idnits 2.17.1 draft-ietf-rtgwg-ipfrr-notvia-addresses-08.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 : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == Line 491 has weird spacing: '... a Ps a ...' -- The document seems to lack a disclaimer for pre-RFC5378 work, but may have content which was first submitted before 10 November 2008. If you have contacted all the original authors and they are all willing to grant the BCP78 rights to the IETF Trust, then this is fine, and you can ignore this comment. If not, you may need to add the pre-RFC5378 disclaimer. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (December 21, 2011) is 4503 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Outdated reference: A later version (-12) exists of draft-ietf-rtgwg-ordered-fib-05 Summary: 0 errors (**), 0 flaws (~~), 3 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group S. Bryant 3 Internet-Draft S. Previdi 4 Intended status: Informational Cisco Systems 5 Expires: June 23, 2012 M. Shand 6 Individual Contributor 7 December 21, 2011 9 IP Fast Reroute Using Not-via Addresses 10 draft-ietf-rtgwg-ipfrr-notvia-addresses-08 12 Abstract 14 This draft describes a mechanism that provides fast reroute in an IP 15 network through encapsulation to "not-via" addresses. A single level 16 of encapsulation is used. The mechanism protects unicast, multicast 17 and LDP traffic against link, router and shared risk group failure, 18 regardless of network topology and metrics. 20 Status of this Memo 22 This Internet-Draft is submitted in full conformance with the 23 provisions of BCP 78 and BCP 79. 25 Internet-Drafts are working documents of the Internet Engineering 26 Task Force (IETF). Note that other groups may also distribute 27 working documents as Internet-Drafts. The list of current Internet- 28 Drafts is at http://datatracker.ietf.org/drafts/current/. 30 Internet-Drafts are draft documents valid for a maximum of six months 31 and may be updated, replaced, or obsoleted by other documents at any 32 time. It is inappropriate to use Internet-Drafts as reference 33 material or to cite them other than as "work in progress." 35 This Internet-Draft will expire on June 23, 2012. 37 Copyright Notice 39 Copyright (c) 2011 IETF Trust and the persons identified as the 40 document authors. All rights reserved. 42 This document is subject to BCP 78 and the IETF Trust's Legal 43 Provisions Relating to IETF Documents 44 (http://trustee.ietf.org/license-info) in effect on the date of 45 publication of this document. Please review these documents 46 carefully, as they describe your rights and restrictions with respect 47 to this document. Code Components extracted from this document must 48 include Simplified BSD License text as described in Section 4.e of 49 the Trust Legal Provisions and are provided without warranty as 50 described in the Simplified BSD License. 52 Table of Contents 54 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 55 2. Overview of Not-via Repairs . . . . . . . . . . . . . . . . . 3 56 2.1. Use of Equal Cost Multi-Path . . . . . . . . . . . . . . . 4 57 2.2. Use of LFA repairs . . . . . . . . . . . . . . . . . . . . 4 58 3. Not-via Repair Path Computation . . . . . . . . . . . . . . . 5 59 3.1. Computing not-via repairs in routing vector protocols . . 6 60 4. Operation of Repairs . . . . . . . . . . . . . . . . . . . . . 6 61 4.1. Node Failure . . . . . . . . . . . . . . . . . . . . . . . 6 62 4.2. Link Failure . . . . . . . . . . . . . . . . . . . . . . . 7 63 4.2.1. Loop Prevention Under Node Failure . . . . . . . . . . 7 64 4.3. Multi-homed Prefixes . . . . . . . . . . . . . . . . . . . 7 65 4.4. Installation of Repair Paths . . . . . . . . . . . . . . . 9 66 5. Compound Failures . . . . . . . . . . . . . . . . . . . . . . 10 67 5.1. Shared Risk Link Groups . . . . . . . . . . . . . . . . . 10 68 5.1.1. Use of LFAs with SRLGs . . . . . . . . . . . . . . . . 14 69 5.2. Local Area Networks . . . . . . . . . . . . . . . . . . . 14 70 5.2.1. Simple LAN Repair . . . . . . . . . . . . . . . . . . 15 71 5.2.2. LAN Component Repair . . . . . . . . . . . . . . . . . 16 72 5.2.3. LAN Repair Using Diagnostics . . . . . . . . . . . . . 17 73 5.3. Multiple Independent Failures . . . . . . . . . . . . . . 17 74 5.3.1. Looping Repairs . . . . . . . . . . . . . . . . . . . 18 75 5.3.2. Outline Solution . . . . . . . . . . . . . . . . . . . 19 76 5.3.3. Looping Repairs . . . . . . . . . . . . . . . . . . . 20 77 5.3.3.1. Dropping Looping Packets . . . . . . . . . . . . . 20 78 5.3.3.2. Computing non-looping Repairs of Repairs . . . . . 21 79 5.3.4. Mixing LFAs and Not-via . . . . . . . . . . . . . . . 23 80 6. Optimizing not-via computations using LFAs . . . . . . . . . . 24 81 7. Multicast . . . . . . . . . . . . . . . . . . . . . . . . . . 25 82 8. Fast Reroute in an MPLS LDP Network. . . . . . . . . . . . . . 25 83 9. Encapsulation . . . . . . . . . . . . . . . . . . . . . . . . 25 84 10. Routing Extensions . . . . . . . . . . . . . . . . . . . . . . 26 85 11. Incremental Deployment . . . . . . . . . . . . . . . . . . . . 26 86 12. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 26 87 13. Security Considerations . . . . . . . . . . . . . . . . . . . 26 88 14. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 27 89 15. Informative References . . . . . . . . . . . . . . . . . . . . 27 90 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 28 92 1. Introduction 94 When a link or a router fails, only the neighbors of the failure are 95 initially aware that the failure has occurred. In a network 96 operating IP fast reroute [RFC5714], the routers that are the 97 neighbors of the failure repair the failure. These repairing routers 98 have to steer packets to their destinations despite the fact that 99 most other routers in the network are unaware of the nature and 100 location of the failure. 102 A common limitation in most IPFRR mechanisms is an inability to 103 indicate the identity of the failure and to explicitly steer the 104 repaired packet round the failure. The extent to which this 105 limitation affects the repair coverage is topology dependent. The 106 mechanism proposed here is to encapsulate the packet to an address 107 that explicitly identifies the network component that the repair must 108 avoid. This produces a repair mechanism, which, provided the network 109 is not partitioned by the failure, will always achieve a repair. 111 2. Overview of Not-via Repairs 113 This section provides a brief overview of the not-via method of 114 IPFRR. 116 A 117 | Bp is the address to use to get 118 | a packet to B not-via P 119 | 120 S----------P----------B. . . . . . . . . .D 121 \ | Bp^ 122 \ | | 123 \ | | 124 \ C | 125 \ | 126 ----------------+ 127 Repair to Bp 129 Figure 1: Not-via repair of router failure 131 Assume that S has a packet for some destination D that it would 132 normally send via P and B, and that S suspects that P has failed. S 133 encapsulates the packet to Bp. The path from S to Bp is the shortest 134 path from S to B not going via P. If the network contains a path from 135 S to B that does not transit router P, i.e. the network is not 136 partitioned by the failure of P, then the packet will be successfully 137 delivered to B. When the packet addressed to Bp arrives at B, B 138 removes the encapsulation and forwards the repaired packet towards 139 its final destination. 141 Note that if the path from B to the final destination includes one or 142 more nodes that are included in the repair path, a packet may back 143 track after the encapsulation is removed. However, because the 144 decapsulating router is always closer to the packet destination than 145 the encapsulating router, the packet will not loop. 147 For complete protection, all of P's neighbors will require a not-via 148 address that allows traffic to be directed to them without traversing 149 P. This is shown in Figure 2. 151 A 152 |Ap 153 | 154 Sp Pa|Pb 155 S----------P----------B 156 Ps|Pc Bp 157 | 158 Cp| 159 C 161 Figure 2: The set of Not-via P Addresses 163 2.1. Use of Equal Cost Multi-Path 165 A router can use an equal cost multi-path (ECMP) repair in place of a 166 not-via repair. 168 A router computing a not-via repair path may subject the repair to 169 ECMP. 171 2.2. Use of LFA repairs 173 The not-via approach provides complete repair coverage and therefore 174 may be used as the sole repair mechanism. There are, however, 175 advantages in using not-via in combination with loop free alternates 176 (LFA) and or downstream paths as documented in [RFC5286]. 178 LFAs are computed on a per destination basis and in general, only a 179 subset of the destinations requiring repair will have a suitable LFA 180 repair. In this case, those destinations which are repairable by 181 LFAs are so repaired and the remainder of the destinations are 182 repaired using the not-via encapsulation. This has the advantage of 183 reducing the volume of traffic that requires encapsulation. On the 184 other hand, the path taken by an LFA repair may be less optimal than 185 that of the equivalent not-via repair for traffic destined to nodes 186 close to the far end of the failure, but may be more optimal for some 187 other traffic. The description in this document assumes that LFAs 188 will be used where available, but the distribution of repairs between 189 the two mechanisms is a local implementation choice. 191 3. Not-via Repair Path Computation 193 The not-via repair mechanism requires that all routers on the path 194 from S to B (Figure 1) have a route to Bp. They can calculate this 195 by failing node P, running a Shortest Path First Algorithm (SPF), and 196 finding the shortest route to B. 198 A router has no simple way of knowing whether it is on the shortest 199 path for any particular repair. It is therefore necessary for every 200 router to calculate the path it would use in the event of any 201 possible router failure. Each router therefore "fails" every router 202 in the network, one at a time, and calculates its own best route to 203 each of the neighbors of that router. In other words, with reference 204 to Figure 1, some router X will consider each router in turn to be P, 205 fail P, and then calculate its own route to each of the not-via P 206 addresses advertised by the neighbors of P. i.e. X calculates its 207 route to Sp, Ap, Bp, and Cp, in each case, not via P. 209 To calculate the repair paths a router has to calculate n-1 SPFs 210 where n is the number of routers in the network. This is expensive 211 to compute. However, the problem is amenable to a solution in which 212 each router (X) proceeds as follows. X first calculates the base 213 topology with all routers functional and determines its normal path 214 to all not-via addresses. This can be performed as part of the 215 normal SPF computation. For each router P in the topology, X then 216 performs the following actions:- 218 1. Removes router P from the topology. 220 2. Performs an incremental SPF (iSPF) [ISPF] on the modified 221 topology. The iSPF process involves detaching the sub-tree 222 affected by the removal of router P, and then re-attaching the 223 detached nodes. However, it is not necessary to run the iSPF to 224 completion. It is sufficient to run the iSPF up to the point 225 where all of the nodes advertising not-via P addresses have been 226 re-attached to the SPT, and then terminate it. 228 3. Reverts to the base topology. 230 This algorithm is significantly less expensive than a set of full 231 SPFs. Thus, although a router has to calculate the repair paths for 232 n-1 failures, the computational effort is much less than n-1 SPFs. 234 Experiments on a selection of real world network topologies with 235 between 40 and 400 nodes suggest that the worst-case computational 236 complexity using the above optimizations is equivalent to performing 237 between 5 and 13 full SPFs. Further optimizations are described in 238 section 6. 240 3.1. Computing not-via repairs in routing vector protocols 242 While this document focuses on link state routing protocols, it is 243 equally possible to compute not-via repairs in distance vector (e.g. 244 RIP) or path vector (e.g. BGP) routing protocols. This can be 245 achieved with very little protocol modification by advertising the 246 not-via address in the normal way, but ensuring that the information 247 about a not-via address Ps is not propagated through the node S. In 248 the case of link protection this simply means that the advertisement 249 from P to S is suppressed, with the result that S and all other nodes 250 compute a route to Ps which doesn't traverse S, as required. 252 In the case of node protection, where P is the protected node, and N 253 is some neighbor, the advertisement of Np must be suppressed not only 254 across the link N->P, but also across any link to P. The simplest way 255 of achieving this is for P itself to perform the suppression of any 256 address of the form Xp. 258 4. Operation of Repairs 260 This section explains the basic operation of the not-via repair of 261 node and link failure. 263 4.1. Node Failure 265 When router P fails (Figure 2) S encapsulates any packet that it 266 would send to B via P to Bp, and then sends the encapsulated packet 267 on the shortest path to Bp. S follows the same procedure for routers 268 A and C in Figure 2. The packet is decapsulated at the repair target 269 (A, B or C) and then forwarded normally to its destination. The 270 repair target can be determined as part of the normal SPF by 271 recording the "next-next-hop" for each destination in addition to the 272 normal next-hop. 274 Notice that with this technique only one level of encapsulation is 275 needed, and that it is possible to repair ANY failure regardless of 276 link metrics and any asymmetry that may be present in the network. 277 The only exception to this is where the failure was a single point of 278 failure that partitioned the network, in which case ANY repair is 279 clearly impossible. 281 4.2. Link Failure 283 The normal mode of operation of the network would be to assume router 284 failure. However, where some destinations are only reachable through 285 the failed router, it is desirable that an attempt be made to repair 286 to those destinations by assuming that only a link failure has 287 occurred. 289 To perform a link repair, S encapsulates to Ps (i.e. it instructs the 290 network to deliver the packet to P not-via S). All of the neighbors 291 of S will have calculated a path to Ps in case S itself had failed. 292 S could therefore give the packet to any of its neighbors (except, of 293 course, P). However, S should preferably send the encapsulated 294 packet on the shortest available path to P. This path is calculated 295 by running an SPF with the link SP failed. Note that this may again 296 be an incremental calculation, which can terminate when address Ps 297 has been reattached. 299 4.2.1. Loop Prevention Under Node Failure 301 It is necessary to consider the behavior of IPFRR solutions when a 302 link repair is attempted in the presence of node failure. In its 303 simplest form the not-via IPFRR solution prevents the formation of 304 loops forming as a result of mutual repair, by never providing a 305 repair path for a not-via address. The repair of packets with not- 306 via addresses is considered in more detail in Section 5.3. Referring 307 to Figure 2, if A was the neighbor of P that was on the link repair 308 path from S to P, and P itself had failed, the repaired packet from S 309 would arrive at A encapsulated to Ps. A would have detected that the 310 AP link had failed and would normally attempt to repair the packet. 311 However, no repair path is provided for any not-via address, and so A 312 would be forced to drop the packet, thus preventing the formation of 313 loop. 315 4.3. Multi-homed Prefixes 317 A multi-homed Prefix (MHP) is a prefix that is reachable via more 318 than one router in the network. Some of these may be repairable 319 using LFAs as described in [RFC5286]. Only those without such a 320 repair need be considered here. 322 When IPFRR router S (Figure 3) discovers that P has failed, it needs 323 to send packets addressed to the MHP X, which is normally reachable 324 through P, to an alternate router, which is still able to reach X. 326 X X X 327 | | | 328 | | | 329 | Sp |Pb | 330 Z...............S----------P----------B...............Y 331 Ps|Pc Bp 332 | 333 Cp| 334 C 336 Figure 3: Multi-homed Prefixes 338 S should choose the closest router that can reach X during the 339 failure as the alternate router. S determines which router to use as 340 the alternate while running the SPF with P failed. This is 341 accomplished by the normal process of re-attaching a leaf node to the 342 core topology (this is sometimes known as a "partial SPF"). 344 First, consider the case where the shortest alternate path to X is 345 via Z. S can reach Z without using the failed router P. However, S 346 cannot just send the packet towards Z, because the other routers in 347 the network will not be aware of the failure of P, and may loop the 348 packet back to S. S therefore encapsulates the packet to Z (using a 349 normal address for Z). When Z receives the encapsulated packet it 350 removes the encapsulation and forwards the packet to X. 352 Now consider the case where the shortest alternate path to X is via 353 Y, which S reaches via P and B. To reach Y, S must first repair the 354 packet to B using the normal not-via repair mechanism. To do this S 355 encapsulates the packet for X to Bp. When B receives the packet it 356 removes the encapsulation and discovers that the packet is intended 357 for MHP X. The situation now reverts to the previous case, in which 358 the shortest alternate path does not require traversal of the 359 failure. B therefore follows the algorithm above and encapsulates 360 the packet to Y (using a normal address for Y). Y removes the 361 encapsulation and forwards the packet to X. 363 It may be that the cost of reaching X using local delivery from the 364 alternate router (i.e. Z or Y) is greater than the cost of reaching 365 X via P. Under those circumstances, the alternate router would 366 normally forward to X via P, which would cause the IPFRR repair to 367 loop. To prevent the repair from looping the alternate router must 368 locally deliver a packet received via a repair encapsulation. This 369 may be specified by using a special address with the above semantics. 370 Note that only one such address is required per node. Notice that 371 using the not-via approach, only one level of encapsulation was 372 needed to repair MHPs to the alternate router. 374 4.4. Installation of Repair Paths 376 The following algorithm is used by node S (Figure 3) to pre- 377 calculate and install repair paths in the FIB, ready for immediate 378 use in the event of a failure. It is assumed that the not-via repair 379 paths have already been calculated as described above. 381 For each neighbor P, consider all destinations which are reachable 382 via P in the current topology:- 384 1. For all destinations with an ECMP or LFA repair (as described in 385 [RFC5286]) install that repair. 387 2. For each destination (DR) that remains, identify in the current 388 topology the next-next-hop (H) (i.e. the neighbor of P that P 389 will use to send the packet to DR). This can be determined 390 during the normal SPF run by recording the additional 391 information. If S has a path to the not-via address Hp (H not 392 via P), install a not-via repair to Hp for the destination DR. 394 3. Identify all remaining destinations (M) which can still be 395 reached when node P fails. These will be multi-homed prefixes 396 that are not repairable by LFA, for which the normal attachment 397 node is P, or a router for which P is a single point of failure, 398 and have an alternative attachment point that is reachable after 399 P has failed. One way of determining these destinations would be 400 to run an SPF rooted at S with node P removed, but an 401 implementation may record alternative attachment points during 402 the normal SPF run. In either case, the next best point of 403 attachment can also be determined for use in step (4) below. 405 4. For each multi-homed prefix (M) identified in step (3):- 407 A. Identify the new attachment node (as shown in Figure 3). 408 This may be:- 410 a. Y, where the next hop towards Y is P, or 412 b. Z, where the next hop towards Z is not P. 414 If the attachment node is Z, install the repair for M as a 415 tunnel to Z' (where Z' is the address of Z that is used to 416 force local forwarding). 418 B. For the subset of prefixes (M) that remain (having attachment 419 point Y), install the repair path previously installed for 420 destination Y. 422 For each destination (DS) that remains, install a not-via repair 423 to Ps (P not via S). Note, these are destinations for which node 424 P is a single point of failure, and can only be repaired by 425 assuming that the apparent failure of node P was simply a failure 426 of the S-P link. Note that, if available, a downstream path to P 427 may be used for such a repair. This cannot generate a persistent 428 loop in the event of the failure of node P, but if one neighbor 429 of P uses a not-via repair and another uses a downstream path, it 430 is possible for a packet sent on the downstream path to be 431 returned to the sending node inside a not-via encapsulation. 432 Since packets destined to not-via addresses are not repaired, the 433 packet will be dropped after executing a single turn loop. 435 5. Compound Failures 437 The following types of failures involve more than one component: 439 1. Shared Risk Link Groups 441 2. Local Area Networks 443 3. Multiple Independent Failures 445 The considerations that apply in each of the above situations are 446 described in the following sections. 448 5.1. Shared Risk Link Groups 450 A Shared Risk Link Group (SRLG) is a set of links whose failure can 451 be caused by a single action such as a conduit cut or line card 452 failure. When repairing the failure of a link that is a member of an 453 SRLG, it must be assumed that all the other links that are also 454 members of the SRLG have also failed. Consequently, any repair path 455 must be computed to avoid not just the adjacent link, but also all 456 the links which are members of the same SRLG. 458 In Figure 4 below, the links S-P and A-B are both members of SRLG 459 "a". The semantics of the not-via address Ps changes from simply "P 460 not-via the link S-P" to be "P not-via the link S-P or any other link 461 with which S-P shares an SRLG" In Figure 4 this is the links that are 462 members of SRLG "a". I.e. links S-P and A-B. Since the information 463 about SRLG membership of all links is available in the Link State 464 Database, all nodes computing routes to the not-via address Ps can 465 infer these semantics, and perform the computation by failing all the 466 links in the SRLG when running the iSPF. 468 Note that it is not necessary for S to consider repairs to any other 469 nodes attached to members of the SRLG (such as B). It is sufficient 470 for S to repair to the other end of the adjacent link (P in this 471 case). 473 a Ps 474 S----------P---------D 475 | | 476 | a | 477 A----------B 478 | | 479 | | 480 C----------E 482 Figure 4: Shared Risk Link Group 484 In some cases, it may be that the links comprising the SRLG occur in 485 series on the path from S to the destination D, as shown in Figure 5. 486 In this case, multiple consecutive repairs may be necessary. S will 487 first repair to Ps, then P will repair to Dp. In both cases, because 488 the links concerned are members of SRLG "a" the paths are computed to 489 avoid all members of SRLG "a". 491 a Ps a Dp 492 S----------P---------D 493 | | | 494 | a | | 495 A----------B | 496 | | | 497 | | | 498 C----------E---------F 500 Figure 5: Shared Risk Link Group members in series 502 While the use of multiple repairs in series introduces some 503 additional overhead, these semantics avoid the potential 504 combinatorial explosion of not-via addresses that could otherwise 505 occur. 507 Note that although multiple repairs are used, only a single level of 508 encapsulation is required. This is because the first repair packet 509 is decapsulated before the packet is re-encapsulated using the not- 510 via address corresponding to the far side of the next link which is a 511 member of the same SRLG. In some cases the de-capsulation and re- 512 encapsulation takes place (at least notionally) at a single node, 513 while in other cases, these functions may be performed by different 514 nodes. This scenario is illustrated in Figure 6 below. 516 a Ps a Dg 517 S----------P---------G--------D 518 | | | | 519 | a | | | 520 A----------B | | 521 | | | | 522 | | | | 523 C----------E---------F--------H 525 Figure 6: Shared Risk Link Group members in series 527 In this case, S first encapsulates to Ps, and node P decapsulates the 528 packet and forwards it "native" to G using its normal FIB entry for 529 destination D. G then repairs the packet to Dg. 531 It can be shown that such multiple repairs can never form a loop 532 because each repair causes the packet to move closer to its 533 destination. 535 It is often the case that a single link may be a member of multiple 536 SRLGs, and those SRLGs may not be isomorphic. This is illustrated in 537 Figure 7 below. 539 ab Ps a Dg 540 S----------P---------G--------D 541 | | | | 542 | a | | | 543 A----------B | | 544 | | | | 545 | b | | b | 546 C----------E---------F--------H 547 | | 548 | | 549 J----------K 551 Figure 7: Multiple Shared Risk Link Groups 553 The link SP is a member of SRLGs "a" and "b". When a failure of the 554 link SP is detected, it must be assumed that BOTH SRLGs have failed. 555 Therefore the not-via path to Ps must be computed by failing all 556 links which are members of SRLG "a" or SRLG "b". I.e. the semantics 557 of Ps is now "P not-via any links which are members of any of the 558 SRLGs of which link SP is a member". This is illustrated in Figure 8 559 below. 561 ab Ps a Dg 562 S----/-----P---------G---/----D 563 | | | | 564 | a | | | 565 A----/-----B | | 566 | | | | 567 | b | | b | 568 C----/-----E---------F---/----H 569 | | 570 | | 571 J----------K 573 Figure 8: Topology used for repair computation for link S-P 575 In this case, the repair path to Ps will be S-A-C-J-K-E-B-P. It may 576 appear that there is no path to D because GD is a member of SRLG "a" 577 and FH is a member of SRLG "b". This is true if BOTH SRLGs "a" and 578 "b" have in fact failed. But that would be an instance of multiple 579 uncorrelated failures which are out of scope for this design. In 580 practice it is likely that there is only a single failure, i.e. 581 either SRLG "a" or SRLG "b" has failed, but not both. These two 582 possibilities are indistinguishable from the point of view of the 583 repairing router S and so it must repair on the assumption that both 584 are unavailable. However, each link repair is considered 585 independently. The repair to Ps delivers the packet to P which then 586 forwards the packet to G. When the packet arrives at G, if SRLG "a" 587 has failed it will be repaired around the path G-F-H-D. This is 588 illustrated in Figure 9 below. If, on the other hand, SRLG "b" has 589 failed, link GD will still be available. In this case the packet 590 will be delivered as normal across the link GD. 592 ab Ps a Dg 593 S----/-----P---------G---/----D 594 | | | | 595 | a | | | 596 A----/-----B | | 597 | | | | 598 | b | | b | 599 C----------E---------F--------H 600 | | 601 | | 602 J----------K 604 Figure 9: Topology used for repair computation for link G-D 606 A repair strategy that assumes the worst-case failure for each link 607 can often result in longer repair paths than necessary. In cases 608 where only a single link fails, rather than the full SRLG, this 609 strategy may occasionally fail to identify a repair even though a 610 viable repair path exists in the network. The use of sub-optimal 611 repair paths is an inevitable consequence of this compromise 612 approach. The failure to identify any repair is a serious 613 deficiency, but is a rare occurrence in a robustly designed network. 614 This problem can be addressed by:- 616 1. Reporting that the link in question is irreparable, so that the 617 network designer can take appropriate action. 619 2. Modifying the design of the network to avoid this possibility. 621 3. Using some form of SRLG diagnostic (for example, by running BFD 622 over alternate repair paths) to determine which SRLG member(s) 623 has actually failed and using this information to select an 624 appropriate pre-computed repair path. However, aside from the 625 complexity of performing the diagnostics, this requires multiple 626 not-via addresses per interface, which has poor scaling 627 properties. 629 4. Using the mechanism described in Section 5.3 631 5.1.1. Use of LFAs with SRLGs 633 Section 4.1 above describes the repair of links which are members of 634 one or more SRLGs. LFAs can be used for the repair of such links 635 provided that any other link with which S-P shares an SRLG is avoided 636 when computing the LFA. This is described for the simple case of 637 "local-SRLGs" in [RFC5286]. 639 5.2. Local Area Networks 641 LANs are a special type of SRLG and are solved using the SRLG 642 mechanisms outlined above. With all SRLGs there is a trade-off 643 between the sophistication of the fault detection and the size of the 644 SRLG. Protecting against link failure of the LAN link(s) is 645 relatively straightforward, but as with all fast reroute mechanisms, 646 the problem becomes more complex when it is desired to protect 647 against the possibility of failure of the nodes attached to the LAN 648 as well as the LAN itself. 650 +--------------Q------C 651 | 652 | 653 | 654 A--------S-------(N)-------------P------B 655 | 656 | 657 | 658 +--------------R------D 660 Figure 10: Local Area Networks 662 Consider the LAN shown in Figure 10. For connectivity purposes, we 663 consider that the LAN is represented by the pseudonode (N). To 664 provide IPFRR protection, S must run a connectivity check to each of 665 its protected LAN adjacencies P, Q, and R, using, for example BFD 666 [RFC5880]. 668 When S discovers that it has lost connectivity to P, it is unsure 669 whether the failure is: 671 o its own interface to the LAN, 673 o the LAN itself, 675 o the LAN interface of P, 677 o the node P. 679 5.2.1. Simple LAN Repair 681 A simple approach to LAN repair is to consider the LAN and all of its 682 connected routers as a single SRLG. Thus, the address P not via the 683 LAN (Pl) would require P to be reached not-via any router connected 684 to the LAN. This is shown in Figure 11. 686 Ql Cl 687 +-------------Q--------C 688 | Qc 689 | 690 As Sl | Pl Bl 691 A--------S-------(N)------------P--------B 692 Sa | Pb 693 | 694 | Rl Dl 695 +-------------R--------D 696 Rd 698 Figure 11: Local Area Networks - LAN SRLG 700 In this case, when S detected that P had failed it would send traffic 701 reached via P and B to B not-via the LAN or any router attached to 702 the LAN (i.e. to Bl). Any destination only reachable through P would 703 be addressed to P not-via the LAN or any router attached to the LAN 704 (except of course P). 706 Whilst this approach is simple, it assumes that a large portion of 707 the network adjacent to the failure has also failed. This will 708 result in the use of sub-optimal repair paths and in some cases the 709 inability to identify a viable repair. 711 5.2.2. LAN Component Repair 713 In this approach, possible failures are considered at a finer 714 granularity, but without the use of diagnostics to identify the 715 specific component that has failed. Because S is unable to diagnose 716 the failure it must repair traffic sent through P and B, to B not- 717 via P,N (i.e. not via P and not via N), on the conservative 718 assumption that both the entire LAN and P have failed. Destinations 719 for which P is a single point of failure must as usual be sent to P 720 using an address that avoids the interface by which P is reached from 721 S, i.e. to P not-via N. Similarly for routers Q and R. 723 Notice that each router that is connected to a LAN must, as usual, 724 advertise one not-via address for each neighbor. In addition, each 725 router on the LAN must advertise an extra address not via the 726 pseudonode (N). 728 Notice also that each neighbor of a router connected to a LAN must 729 advertise two not-via addresses, the usual one not via the neighbor 730 and an additional one, not via either the neighbor or the pseudonode. 731 The required set of LAN address assignments is shown in Figure 12 732 below. Each router on the LAN, and each of its neighbors, is 733 advertising exactly one address more than it would otherwise have 734 advertised if this degree of connectivity had been achieved using 735 point-to-point links. 737 Qs Qp Qc Cqn 738 +--------------Q---------C 739 | Qr Qn Cq 740 | 741 Asn Sa Sp Sq | Ps Pq Pb Bpn 742 A--------S-------(N)-------------P---------B 743 As Sr Sn | Pr Pn Bp 744 | 745 | Rs Rp Pd Drn 746 +--------------R---------D 747 Rq Rn Dr 749 Figure 12: Local Area Networks 751 5.2.3. LAN Repair Using Diagnostics 753 A more specific LAN repair can be undertaken by using diagnostics. 754 In order to explicitly diagnose the failed network component, S 755 correlates the connectivity reports from P and one or more of the 756 other routers on the LAN, in this case, Q and R. If it lost 757 connectivity to P alone, it could deduce that the LAN was still 758 functioning and that the fault lay with either P, or the interface 759 connecting P to the LAN. It would then repair to B not via P (and P 760 not-via N for destinations for which P is a single point of failure) 761 in the usual way. If S lost connectivity to more than one router on 762 the LAN, it could conclude that the fault lay only with the LAN, and 763 could repair to P, Q and R not-via N, again in the usual way. 765 5.3. Multiple Independent Failures 767 IPFRR repair of multiple simultaneous failures which are not members 768 of a known SRLG is complicated by the problem that the use of 769 multiple concurrent repairs may result in looping repair paths. As 770 described in Section 4.2.1, the simplest method of preventing such 771 loops, is to ensure that packets addressed to a not-via address are 772 not repaired but instead are dropped. It is possible that a network 773 may experience multiple simultaneous failures. This may be due to 774 simple statistical effects, but the more likely cause is 775 unanticipated SRLGs. When multiple failures which are not part of an 776 anticipated group are detected, repairs are abandoned and the network 777 reverts to normal convergence. Although safe, this approach is 778 somewhat draconian, since there are many circumstances were multiple 779 repairs do not induce loops. 781 This section describes the properties of multiple unrelated failures 782 and proposes some methods that may be used to address this problem. 784 5.3.1. Looping Repairs 786 Let us assume that the repair mechanism is based on solely on not-via 787 repairs. LFA or downstream routes may be incorporated, and will be 788 dealt with later. 790 A------//------B------------D 791 / \ 792 / \ 793 F G 794 \ / 795 \ / 796 X------//------Y 798 Figure 13: The General Case of Multiple Failures 800 The essential case is as illustrated in Figure 13. Note that 801 depending on the repair case under consideration, there may be paths 802 present in Figure 13, that are in addition to those shown in the 803 figure. For example there may be paths between A and B, and/or 804 between X and Y. These paths are omitted for graphical clarity. 806 There are three cases to consider: 808 1) Consider the general case of a pair of protected links A-B and 809 X-Y as shown in the network fragment shown Figure 13. If the 810 repair path for A-B does not traverse X-Y and the repair path for 811 X-Y does not traverse A-B, this case is completely safe and will 812 not cause looping or packet loss. 814 A more common variation of this case is shown in Figure 14, which 815 shows two failures in different parts of the network in which a 816 packet from A to D traverses two concatenated repairs. 818 A------//------B------------X------//------Y------D 819 | | | | 820 | | | | 821 M--------------+ N--------------+ 823 Figure 14: Concatenated Repairs 825 2) In Figure 13, the repair for A-B traverses X-Y, but the repair 826 for X-Y does not traverse A-B. This case occurs when the not-via 827 path from A to B traverses link X-Y, but the not-via path from X 828 to Y traverses some path not shown in Figure 13. Without the 829 multi-failure mechanism described in this section the repaired 830 packet for A-B would be dropped when it reached X-Y, since the 831 repair of repaired packets would be forbidden. However, if this 832 packet were allowed to be repaired, the path to D would be 833 complete and no harm would be done, although two levels of 834 encapsulation would be required. 836 3) The repair for A-B traverses X-Y AND the repair for X-Y 837 traverses A-B. In this case unrestricted repair would result in 838 looping packets and increasing levels of encapsulation. 840 The challenge in applying IPFRR to a network that is undergoing 841 multiple failures is, therefore, to identify which of these cases 842 exist in the network and react accordingly. 844 5.3.2. Outline Solution 846 When A is computing the not-via repair path for A-B (i.e. the path 847 for packets addressed to Ba, read as "B not-via A") it is aware of 848 the list of nodes which this path traverses. This can be recorded by 849 a simple addition to the SPF process, and the not-via addresses 850 associated with each forward link can be determined. If the path 851 were A, F, X, Y, G, B, (Figure 13) the list of not-via addresses 852 would be: Fa, Xf, Yx, Gy, Bg. Under standard not-via operation, A 853 would populate its FIB such that all normal addresses normally 854 reachable via A-B would be encapsulated to Ba when A-B fails, but 855 traffic addressed to any not-via address arriving at A would be 856 dropped. The new procedure modifies this such that any traffic for a 857 not-via address normally reachable over A-B is also encapsulated to 858 Ba unless the not-via address is one of those previously identified 859 as being on the path to Ba, for example Yx, in which case the packet 860 is dropped. 862 The above procedure allows cases 1 and 2 above to be repaired, while 863 preventing the loop which would result from case 3. 865 Note that this is accomplished by pre-computing the required FIB 866 entries, and does not require any detailed packet inspection. The 867 same result could be achieved by checking for multiple levels of 868 encapsulation and dropping any attempt to triple encapsulate. 869 However, this would require more detailed inspection of the packet, 870 and causes difficulties when more than 2 "simultaneous" failures are 871 contemplated. 873 So far we have permitted benign repairs to coexist, albeit sometimes 874 requiring multiple encapsulation. Note that in many cases there will 875 be no performance impact since unless both failures are on the same 876 node, the two encapsulations or two decapsulations will be performed 877 at different nodes. There is however the issue of the MTU impact of 878 multiple encapsulations. 880 In the following sub-section we consider the various strategies that 881 may be applied to case 3 - mutual repairs that would loop. 883 5.3.3. Looping Repairs 885 In case 3, the simplest approach is to simply not install repairs for 886 repair paths that might loop. In this case, although the potentially 887 looping traffic is dropped, the traffic is not repaired. If we 888 assume that a hold-down is applied before reconvergence in case the 889 link failure was just a short glitch, and if a loop free convergence 890 mechanism further delays convergence, then the traffic will be 891 dropped for an extended period. In these circumstances it would be 892 better to "abandon all hope" (AAH) [I-D.ietf-rtgwg-ordered-fib] 893 (Appendix A) and immediately invoke normal re-convergence. 895 Note that it is not sufficient to expedite the issuance of an LSP 896 reporting the failure, since this may be treated as a permitted 897 simultaneous failure by the ordered FIB (oFIB) algorithm 898 [I-D.ietf-rtgwg-ordered-fib]. It is therefore necessary to 899 explicitly trigger an oFIB AAH. 901 5.3.3.1. Dropping Looping Packets 903 One approach to case 3 is to allow the repair, and to experimentally 904 discover the incompatibility of the repairs if and when they occur. 905 With this method we permit the repair in case 3 and trigger AAH when 906 a packet drop count on the not-via address has been incremented. 907 Alternatively, it is possible to wait until the LSP describing the 908 change is issued normally (i.e. when X announces the failure of X-Y). 909 When the repairing node A, which has precomputed that X-Y failures 910 are mutually incompatible with its own repairs receives this LSP it 911 can then issue the AAH. This has the disadvantage that it does not 912 overcome the hold-down delay, but it requires no "data-driven" 913 operation, and it still has the required effect of abandoning the 914 oFIB which is probably the longer of the delays (although with 915 signalled oFIB this should be sub-second). 917 Whilst both of the experimental approaches described above are 918 feasible, they tend to induce AAH in the presence of otherwise 919 feasible repairs, and they are contrary to the philosophy of repair 920 pre-determination that has been applied to existing IPFRR solutions. 922 5.3.3.2. Computing non-looping Repairs of Repairs 924 An alternative approach to simply dropping the looping packets, or to 925 detecting the loop after it has occurred, is to use secondary SRLGs. 926 With a link state routing protocol it is possible to precompute the 927 incompatibility of the repairs in advance and to compute an 928 alternative SRLG repair path. Although this does considerably 929 increase the computational complexity it may be possible to compute 930 repair paths that avoid the need to simply drop the offending 931 packets. 933 This approach requires us to identify the mutually incompatible 934 failures, and advertise them as "secondary SRLGs". When computing 935 the repair paths for the affected not-via addresses these links are 936 simultaneously failed. Note that the assumed simultaneous failure 937 and resulting repair path only applies to the repair path computed 938 for the conflicting not-via addresses, and is not used for normal 939 addresses. This implies that although there will be a longer repair 940 path when there is more than one failure, if there is a single 941 failure the repair path length will be "normal". 943 Ideally we would wish to only invoke secondary SRLG computation when 944 we are sure that the repair paths are mutually incompatible. 945 Consider the case of node A in Figure 13. A first identifies that 946 the repair path for A-B is via F-X-Y-G-B. It then explores this path 947 determining the repair path for each link in the path. Thus, for 948 example, it performs a check at X by running an SPF rooted at X with 949 the X-Y link removed to determine whether A-B is indeed on X's repair 950 path for packets addressed to Yx. 952 Some optimizations are possible in this calculation, which appears at 953 first sight to be order hk (where h is the average hop length of 954 repair paths and k is the average number of neighbours of a router). 955 When A is computing its set of repair paths, it does so for all its k 956 neighbours. In each case it identifies a list of node pairs 957 traversed by each repair. These lists may often have one or more 958 node pairs in common, so the actual number of link failures which 959 require investigation is the union of these sets. It is then 960 necessary to run an SPF rooted at the first node of each pair (the 961 first node because the pairings are ordered representing the 962 direction of the path), with the link to the second node removed. 963 This SPF, while not an incremental, can be terminated as soon as the 964 not-via address is reached. For example, when running the SPF rooted 965 at X, with the link X-Y removed, the SPF can be terminated when Yx is 966 reached. Once the path has been found, the path is checked to 967 determine if it traverses any of A's links in the direction away from 968 A. Note that, because the node pair XY may exist in the list for more 969 than one of A's links (i.e. it lies on more than one repair path), it 970 is necessary to identify the correct list, and hence link which has a 971 mutually looping repair path. That link of A is then advertised by A 972 as a secondary SRLG paired with the link X-Y. Also note that X will 973 be running this algorithm as well, and will identify that XY is 974 paired with A-B and so advertise it. This could perhaps be used as a 975 further check. 977 The ordering of the pairs in the lists is important. i.e. X-Y and 978 Y-X are dealt with separately. If and only if the repairs are 979 mutually incompatible, we need to advertise the pair of links as a 980 secondary SRLG, and then ALL nodes compute repair paths around both 981 failures using an additional not-via address with the semantics not- 982 via A-B AND not-via X-Y. 984 A further possibility is that because we are going to the trouble of 985 advertising these SRLG sets, we could also advertise the new repair 986 path and only get the nodes on that path to perform the necessary 987 computation. Note also that once we have reached Q space with 988 respect to the two failures we need no longer continue the 989 computation, so we only need to notify the nodes on the path that are 990 not in Q-space. 992 One cause of mutually looping repair paths is the existence of nodes 993 with only two links, or sections of the network which are only bi- 994 connected. In these cases, repair is clearly impossible - the 995 failure of both links partitions the network. It would be 996 advantageous to be able to identify these cases, and inhibit the 997 fruitless advertisement of the secondary SRLG information. This 998 could be achieved by the node detecting the requirement for a 999 secondary SRLG, first running the not-via computation with both links 1000 removed. If this does not result in a path, it is clear that the 1001 network would be partitioned by such a failure, and so no 1002 advertisement is required. 1004 5.3.4. Mixing LFAs and Not-via 1006 So far in this section we have assumed that all repairs use not-via 1007 tunnels. However, in practise we may wish to use LFAs or downstream 1008 routes where available. This complicates the issue, because their 1009 use results in packets which are being repaired, but NOT addressed to 1010 not-via addresses. If BOTH links are using downstream routes there 1011 is no possibility of looping, since it is impossible to have a pair 1012 of nodes which are both downstream of each other [RFC5286]. 1014 Loops can however occur when LFAs are used. An obvious example is 1015 the well known node repair problem with LFAs [RFC5286]. If one link 1016 is using a downstream route, while the other is using a not-via 1017 tunnel, the potential mechanism described above would work provided 1018 it were possible to determine the nodes on the path of the downstream 1019 route. Some methods of computing downstream routes do not provide 1020 this path information. If the path information is however available, 1021 the link using a downstream route will have a discard FIB entry for 1022 the not-via address of the other link. The consequence is that 1023 potentially looping packets will be discarded when they attempt to 1024 cross this link. 1026 In the case where the mutual repairs are both using not-via repairs, 1027 the loop will be broken when the packet arrives at the second 1028 failure. However packets are unconditionally repaired by means of a 1029 downstream routes, and thus when the mutual pair consists of a 1030 downstream route and a not-via repair, the looping packet will only 1031 be dropped when it gets back to the first failure. i.e. it will 1032 execute a single turn of the loop before being dropped. 1034 There is a further complication with downstream routes, since 1035 although the path may be computed to the far side of the failure, the 1036 packet may "peel off" to its destination before reaching the far side 1037 of the failure. In this case it may traverse some other link which 1038 has failed and was not accounted for on the computed path. If the 1039 A-B repair (Figure 1) is a downstream route and the X-Y repair is a 1040 not-via repair, we can have the situation where the X-Y repair 1041 packets encapsulated to Yx follow a path which attempts to traverse 1042 A-B. If the A-B repair path for "normal" addresses is a downstream 1043 route, it cannot be assumed that the repair path for packets 1044 addressed to Yx can be sent to the same neighbour. This is because 1045 the validity of a downstream route must be ascertained in the 1046 topology represented by Yx, i.e. that with the link X-Y failed. This 1047 is not the same topology that was used for the normal downstream 1048 calculation, and use of the normal downstream route for the 1049 encapsulated packets may result in an undetected loop. If it is 1050 computationally feasible to check the downstream route in this 1051 topology (i.e. for any not-via address Qp which traverses A-B we must 1052 perform the downstream calculation for that not-via address in the 1053 topology with link Q-P failed.), then the downstream repair for Yx 1054 can safely be used. These packets cannot re-visit X-Y, since by 1055 definition they will avoid that link. Alternatively, the packet 1056 could be always repaired in a not-via tunnel. i.e. even though the 1057 normal repair for traffic traversing A-B would be to use a downstream 1058 route, we could insist that such traffic addressed to a not-via 1059 address must use a tunnel to Ba. Such a tunnel would only be 1060 installed for an address Qp if it were established that it did not 1061 traverse Q-P (using the rules described above). 1063 6. Optimizing not-via computations using LFAs 1065 If repairing node S has an LFA to the repair endpoint it is not 1066 necessary for any router to perform the incremental SPF with the link 1067 SP removed in order to compute the route to the not-via address Ps. 1068 This is because the correct routes will already have been computed as 1069 a result of the SPF on the base topology. Node S can signal this 1070 condition to all other routers by including a bit in its LSP or LSA 1071 associated with each LFA protected link. Routers computing not-via 1072 routes can then omit the running of the iSPF for links with this bit 1073 set. 1075 When running the iSPF for a particular link AB, the calculating 1076 router first checks whether the link AB is present in the existing 1077 SPT. If the link is not present in the SPT, no further work is 1078 required. This check is a normal part of the iSPF computation. 1080 If the link is present in the SPT, this optimization introduces a 1081 further check to determine whether the link is marked as protected by 1082 an LFA in the direction in which the link appears in the SPT. If so 1083 the iSPF need not be performed. For example, if the link appears in 1084 the SPT in the direction A->B and A has indicated that the link AB is 1085 protected by an LFA no further action is required for this link. 1087 If the receipt of this information is delayed, the correct operation 1088 of the protocol is not compromised provided that the necessity to 1089 perform a not-via computation is re-evaluated whenever new 1090 information arrives. 1092 This optimization is not particularly beneficial to nodes close to 1093 the repair since, as has been observed above, the computation for 1094 nodes on the LFA path is trivial. However, for nodes upstream of the 1095 link SP for which S-P is in the path to P, there is a significant 1096 reduction in the computation required. 1098 7. Multicast 1100 Multicast traffic can be repaired in a similar way to unicast. The 1101 multicast forwarder is able to use the not-via address to which the 1102 multicast packet was addressed as an indication of the expected 1103 receive interface and hence to correctly run the required RPF check. 1105 In some cases, all the destinations, including the repair endpoint, 1106 are repairable by an LFA. In this case, all unicast traffic may be 1107 repaired without encapsulation. Multicast traffic still requires 1108 encapsulation, but for the nodes on the LFA repair path the 1109 computation of the not-via forwarding entry is unnecessary since, by 1110 definition, their normal path to the repair endpoint is not via the 1111 failure. 1113 A more complete description of multicast operation is for further 1114 study. 1116 8. Fast Reroute in an MPLS LDP Network. 1118 Not-via addresses are IP addresses and LDP [RFC5036] will distribute 1119 labels for them in the usual way. The not-via repair mechanism may 1120 therefore be used to provide fast re-route in an MPLS network by 1121 first pushing the label which the repair endpoint uses to forward the 1122 packet, and then pushing the label corresponding to the not-via 1123 address needed to effect the repair. Referring once again to 1124 Figure 1, if S has a packet destined for D that it must reach via P 1125 and B, S first pushes B's label for D. S then pushes the label that 1126 its next hop to Bp needs to reach Bp. 1128 Note that in an MPLS LDP network it is necessary for S to have the 1129 repair endpoint's label for the destination. When S is effecting a 1130 link repair it already has this. In the case of a node repair, S 1131 either needs to set up a directed LDP session with each of its 1132 neighbor's neighbors, or it needs to use the next-next hop label 1133 distribution mechanism proposed in [I-D.shen-mpls-ldp-nnhop-label]. 1135 9. Encapsulation 1137 Any IETF specified IP in IP encapsulation may be used to carry a not- 1138 via repair. IP in IP [RFC2003], GRE [RFC1701] and L2TPv3 [RFC3931], 1139 all have the necessary and sufficient properties. The requirement is 1140 that both the encapsulating router and the router to which the 1141 encapsulated packet is addressed have a common ability to process the 1142 chosen encapsulation type. When an MPLS LDP network is being 1143 protected, the encapsulation would normally be an additional MPLS 1144 label. In an MPLS enabled IP network an MPLS label may be used in 1145 place of an IP in IP encapsulation in the case above. 1147 10. Routing Extensions 1149 IPFRR requires IGP extensions. Each IPFRR router that is directly 1150 connected to a protected network component must advertise a not-via 1151 address for that component. This must be advertised in such a way 1152 that the association between the protected component (link, router or 1153 SRLG) and the not-via address can be determined by the other routers 1154 in the network. 1156 It is necessary that not-via capable routers advertise in the IGP 1157 that they will calculate not-via routes. 1159 It is necessary for routers to advertise the type of encapsulation 1160 that they support (MPLS, GRE, L2TPv3 etc). However, the deployment 1161 of mixed IP encapsulation types within a network is discouraged. 1163 11. Incremental Deployment 1165 Incremental deployment is supported by excluding routers that are not 1166 calculating not-via routes (as indicated by their capability 1167 information flooded with their link state information) from the base 1168 topology used for the computation of repair paths. In that way 1169 repairs may be steered around islands of routers that are not IPFRR 1170 capable. Routers that are protecting a network component need to 1171 have the capability to encapsulate and decapsulate packets. However, 1172 routers that are on the repair path only need to be capable of 1173 calculating not-via paths and including the not-via addresses in 1174 their FIB i.e. these routers do not need any changes to their 1175 forwarding mechanism. 1177 12. IANA Considerations 1179 There are no IANA considerations that arise from this draft. 1181 13. Security Considerations 1183 The repair endpoints present vulnerability in that they might be used 1184 as a method of disguising the delivery of a packet to a point in the 1185 network. The primary method of protection should be through the use 1186 of a private address space for the not-via addresses. These 1187 addresses must not be advertised outside the area, and should be 1188 filtered at the network entry points. In addition, a mechanism might 1189 be developed that allowed the use of the mild security available 1190 through the use of a key [RFC1701] [RFC3931]. With the deployment of 1191 such mechanisms, the repair endpoints would not increase the security 1192 risk beyond that of existing IP tunnel mechanisms. An attacker may 1193 attempt to overload a router by addressing an excessive traffic load 1194 to the de-capsulation endpoint. Typically, routers take a 50% 1195 performance penalty in decapsulating a packet. The attacker could 1196 not be certain that the router would be impacted, and the extremely 1197 high volume of traffic needed, would easily be detected as an 1198 anomaly. If an attacker were able to influence the availability of a 1199 link, they could cause the network to invoke the not-via repair 1200 mechanism. A network protected by not-via IPFRR is less vulnerable 1201 to such an attack than a network that undertook a full convergence in 1202 response to a link up/down event. 1204 14. Acknowledgements 1206 The authors would like to acknowledge contributions made by Alia 1207 Atlas and John Harper. 1209 15. Informative References 1211 [I-D.ietf-rtgwg-ordered-fib] 1212 Shand, M., Bryant, S., Previdi, S., and C. Filsfils, 1213 "Loop-free convergence using oFIB", 1214 draft-ietf-rtgwg-ordered-fib-05 (work in progress), 1215 April 2011. 1217 [I-D.shen-mpls-ldp-nnhop-label] 1218 Shen, N., "Discovering LDP Next-Nexthop Labels", 1219 draft-shen-mpls-ldp-nnhop-label-02 (work in progress), 1220 May 2005. 1222 [ISPF] McQuillan, J., Richer, I., and E. Rosen, "ARPANET Routing 1223 Algorithm Improvements"", BBN Technical Report 3803, 1978. 1225 [RFC1701] Hanks, S., Li, T., Farinacci, D., and P. Traina, "Generic 1226 Routing Encapsulation (GRE)", RFC 1701, October 1994. 1228 [RFC2003] Perkins, C., "IP Encapsulation within IP", RFC 2003, 1229 October 1996. 1231 [RFC3931] Lau, J., Townsley, M., and I. Goyret, "Layer Two Tunneling 1232 Protocol - Version 3 (L2TPv3)", RFC 3931, March 2005. 1234 [RFC5036] Andersson, L., Minei, I., and B. Thomas, "LDP 1235 Specification", RFC 5036, October 2007. 1237 [RFC5286] Atlas, A. and A. Zinin, "Basic Specification for IP Fast 1238 Reroute: Loop-Free Alternates", RFC 5286, September 2008. 1240 [RFC5714] Shand, M. and S. Bryant, "IP Fast Reroute Framework", 1241 RFC 5714, January 2010. 1243 [RFC5880] Katz, D. and D. Ward, "Bidirectional Forwarding Detection 1244 (BFD)", RFC 5880, June 2010. 1246 Authors' Addresses 1248 Stewart Bryant 1249 Cisco Systems 1250 250, Longwater Avenue. 1251 Reading, Berks RG2 6GB 1252 UK 1254 Email: stbryant@cisco.com 1256 Stefano Previdi 1257 Cisco Systems 1258 Via Del Serafico, 200 1259 00142 Rome, 1260 Italy 1262 Email: sprevidi@cisco.com 1264 Mike Shand 1265 Individual Contributor 1267 Email: imc.shand@googlemail.com