers to 1229 support BGP PIC. 1231 8. Security Considerations 1233 The behavior described in this document is internal functionality 1234 to a router that result in significant improvement to convergence 1235 time as well as reduction in CPU and memory used by FIB while not 1236 showing change in basic routing and forwarding functionality. As 1237 such no additional security risk is introduced by using the 1238 mechanisms proposed in this document. 1240 9. IANA Considerations 1242 No requirements for IANA 1244 10. Conclusions 1246 This document proposes a hierarchical and shared forwarding chain 1247 structure that allows achieving BGP prefix independent 1248 convergence, and in the case of locally detected failures, sub-50 1249 msec convergence. A router can construct the forwarding chains in 1250 a completely transparent manner with zero operator intervention 1251 thereby supporting smooth and incremental deployment. 1253 11. References 1255 11.1. Normative References 1257 [1] Rekhter, Y., Li, T., and S. Hares, "A Border Gateway Protocol 1258 4 (BGP-4), RFC 4271, January 2006 1260 [2] Bates, T., Chandra, R., Katz, D., and Rekhter Y., 1261 "Multiprotocol Extensions for BGP", RFC 4760, January 2007 1263 [3] Y. Rekhter and E. Rosen, " Carrying Label Information in BGP- 1264 4", RFC 8277, October 2017 1266 [4] Andersson, L., Minei, I., and B. Thomas, "LDP Specification", 1267 RFC 5036, October 2007 1269 11.2. Informative References 1271 [5] Marques,P., Fernando, R., Chen, E, Mohapatra, P., Gredler, H., 1272 "Advertisement of the best external route in BGP", draft-ietf- 1273 idr-best-external-05.txt, January 2012. 1275 [6] Wu, J., Cui, Y., Metz, C., and E. Rosen, "Softwire Mesh 1276 Framework", RFC 5565, June 2009. 1278 [7] Rosen, E. and Y. Rekhter, "BGP/MPLS IP Virtual Private 1279 Networks (VPNs)", RFC 4364, February 2006. 1281 [8] De Clercq, J. , Ooms, D., Prevost, S., Le Faucheur, F., 1282 "Connecting IPv6 Islands over IPv4 MPLS Using IPv6 Provider 1283 Edge Routers (6PE)", RFC 4798, February 2007 1285 [9] O. Bonaventure, C. Filsfils, and P. Francois. "Achieving sub- 1286 50 milliseconds recovery upon bgp peering link failures, " 1287 IEEE/ACM Transactions on Networking, 15(5):1123-1135, 2007 1289 [10] D. Walton, A. Retana, E. Chen, J. Scudder, "Advertisement of 1290 Multiple Paths in BGP", RFC 7911, July 2016 1292 [11] R. Raszuk, R. Fernando, K. Patel, D. McPherson, K. Kumaki, 1293 "Distribution of diverse BGP paths", RFC 6774, November 2012 1295 [12] P. Mohapatra, R. Fernando, C. Filsfils, and R. Raszuk, "Fast 1296 Connectivity Restoration Using BGP Add-path", draft-pmohapat- 1297 idr-fast-conn-restore-03, Jan 2013 1299 [13] A. Bashandy, C. Filsfils, S. Previdi, B. Decraene, S. 1300 Litkowski, M. Horneffer, R. Shakir, "Segment Routing with MPLS 1301 data plane", draft-ietf-spring-segment-routing-mpls-12 (work 1302 in progress), February 2018 1304 [14] A. Bashandy, C. Filsfils, B. Decraene, P. Francois, " Topology 1305 Independent Fast Reroute using Segment Routing", draft- 1306 bashandy-rtgwg-segment-routing-ti-lfa-02 (work in progress), 1307 August 2018 1309 [15] M. Shand and S. Bryant, "IP Fast Reroute Framework", RFC 5714, 1310 January 2010 1312 [16] S. Bryant, C. Filsfils, S. Previdi, M. Shand, N So, " Remote 1313 Loop-Free Alternate (LFA) Fast Reroute (FRR)", RFC 7490 April 1314 2015 1316 [17] A. Atlas, C. Bowers, G. Enyedi, " An Architecture for IP/LDP 1317 Fast-Reroute Using Maximally Redundant Trees", RFC 7812, June 1318 2016 1320 12. Acknowledgments 1322 Special thanks to Neeraj Malhotra, Yuri Tsier for the valuable 1323 help 1325 Special thanks to Bruno Decraene for the valuable comments 1327 This document was prepared using 2-Word-v2.0.template.dot. 1329 Authors' Addresses 1331 Ahmed Bashandy 1332 Arrcus 1333 Email: abashandy.ietf@gmail.com 1335 Clarence Filsfils 1336 Cisco Systems 1337 Brussels, Belgium 1338 Email: cfilsfil@cisco.com 1340 Prodosh Mohapatra 1341 Sproute Networks 1342 Email: mpradosh@yahoo.com 1344 Appendix A. Perspective 1346 The following table puts the BGP PIC benefits in perspective 1347 assuming 1349 o 1M impacted BGP prefixes 1351 o IGP convergence ~ 500 msec 1353 o local protection ~ 50msec 1355 o FIB Update per BGP destination ~ 100usec conservative, 1357 ~ 10usec optimistic 1359 o BGP Convergence per BGP destination ~ 200usec conservative, 1361 ~ 100usec optimistic 1363 Without PIC With PIC 1365 Local IGP Failure 10 to 100sec 50msec 1367 Local BGP Failure 100 to 200sec 50msec 1369 Remote IGP Failure 10 to 100sec 500msec 1371 Local BGP Failure 100 to 200sec 500msec 1373 Upon local IGP next-hop failure or remote IGP next-hop failure, the 1374 existing primary BGP next-hop is intact and usable hence the 1375 resiliency only depends on the ability of the FIB mechanism to 1376 reflect the new path to the BGP next-hop to the depending BGP 1377 destinations. Without BGP PIC, a conservative back-of-the-envelope 1378 estimation for this FIB update is 100usec per BGP destination. An 1379 optimistic estimation is 10usec per entry. 1381 Upon local BGP next-hop failure or remote BGP next-hop failure, 1382 without the BGP PIC mechanism, a new BGP Best-Path needs to be 1383 recomputed and new updates need to be sent to peers. This depends on 1384 BGP processing time that will be shared between best-path 1385 computation, RIB update and peer update. A conservative back-of-the- 1386 envelope estimation for this is 200usec per BGP destination. An 1387 optimistic estimation is 100usec per entry.