idnits 2.17.1 draft-ietf-rtgwg-ipfrr-notvia-addresses-07.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 490 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 (April 20, 2011) is 4754 days in the past. Is this intentional? Checking references for intended status: Experimental ---------------------------------------------------------------------------- == Outdated reference: A later version (-12) exists of draft-ietf-rtgwg-ordered-fib-05 Summary: 0 errors (**), 0 flaws (~~), 3 warnings (==), 2 comments (--).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 M. Shand 3 Internet-Draft S. Bryant 4 Intended status: Experimental S. Previdi 5 Expires: October 22, 2011 Cisco Systems 6 April 20, 2011 8 IP Fast Reroute Using Not-via Addresses 9 draft-ietf-rtgwg-ipfrr-notvia-addresses-07 11 Abstract 13 This draft describes a mechanism that provides fast reroute in an IP 14 network through encapsulation to "not-via" addresses. A single level 15 of encapsulation is used. The mechanism protects unicast, multicast 16 and LDP traffic against link, router and shared risk group failure, 17 regardless of network topology and metrics. 19 Status of this Memo 21 This Internet-Draft is submitted in full conformance with the 22 provisions of BCP 78 and BCP 79. 24 Internet-Drafts are working documents of the Internet Engineering 25 Task Force (IETF). Note that other groups may also distribute 26 working documents as Internet-Drafts. The list of current Internet- 27 Drafts is at http://datatracker.ietf.org/drafts/current/. 29 Internet-Drafts are draft documents valid for a maximum of six months 30 and may be updated, replaced, or obsoleted by other documents at any 31 time. It is inappropriate to use Internet-Drafts as reference 32 material or to cite them other than as "work in progress." 34 This Internet-Draft will expire on October 22, 2011. 36 Copyright Notice 38 Copyright (c) 2011 IETF Trust and the persons identified as the 39 document authors. All rights reserved. 41 This document is subject to BCP 78 and the IETF Trust's Legal 42 Provisions Relating to IETF Documents 43 (http://trustee.ietf.org/license-info) in effect on the date of 44 publication of this document. Please review these documents 45 carefully, as they describe your rights and restrictions with respect 46 to this document. Code Components extracted from this document must 47 include Simplified BSD License text as described in Section 4.e of 48 the Trust Legal Provisions and are provided without warranty as 49 described in the Simplified BSD License. 51 Table of Contents 53 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 54 2. Overview of Not-via Repairs . . . . . . . . . . . . . . . . . 3 55 2.1. Use of Equal Cost Multi-Path . . . . . . . . . . . . . . . 4 56 2.2. Use of LFA repairs . . . . . . . . . . . . . . . . . . . . 4 57 3. Not-via Repair Path Computation . . . . . . . . . . . . . . . 5 58 3.1. Computing not-via repairs in routing vector protocols . . 6 59 4. Operation of Repairs . . . . . . . . . . . . . . . . . . . . . 6 60 4.1. Node Failure . . . . . . . . . . . . . . . . . . . . . . . 6 61 4.2. Link Failure . . . . . . . . . . . . . . . . . . . . . . . 7 62 4.2.1. Loop Prevention Under Node Failure . . . . . . . . . . 7 63 4.3. Multi-homed Prefixes . . . . . . . . . . . . . . . . . . . 7 64 4.4. Installation of Repair Paths . . . . . . . . . . . . . . . 9 65 5. Compound Failures . . . . . . . . . . . . . . . . . . . . . . 10 66 5.1. Shared Risk Link Groups . . . . . . . . . . . . . . . . . 10 67 5.1.1. Use of LFAs with SRLGs . . . . . . . . . . . . . . . . 14 68 5.2. Local Area Networks . . . . . . . . . . . . . . . . . . . 14 69 5.2.1. Simple LAN Repair . . . . . . . . . . . . . . . . . . 15 70 5.2.2. LAN Component Repair . . . . . . . . . . . . . . . . . 16 71 5.2.3. LAN Repair Using Diagnostics . . . . . . . . . . . . . 17 72 5.3. Multiple Independent Failures . . . . . . . . . . . . . . 17 73 5.3.1. Looping Repairs . . . . . . . . . . . . . . . . . . . 18 74 5.3.2. Outline Solution . . . . . . . . . . . . . . . . . . . 19 75 5.3.3. Looping Repairs . . . . . . . . . . . . . . . . . . . 20 76 5.3.3.1. Dropping Looping Packets . . . . . . . . . . . . . 20 77 5.3.3.2. Computing non-looping Repairs of Repairs . . . . . 21 78 5.3.4. Mixing LFAs and Not-via . . . . . . . . . . . . . . . 23 79 6. Optimizing not-via computations using LFAs . . . . . . . . . . 24 80 7. Multicast . . . . . . . . . . . . . . . . . . . . . . . . . . 25 81 8. Fast Reroute in an MPLS LDP Network. . . . . . . . . . . . . . 25 82 9. Encapsulation . . . . . . . . . . . . . . . . . . . . . . . . 25 83 10. Routing Extensions . . . . . . . . . . . . . . . . . . . . . . 26 84 11. Incremental Deployment . . . . . . . . . . . . . . . . . . . . 26 85 12. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 26 86 13. Security Considerations . . . . . . . . . . . . . . . . . . . 26 87 14. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 27 88 15. Informative References . . . . . . . . . . . . . . . . . . . . 27 89 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 28 91 1. Introduction 93 When a link or a router fails, only the neighbors of the failure are 94 initially aware that the failure has occurred. In a network 95 operating IP fast reroute [RFC5714], the routers that are the 96 neighbors of the failure repair the failure. These repairing routers 97 have to steer packets to their destinations despite the fact that 98 most other routers in the network are unaware of the nature and 99 location of the failure. 101 A common limitation in most IPFRR mechanisms is an inability to 102 indicate the identity of the failure and to explicitly steer the 103 repaired packet round the failure. The extent to which this 104 limitation affects the repair coverage is topology dependent. The 105 mechanism proposed here is to encapsulate the packet to an address 106 that explicitly identifies the network component that the repair must 107 avoid. This produces a repair mechanism, which, provided the network 108 is not partitioned by the failure, will always achieve a repair. 110 2. Overview of Not-via Repairs 112 This section provides a brief overview of the not-via method of 113 IPFRR. 115 A 116 | Bp is the address to use to get 117 | a packet to B not-via P 118 | 119 S----------P----------B. . . . . . . . . .D 120 \ | Bp^ 121 \ | | 122 \ | | 123 \ C | 124 \ | 125 ----------------+ 126 Repair to Bp 128 Figure 1: Not-via repair of router failure 130 Assume that S has a packet for some destination D that it would 131 normally send via P and B, and that S suspects that P has failed. S 132 encapsulates the packet to Bp. The path from S to Bp is the shortest 133 path from S to B not going via P. If the network contains a path from 134 S to B that does not transit router P, i.e. the network is not 135 partitioned by the failure of P, then the packet will be successfully 136 delivered to B. When the packet addressed to Bp arrives at B, B 137 removes the encapsulation and forwards the repaired packet towards 138 its final destination. 140 Note that if the path from B to the final destination includes one or 141 more nodes that are included in the repair path, a packet may back 142 track after the encapsulation is removed. However, because the 143 decapsulating router is always closer to the packet destination than 144 the encapsulating router, the packet will not loop. 146 For complete protection, all of P's neighbors will require a not-via 147 address that allows traffic to be directed to them without traversing 148 P. This is shown in Figure 2. 150 A 151 |Ap 152 | 153 Sp Pa|Pb 154 S----------P----------B 155 Ps|Pc Bp 156 | 157 Cp| 158 C 160 Figure 2: The set of Not-via P Addresses 162 2.1. Use of Equal Cost Multi-Path 164 A router can use an equal cost multi-path (ECMP) repair in place of a 165 not-via repair. 167 A router computing a not-via repair path may subject the repair to 168 ECMP. 170 2.2. Use of LFA repairs 172 The not-via approach provides complete repair coverage and therefore 173 may be used as the sole repair mechanism. There are, however, 174 advantages in using not-via in combination with loop free alternates 175 (LFA) and or downstream paths as documented in [RFC5286]. 177 LFAs are computed on a per destination basis and in general, only a 178 subset of the destinations requiring repair will have a suitable LFA 179 repair. In this case, those destinations which are repairable by 180 LFAs are so repaired and the remainder of the destinations are 181 repaired using the not-via encapsulation. This has the advantage of 182 reducing the volume of traffic that requires encapsulation. On the 183 other hand, the path taken by an LFA repair may be less optimal than 184 that of the equivalent not-via repair for traffic destined to nodes 185 close to the far end of the failure, but may be more optimal for some 186 other traffic. The description in this document assumes that LFAs 187 will be used where available, but the distribution of repairs between 188 the two mechanisms is a local implementation choice. 190 3. Not-via Repair Path Computation 192 The not-via repair mechanism requires that all routers on the path 193 from S to B (Figure 1) have a route to Bp. They can calculate this 194 by failing node P, running a Shortest Path First Algorithm (SPF), and 195 finding the shortest route to B. 197 A router has no simple way of knowing whether it is on the shortest 198 path for any particular repair. It is therefore necessary for every 199 router to calculate the path it would use in the event of any 200 possible router failure. Each router therefore "fails" every router 201 in the network, one at a time, and calculates its own best route to 202 each of the neighbors of that router. In other words, with reference 203 to Figure 1, some router X will consider each router in turn to be P, 204 fail P, and then calculate its own route to each of the not-via P 205 addresses advertised by the neighbors of P. i.e. X calculates its 206 route to Sp, Ap, Bp, and Cp, in each case, not via P. 208 To calculate the repair paths a router has to calculate n-1 SPFs 209 where n is the number of routers in the network. This is expensive 210 to compute. However, the problem is amenable to a solution in which 211 each router (X) proceeds as follows. X first calculates the base 212 topology with all routers functional and determines its normal path 213 to all not-via addresses. This can be performed as part of the 214 normal SPF computation. For each router P in the topology, X then 215 performs the following actions:- 217 1. Removes router P from the topology. 219 2. Performs an incremental SPF (iSPF) [ISPF] on the modified 220 topology. The iSPF process involves detaching the sub-tree 221 affected by the removal of router P, and then re-attaching the 222 detached nodes. However, it is not necessary to run the iSPF to 223 completion. It is sufficient to run the iSPF up to the point 224 where all of the nodes advertising not-via P addresses have been 225 re-attached to the SPT, and then terminate it. 227 3. Reverts to the base topology. 229 This algorithm is significantly less expensive than a set of full 230 SPFs. Thus, although a router has to calculate the repair paths for 231 n-1 failures, the computational effort is much less than n-1 SPFs. 233 Experiments on a selection of real world network topologies with 234 between 40 and 400 nodes suggest that the worst-case computational 235 complexity using the above optimizations is equivalent to performing 236 between 5 and 13 full SPFs. Further optimizations are described in 237 section 6. 239 3.1. Computing not-via repairs in routing vector protocols 241 While this document focuses on link state routing protocols, it is 242 equally possible to compute not-via repairs in distance vector (e.g. 243 RIP) or path vector (e.g. BGP) routing protocols. This can be 244 achieved with very little protocol modification by advertising the 245 not-via address in the normal way, but ensuring that the information 246 about a not-via address Ps is not propagated through the node S. In 247 the case of link protection this simply means that the advertisement 248 from P to S is suppressed, with the result that S and all other nodes 249 compute a route to Ps which doesn't traverse S, as required. 251 In the case of node protection, where P is the protected node, and N 252 is some neighbor, the advertisement of Np must be suppressed not only 253 across the link N->P, but also across any link to P. The simplest way 254 of achieving this is for P itself to perform the suppression of any 255 address of the form Xp. 257 4. Operation of Repairs 259 This section explains the basic operation of the not-via repair of 260 node and link failure. 262 4.1. Node Failure 264 When router P fails (Figure 2) S encapsulates any packet that it 265 would send to B via P to Bp, and then sends the encapsulated packet 266 on the shortest path to Bp. S follows the same procedure for routers 267 A and C in Figure 2. The packet is decapsulated at the repair target 268 (A, B or C) and then forwarded normally to its destination. The 269 repair target can be determined as part of the normal SPF by 270 recording the "next-next-hop" for each destination in addition to the 271 normal next-hop. 273 Notice that with this technique only one level of encapsulation is 274 needed, and that it is possible to repair ANY failure regardless of 275 link metrics and any asymmetry that may be present in the network. 276 The only exception to this is where the failure was a single point of 277 failure that partitioned the network, in which case ANY repair is 278 clearly impossible. 280 4.2. Link Failure 282 The normal mode of operation of the network would be to assume router 283 failure. However, where some destinations are only reachable through 284 the failed router, it is desirable that an attempt be made to repair 285 to those destinations by assuming that only a link failure has 286 occurred. 288 To perform a link repair, S encapsulates to Ps (i.e. it instructs the 289 network to deliver the packet to P not-via S). All of the neighbors 290 of S will have calculated a path to Ps in case S itself had failed. 291 S could therefore give the packet to any of its neighbors (except, of 292 course, P). However, S should preferably send the encapsulated 293 packet on the shortest available path to P. This path is calculated 294 by running an SPF with the link SP failed. Note that this may again 295 be an incremental calculation, which can terminate when address Ps 296 has been reattached. 298 4.2.1. Loop Prevention Under Node Failure 300 It is necessary to consider the behavior of IPFRR solutions when a 301 link repair is attempted in the presence of node failure. In its 302 simplest form the not-via IPFRR solution prevents the formation of 303 loops forming as a result of mutual repair, by never providing a 304 repair path for a not-via address. The repair of packets with not- 305 via addresses is considered in more detail in Section 5.3. Referring 306 to Figure 2, if A was the neighbor of P that was on the link repair 307 path from S to P, and P itself had failed, the repaired packet from S 308 would arrive at A encapsulated to Ps. A would have detected that the 309 AP link had failed and would normally attempt to repair the packet. 310 However, no repair path is provided for any not-via address, and so A 311 would be forced to drop the packet, thus preventing the formation of 312 loop. 314 4.3. Multi-homed Prefixes 316 A multi-homed Prefix (MHP) is a prefix that is reachable via more 317 than one router in the network. Some of these may be repairable 318 using LFAs as described in [RFC5286]. Only those without such a 319 repair need be considered here. 321 When IPFRR router S (Figure 3) discovers that P has failed, it needs 322 to send packets addressed to the MHP X, which is normally reachable 323 through P, to an alternate router, which is still able to reach X. 325 X X X 326 | | | 327 | | | 328 | Sp |Pb | 329 Z...............S----------P----------B...............Y 330 Ps|Pc Bp 331 | 332 Cp| 333 C 335 Figure 3: Multi-homed Prefixes 337 S should choose the closest router that can reach X during the 338 failure as the alternate router. S determines which router to use as 339 the alternate while running the SPF with P failed. This is 340 accomplished by the normal process of re-attaching a leaf node to the 341 core topology (this is sometimes known as a "partial SPF"). 343 First, consider the case where the shortest alternate path to X is 344 via Z. S can reach Z without using the failed router P. However, S 345 cannot just send the packet towards Z, because the other routers in 346 the network will not be aware of the failure of P, and may loop the 347 packet back to S. S therefore encapsulates the packet to Z (using a 348 normal address for Z). When Z receives the encapsulated packet it 349 removes the encapsulation and forwards the packet to X. 351 Now consider the case where the shortest alternate path to X is via 352 Y, which S reaches via P and B. To reach Y, S must first repair the 353 packet to B using the normal not-via repair mechanism. To do this S 354 encapsulates the packet for X to Bp. When B receives the packet it 355 removes the encapsulation and discovers that the packet is intended 356 for MHP X. The situation now reverts to the previous case, in which 357 the shortest alternate path does not require traversal of the 358 failure. B therefore follows the algorithm above and encapsulates 359 the packet to Y (using a normal address for Y). Y removes the 360 encapsulation and forwards the packet to X. 362 It may be that the cost of reaching X using local delivery from the 363 alternate router (i.e. Z or Y) is greater than the cost of reaching 364 X via P. Under those circumstances, the alternate router would 365 normally forward to X via P, which would cause the IPFRR repair to 366 loop. To prevent the repair from looping the alternate router must 367 locally deliver a packet received via a repair encapsulation. This 368 may be specified by using a special address with the above semantics. 369 Note that only one such address is required per node. Notice that 370 using the not-via approach, only one level of encapsulation was 371 needed to repair MHPs to the alternate router. 373 4.4. Installation of Repair Paths 375 The following algorithm is used by node S (Figure 3) to pre- 376 calculate and install repair paths in the FIB, ready for immediate 377 use in the event of a failure. It is assumed that the not-via repair 378 paths have already been calculated as described above. 380 For each neighbor P, consider all destinations which are reachable 381 via P in the current topology:- 383 1. For all destinations with an ECMP or LFA repair (as described in 384 [RFC5286]) install that repair. 386 2. For each destination (DR) that remains, identify in the current 387 topology the next-next-hop (H) (i.e. the neighbor of P that P 388 will use to send the packet to DR). This can be determined 389 during the normal SPF run by recording the additional 390 information. If S has a path to the not-via address Hp (H not 391 via P), install a not-via repair to Hp for the destination DR. 393 3. Identify all remaining destinations (M) which can still be 394 reached when node P fails. These will be multi-homed prefixes 395 that are not repairable by LFA, for which the normal attachment 396 node is P, or a router for which P is a single point of failure, 397 and have an alternative attachment point that is reachable after 398 P has failed. One way of determining these destinations would be 399 to run an SPF rooted at S with node P removed, but an 400 implementation may record alternative attachment points during 401 the normal SPF run. In either case, the next best point of 402 attachment can also be determined for use in step (4) below. 404 4. For each multi-homed prefix (M) identified in step (3):- 406 A. Identify the new attachment node (as shown in Figure 3). 407 This may be:- 409 a. Y, where the next hop towards Y is P, or 411 b. Z, where the next hop towards Z is not P. 413 If the attachment node is Z, install the repair for M as a 414 tunnel to Z' (where Z' is the address of Z that is used to 415 force local forwarding). 417 B. For the subset of prefixes (M) that remain (having attachment 418 point Y), install the repair path previously installed for 419 destination Y. 421 For each destination (DS) that remains, install a not-via repair 422 to Ps (P not via S). Note, these are destinations for which node 423 P is a single point of failure, and can only be repaired by 424 assuming that the apparent failure of node P was simply a failure 425 of the S-P link. Note that, if available, a downstream path to P 426 may be used for such a repair. This cannot generate a persistent 427 loop in the event of the failure of node P, but if one neighbor 428 of P uses a not-via repair and another uses a downstream path, it 429 is possible for a packet sent on the downstream path to be 430 returned to the sending node inside a not-via encapsulation. 431 Since packets destined to not-via addresses are not repaired, the 432 packet will be dropped after executing a single turn loop. 434 5. Compound Failures 436 The following types of failures involve more than one component: 438 1. Shared Risk Link Groups 440 2. Local Area Networks 442 3. Multiple Independent Failures 444 The considerations that apply in each of the above situations are 445 described in the following sections. 447 5.1. Shared Risk Link Groups 449 A Shared Risk Link Group (SRLG) is a set of links whose failure can 450 be caused by a single action such as a conduit cut or line card 451 failure. When repairing the failure of a link that is a member of an 452 SRLG, it must be assumed that all the other links that are also 453 members of the SRLG have also failed. Consequently, any repair path 454 must be computed to avoid not just the adjacent link, but also all 455 the links which are members of the same SRLG. 457 In Figure 4 below, the links S-P and A-B are both members of SRLG 458 "a". The semantics of the not-via address Ps changes from simply "P 459 not-via the link S-P" to be "P not-via the link S-P or any other link 460 with which S-P shares an SRLG" In Figure 4 this is the links that are 461 members of SRLG "a". I.e. links S-P and A-B. Since the information 462 about SRLG membership of all links is available in the Link State 463 Database, all nodes computing routes to the not-via address Ps can 464 infer these semantics, and perform the computation by failing all the 465 links in the SRLG when running the iSPF. 467 Note that it is not necessary for S to consider repairs to any other 468 nodes attached to members of the SRLG (such as B). It is sufficient 469 for S to repair to the other end of the adjacent link (P in this 470 case). 472 a Ps 473 S----------P---------D 474 | | 475 | a | 476 A----------B 477 | | 478 | | 479 C----------E 481 Figure 4: Shared Risk Link Group 483 In some cases, it may be that the links comprising the SRLG occur in 484 series on the path from S to the destination D, as shown in Figure 5. 485 In this case, multiple consecutive repairs may be necessary. S will 486 first repair to Ps, then P will repair to Dp. In both cases, because 487 the links concerned are members of SRLG "a" the paths are computed to 488 avoid all members of SRLG "a". 490 a Ps a Dp 491 S----------P---------D 492 | | | 493 | a | | 494 A----------B | 495 | | | 496 | | | 497 C----------E---------F 499 Figure 5: Shared Risk Link Group members in series 501 While the use of multiple repairs in series introduces some 502 additional overhead, these semantics avoid the potential 503 combinatorial explosion of not-via addresses that could otherwise 504 occur. 506 Note that although multiple repairs are used, only a single level of 507 encapsulation is required. This is because the first repair packet 508 is decapsulated before the packet is re-encapsulated using the not- 509 via address corresponding to the far side of the next link which is a 510 member of the same SRLG. In some cases the de-capsulation and re- 511 encapsulation takes place (at least notionally) at a single node, 512 while in other cases, these functions may be performed by different 513 nodes. This scenario is illustrated in Figure 6 below. 515 a Ps a Dg 516 S----------P---------G--------D 517 | | | | 518 | a | | | 519 A----------B | | 520 | | | | 521 | | | | 522 C----------E---------F--------H 524 Figure 6: Shared Risk Link Group members in series 526 In this case, S first encapsulates to Ps, and node P decapsulates the 527 packet and forwards it "native" to G using its normal FIB entry for 528 destination D. G then repairs the packet to Dg. 530 It can be shown that such multiple repairs can never form a loop 531 because each repair causes the packet to move closer to its 532 destination. 534 It is often the case that a single link may be a member of multiple 535 SRLGs, and those SRLGs may not be isomorphic. This is illustrated in 536 Figure 7 below. 538 ab Ps a Dg 539 S----------P---------G--------D 540 | | | | 541 | a | | | 542 A----------B | | 543 | | | | 544 | b | | b | 545 C----------E---------F--------H 546 | | 547 | | 548 J----------K 550 Figure 7: Multiple Shared Risk Link Groups 552 The link SP is a member of SRLGs "a" and "b". When a failure of the 553 link SP is detected, it must be assumed that BOTH SRLGs have failed. 554 Therefore the not-via path to Ps must be computed by failing all 555 links which are members of SRLG "a" or SRLG "b". I.e. the semantics 556 of Ps is now "P not-via any links which are members of any of the 557 SRLGs of which link SP is a member". This is illustrated in Figure 8 558 below. 560 ab Ps a Dg 561 S----/-----P---------G---/----D 562 | | | | 563 | a | | | 564 A----/-----B | | 565 | | | | 566 | b | | b | 567 C----/-----E---------F---/----H 568 | | 569 | | 570 J----------K 572 Figure 8: Topology used for repair computation for link S-P 574 In this case, the repair path to Ps will be S-A-C-J-K-E-B-P. It may 575 appear that there is no path to D because GD is a member of SRLG "a" 576 and FH is a member of SRLG "b". This is true if BOTH SRLGs "a" and 577 "b" have in fact failed. But that would be an instance of multiple 578 uncorrelated failures which are out of scope for this design. In 579 practice it is likely that there is only a single failure, i.e. 580 either SRLG "a" or SRLG "b" has failed, but not both. These two 581 possibilities are indistinguishable from the point of view of the 582 repairing router S and so it must repair on the assumption that both 583 are unavailable. However, each link repair is considered 584 independently. The repair to Ps delivers the packet to P which then 585 forwards the packet to G. When the packet arrives at G, if SRLG "a" 586 has failed it will be repaired around the path G-F-H-D. This is 587 illustrated in Figure 9 below. If, on the other hand, SRLG "b" has 588 failed, link GD will still be available. In this case the packet 589 will be delivered as normal across the link GD. 591 ab Ps a Dg 592 S----/-----P---------G---/----D 593 | | | | 594 | a | | | 595 A----/-----B | | 596 | | | | 597 | b | | b | 598 C----------E---------F--------H 599 | | 600 | | 601 J----------K 603 Figure 9: Topology used for repair computation for link G-D 605 A repair strategy that assumes the worst-case failure for each link 606 can often result in longer repair paths than necessary. In cases 607 where only a single link fails, rather than the full SRLG, this 608 strategy may occasionally fail to identify a repair even though a 609 viable repair path exists in the network. The use of sub-optimal 610 repair paths is an inevitable consequence of this compromise 611 approach. The failure to identify any repair is a serious 612 deficiency, but is a rare occurrence in a robustly designed network. 613 This problem can be addressed by:- 615 1. Reporting that the link in question is irreparable, so that the 616 network designer can take appropriate action. 618 2. Modifying the design of the network to avoid this possibility. 620 3. Using some form of SRLG diagnostic (for example, by running BFD 621 over alternate repair paths) to determine which SRLG member(s) 622 has actually failed and using this information to select an 623 appropriate pre-computed repair path. However, aside from the 624 complexity of performing the diagnostics, this requires multiple 625 not-via addresses per interface, which has poor scaling 626 properties. 628 4. Using the mechanism described in Section 5.3 630 5.1.1. Use of LFAs with SRLGs 632 Section 4.1 above describes the repair of links which are members of 633 one or more SRLGs. LFAs can be used for the repair of such links 634 provided that any other link with which S-P shares an SRLG is avoided 635 when computing the LFA. This is described for the simple case of 636 "local-SRLGs" in [RFC5286]. 638 5.2. Local Area Networks 640 LANs are a special type of SRLG and are solved using the SRLG 641 mechanisms outlined above. With all SRLGs there is a trade-off 642 between the sophistication of the fault detection and the size of the 643 SRLG. Protecting against link failure of the LAN link(s) is 644 relatively straightforward, but as with all fast reroute mechanisms, 645 the problem becomes more complex when it is desired to protect 646 against the possibility of failure of the nodes attached to the LAN 647 as well as the LAN itself. 649 +--------------Q------C 650 | 651 | 652 | 653 A--------S-------(N)-------------P------B 654 | 655 | 656 | 657 +--------------R------D 659 Figure 10: Local Area Networks 661 Consider the LAN shown in Figure 10. For connectivity purposes, we 662 consider that the LAN is represented by the pseudonode (N). To 663 provide IPFRR protection, S must run a connectivity check to each of 664 its protected LAN adjacencies P, Q, and R, using, for example BFD 665 [RFC5880]. 667 When S discovers that it has lost connectivity to P, it is unsure 668 whether the failure is: 670 o its own interface to the LAN, 672 o the LAN itself, 674 o the LAN interface of P, 676 o the node P. 678 5.2.1. Simple LAN Repair 680 A simple approach to LAN repair is to consider the LAN and all of its 681 connected routers as a single SRLG. Thus, the address P not via the 682 LAN (Pl) would require P to be reached not-via any router connected 683 to the LAN. This is shown in Figure 11. 685 Ql Cl 686 +-------------Q--------C 687 | Qc 688 | 689 As Sl | Pl Bl 690 A--------S-------(N)------------P--------B 691 Sa | Pb 692 | 693 | Rl Dl 694 +-------------R--------D 695 Rd 697 Figure 11: Local Area Networks - LAN SRLG 699 In this case, when S detected that P had failed it would send traffic 700 reached via P and B to B not-via the LAN or any router attached to 701 the LAN (i.e. to Bl). Any destination only reachable through P would 702 be addressed to P not-via the LAN or any router attached to the LAN 703 (except of course P). 705 Whilst this approach is simple, it assumes that a large portion of 706 the network adjacent to the failure has also failed. This will 707 result in the use of sub-optimal repair paths and in some cases the 708 inability to identify a viable repair. 710 5.2.2. LAN Component Repair 712 In this approach, possible failures are considered at a finer 713 granularity, but without the use of diagnostics to identify the 714 specific component that has failed. Because S is unable to diagnose 715 the failure it must repair traffic sent through P and B, to B not- 716 via P,N (i.e. not via P and not via N), on the conservative 717 assumption that both the entire LAN and P have failed. Destinations 718 for which P is a single point of failure must as usual be sent to P 719 using an address that avoids the interface by which P is reached from 720 S, i.e. to P not-via N. Similarly for routers Q and R. 722 Notice that each router that is connected to a LAN must, as usual, 723 advertise one not-via address for each neighbor. In addition, each 724 router on the LAN must advertise an extra address not via the 725 pseudonode (N). 727 Notice also that each neighbor of a router connected to a LAN must 728 advertise two not-via addresses, the usual one not via the neighbor 729 and an additional one, not via either the neighbor or the pseudonode. 730 The required set of LAN address assignments is shown in Figure 12 731 below. Each router on the LAN, and each of its neighbors, is 732 advertising exactly one address more than it would otherwise have 733 advertised if this degree of connectivity had been achieved using 734 point-to-point links. 736 Qs Qp Qc Cqn 737 +--------------Q---------C 738 | Qr Qn Cq 739 | 740 Asn Sa Sp Sq | Ps Pq Pb Bpn 741 A--------S-------(N)-------------P---------B 742 As Sr Sn | Pr Pn Bp 743 | 744 | Rs Rp Pd Drn 745 +--------------R---------D 746 Rq Rn Dr 748 Figure 12: Local Area Networks 750 5.2.3. LAN Repair Using Diagnostics 752 A more specific LAN repair can be undertaken by using diagnostics. 753 In order to explicitly diagnose the failed network component, S 754 correlates the connectivity reports from P and one or more of the 755 other routers on the LAN, in this case, Q and R. If it lost 756 connectivity to P alone, it could deduce that the LAN was still 757 functioning and that the fault lay with either P, or the interface 758 connecting P to the LAN. It would then repair to B not via P (and P 759 not-via N for destinations for which P is a single point of failure) 760 in the usual way. If S lost connectivity to more than one router on 761 the LAN, it could conclude that the fault lay only with the LAN, and 762 could repair to P, Q and R not-via N, again in the usual way. 764 5.3. Multiple Independent Failures 766 IPFRR repair of multiple simultaneous failures which are not members 767 of a known SRLG is complicated by the problem that the use of 768 multiple concurrent repairs may result in looping repair paths. As 769 described in Section 4.2.1, the simplest method of preventing such 770 loops, is to ensure that packets addressed to a not-via address are 771 not repaired but instead are dropped. It is possible that a network 772 may experience multiple simultaneous failures. This may be due to 773 simple statistical effects, but the more likely cause is 774 unanticipated SRLGs. When multiple failures which are not part of an 775 anticipated group are detected, repairs are abandoned and the network 776 reverts to normal convergence. Although safe, this approach is 777 somewhat draconian, since there are many circumstances were multiple 778 repairs do not induce loops. 780 This section describes the properties of multiple unrelated failures 781 and proposes some methods that may be used to address this problem. 783 5.3.1. Looping Repairs 785 Let us assume that the repair mechanism is based on solely on not-via 786 repairs. LFA or downstream routes may be incorporated, and will be 787 dealt with later. 789 A------//------B------------D 790 / \ 791 / \ 792 F G 793 \ / 794 \ / 795 X------//------Y 797 Figure 13: The General Case of Multiple Failures 799 The essential case is as illustrated in Figure 13. Note that 800 depending on the repair case under consideration, there may be paths 801 present in Figure 13, that are in addition to those shown in the 802 figure. For example there may be paths between A and B, and/or 803 between X and Y. These paths are omitted for graphical clarity. 805 There are three cases to consider: 807 1) Consider the general case of a pair of protected links A-B and 808 X-Y as shown in the network fragment shown Figure 13. If the 809 repair path for A-B does not traverse X-Y and the repair path for 810 X-Y does not traverse A-B, this case is completely safe and will 811 not cause looping or packet loss. 813 A more common variation of this case is shown in Figure 14, which 814 shows two failures in different parts of the network in which a 815 packet from A to D traverses two concatenated repairs. 817 A------//------B------------X------//------Y------D 818 | | | | 819 | | | | 820 M--------------+ N--------------+ 822 Figure 14: Concatenated Repairs 824 2) In Figure 13, the repair for A-B traverses X-Y, but the repair 825 for X-Y does not traverse A-B. This case occurs when the not-via 826 path from A to B traverses link X-Y, but the not-via path from X 827 to Y traverses some path not shown in Figure 13. Without the 828 multi-failure mechanism described in this section the repaired 829 packet for A-B would be dropped when it reached X-Y, since the 830 repair of repaired packets would be forbidden. However, if this 831 packet were allowed to be repaired, the path to D would be 832 complete and no harm would be done, although two levels of 833 encapsulation would be required. 835 3) The repair for A-B traverses X-Y AND the repair for X-Y 836 traverses A-B. In this case unrestricted repair would result in 837 looping packets and increasing levels of encapsulation. 839 The challenge in applying IPFRR to a network that is undergoing 840 multiple failures is, therefore, to identify which of these cases 841 exist in the network and react accordingly. 843 5.3.2. Outline Solution 845 When A is computing the not-via repair path for A-B (i.e. the path 846 for packets addressed to Ba, read as "B not-via A") it is aware of 847 the list of nodes which this path traverses. This can be recorded by 848 a simple addition to the SPF process, and the not-via addresses 849 associated with each forward link can be determined. If the path 850 were A, F, X, Y, G, B, (Figure 13) the list of not-via addresses 851 would be: Fa, Xf, Yx, Gy, Bg. Under standard not-via operation, A 852 would populate its FIB such that all normal addresses normally 853 reachable via A-B would be encapsulated to Ba when A-B fails, but 854 traffic addressed to any not-via address arriving at A would be 855 dropped. The new procedure modifies this such that any traffic for a 856 not-via address normally reachable over A-B is also encapsulated to 857 Ba unless the not-via address is one of those previously identified 858 as being on the path to Ba, for example Yx, in which case the packet 859 is dropped. 861 The above procedure allows cases 1 and 2 above to be repaired, while 862 preventing the loop which would result from case 3. 864 Note that this is accomplished by pre-computing the required FIB 865 entries, and does not require any detailed packet inspection. The 866 same result could be achieved by checking for multiple levels of 867 encapsulation and dropping any attempt to triple encapsulate. 868 However, this would require more detailed inspection of the packet, 869 and causes difficulties when more than 2 "simultaneous" failures are 870 contemplated. 872 So far we have permitted benign repairs to coexist, albeit sometimes 873 requiring multiple encapsulation. Note that in many cases there will 874 be no performance impact since unless both failures are on the same 875 node, the two encapsulations or two decapsulations will be performed 876 at different nodes. There is however the issue of the MTU impact of 877 multiple encapsulations. 879 In the following sub-section we consider the various strategies that 880 may be applied to case 3 - mutual repairs that would loop. 882 5.3.3. Looping Repairs 884 In case 3, the simplest approach is to simply not install repairs for 885 repair paths that might loop. In this case, although the potentially 886 looping traffic is dropped, the traffic is not repaired. If we 887 assume that a hold-down is applied before reconvergence in case the 888 link failure was just a short glitch, and if a loop free convergence 889 mechanism further delays convergence, then the traffic will be 890 dropped for an extended period. In these circumstances it would be 891 better to "abandon all hope" (AAH) [I-D.ietf-rtgwg-ordered-fib] 892 (Appendix A) and immediately invoke normal re-convergence. 894 Note that it is not sufficient to expedite the issuance of an LSP 895 reporting the failure, since this may be treated as a permitted 896 simultaneous failure by the ordered FIB (oFIB) algorithm 897 [I-D.ietf-rtgwg-ordered-fib]. It is therefore necessary to 898 explicitly trigger an oFIB AAH. 900 5.3.3.1. Dropping Looping Packets 902 One approach to case 3 is to allow the repair, and to experimentally 903 discover the incompatibility of the repairs if and when they occur. 904 With this method we permit the repair in case 3 and trigger AAH when 905 a packet drop count on the not-via address has been incremented. 906 Alternatively, it is possible to wait until the LSP describing the 907 change is issued normally (i.e. when X announces the failure of X-Y). 908 When the repairing node A, which has precomputed that X-Y failures 909 are mutually incompatible with its own repairs receives this LSP it 910 can then issue the AAH. This has the disadvantage that it does not 911 overcome the hold-down delay, but it requires no "data-driven" 912 operation, and it still has the required effect of abandoning the 913 oFIB which is probably the longer of the delays (although with 914 signalled oFIB this should be sub-second). 916 Whilst both of the experimental approaches described above are 917 feasible, they tend to induce AAH in the presence of otherwise 918 feasible repairs, and they are contrary to the philosophy of repair 919 pre-determination that has been applied to existing IPFRR solutions. 921 5.3.3.2. Computing non-looping Repairs of Repairs 923 An alternative approach to simply dropping the looping packets, or to 924 detecting the loop after it has occurred, is to use secondary SRLGs. 925 With a link state routing protocol it is possible to precompute the 926 incompatibility of the repairs in advance and to compute an 927 alternative SRLG repair path. Although this does considerably 928 increase the computational complexity it may be possible to compute 929 repair paths that avoid the need to simply drop the offending 930 packets. 932 This approach requires us to identify the mutually incompatible 933 failures, and advertise them as "secondary SRLGs". When computing 934 the repair paths for the affected not-via addresses these links are 935 simultaneously failed. Note that the assumed simultaneous failure 936 and resulting repair path only applies to the repair path computed 937 for the conflicting not-via addresses, and is not used for normal 938 addresses. This implies that although there will be a longer repair 939 path when there is more than one failure, if there is a single 940 failure the repair path length will be "normal". 942 Ideally we would wish to only invoke secondary SRLG computation when 943 we are sure that the repair paths are mutually incompatible. 944 Consider the case of node A in Figure 13. A first identifies that 945 the repair path for A-B is via F-X-Y-G-B. It then explores this path 946 determining the repair path for each link in the path. Thus, for 947 example, it performs a check at X by running an SPF rooted at X with 948 the X-Y link removed to determine whether A-B is indeed on X's repair 949 path for packets addressed to Yx. 951 Some optimizations are possible in this calculation, which appears at 952 first sight to be order hk (where h is the average hop length of 953 repair paths and k is the average number of neighbours of a router). 954 When A is computing its set of repair paths, it does so for all its k 955 neighbours. In each case it identifies a list of node pairs 956 traversed by each repair. These lists may often have one or more 957 node pairs in common, so the actual number of link failures which 958 require investigation is the union of these sets. It is then 959 necessary to run an SPF rooted at the first node of each pair (the 960 first node because the pairings are ordered representing the 961 direction of the path), with the link to the second node removed. 962 This SPF, while not an incremental, can be terminated as soon as the 963 not-via address is reached. For example, when running the SPF rooted 964 at X, with the link X-Y removed, the SPF can be terminated when Yx is 965 reached. Once the path has been found, the path is checked to 966 determine if it traverses any of A's links in the direction away from 967 A. Note that, because the node pair XY may exist in the list for more 968 than one of A's links (i.e. it lies on more than one repair path), it 969 is necessary to identify the correct list, and hence link which has a 970 mutually looping repair path. That link of A is then advertised by A 971 as a secondary SRLG paired with the link X-Y. Also note that X will 972 be running this algorithm as well, and will identify that XY is 973 paired with A-B and so advertise it. This could perhaps be used as a 974 further check. 976 The ordering of the pairs in the lists is important. i.e. X-Y and 977 Y-X are dealt with separately. If and only if the repairs are 978 mutually incompatible, we need to advertise the pair of links as a 979 secondary SRLG, and then ALL nodes compute repair paths around both 980 failures using an additional not-via address with the semantics not- 981 via A-B AND not-via X-Y. 983 A further possibility is that because we are going to the trouble of 984 advertising these SRLG sets, we could also advertise the new repair 985 path and only get the nodes on that path to perform the necessary 986 computation. Note also that once we have reached Q space with 987 respect to the two failures we need no longer continue the 988 computation, so we only need to notify the nodes on the path that are 989 not in Q-space. 991 One cause of mutually looping repair paths is the existence of nodes 992 with only two links, or sections of the network which are only bi- 993 connected. In these cases, repair is clearly impossible - the 994 failure of both links partitions the network. It would be 995 advantageous to be able to identify these cases, and inhibit the 996 fruitless advertisement of the secondary SRLG information. This 997 could be achieved by the node detecting the requirement for a 998 secondary SRLG, first running the not-via computation with both links 999 removed. If this does not result in a path, it is clear that the 1000 network would be partitioned by such a failure, and so no 1001 advertisement is required. 1003 5.3.4. Mixing LFAs and Not-via 1005 So far in this section we have assumed that all repairs use not-via 1006 tunnels. However, in practise we may wish to use LFAs or downstream 1007 routes where available. This complicates the issue, because their 1008 use results in packets which are being repaired, but NOT addressed to 1009 not-via addresses. If BOTH links are using downstream routes there 1010 is no possibility of looping, since it is impossible to have a pair 1011 of nodes which are both downstream of each other [RFC5286]. 1013 Loops can however occur when LFAs are used. An obvious example is 1014 the well known node repair problem with LFAs [RFC5286]. If one link 1015 is using a downstream route, while the other is using a not-via 1016 tunnel, the potential mechanism described above would work provided 1017 it were possible to determine the nodes on the path of the downstream 1018 route. Some methods of computing downstream routes do not provide 1019 this path information. If the path information is however available, 1020 the link using a downstream route will have a discard FIB entry for 1021 the not-via address of the other link. The consequence is that 1022 potentially looping packets will be discarded when they attempt to 1023 cross this link. 1025 In the case where the mutual repairs are both using not-via repairs, 1026 the loop will be broken when the packet arrives at the second 1027 failure. However packets are unconditionally repaired by means of a 1028 downstream routes, and thus when the mutual pair consists of a 1029 downstream route and a not-via repair, the looping packet will only 1030 be dropped when it gets back to the first failure. i.e. it will 1031 execute a single turn of the loop before being dropped. 1033 There is a further complication with downstream routes, since 1034 although the path may be computed to the far side of the failure, the 1035 packet may "peel off" to its destination before reaching the far side 1036 of the failure. In this case it may traverse some other link which 1037 has failed and was not accounted for on the computed path. If the 1038 A-B repair (Figure 1) is a downstream route and the X-Y repair is a 1039 not-via repair, we can have the situation where the X-Y repair 1040 packets encapsulated to Yx follow a path which attempts to traverse 1041 A-B. If the A-B repair path for "normal" addresses is a downstream 1042 route, it cannot be assumed that the repair path for packets 1043 addressed to Yx can be sent to the same neighbour. This is because 1044 the validity of a downstream route must be ascertained in the 1045 topology represented by Yx, i.e. that with the link X-Y failed. This 1046 is not the same topology that was used for the normal downstream 1047 calculation, and use of the normal downstream route for the 1048 encapsulated packets may result in an undetected loop. If it is 1049 computationally feasible to check the downstream route in this 1050 topology (i.e. for any not-via address Qp which traverses A-B we must 1051 perform the downstream calculation for that not-via address in the 1052 topology with link Q-P failed.), then the downstream repair for Yx 1053 can safely be used. These packets cannot re-visit X-Y, since by 1054 definition they will avoid that link. Alternatively, the packet 1055 could be always repaired in a not-via tunnel. i.e. even though the 1056 normal repair for traffic traversing A-B would be to use a downstream 1057 route, we could insist that such traffic addressed to a not-via 1058 address must use a tunnel to Ba. Such a tunnel would only be 1059 installed for an address Qp if it were established that it did not 1060 traverse Q-P (using the rules described above). 1062 6. Optimizing not-via computations using LFAs 1064 If repairing node S has an LFA to the repair endpoint it is not 1065 necessary for any router to perform the incremental SPF with the link 1066 SP removed in order to compute the route to the not-via address Ps. 1067 This is because the correct routes will already have been computed as 1068 a result of the SPF on the base topology. Node S can signal this 1069 condition to all other routers by including a bit in its LSP or LSA 1070 associated with each LFA protected link. Routers computing not-via 1071 routes can then omit the running of the iSPF for links with this bit 1072 set. 1074 When running the iSPF for a particular link AB, the calculating 1075 router first checks whether the link AB is present in the existing 1076 SPT. If the link is not present in the SPT, no further work is 1077 required. This check is a normal part of the iSPF computation. 1079 If the link is present in the SPT, this optimization introduces a 1080 further check to determine whether the link is marked as protected by 1081 an LFA in the direction in which the link appears in the SPT. If so 1082 the iSPF need not be performed. For example, if the link appears in 1083 the SPT in the direction A->B and A has indicated that the link AB is 1084 protected by an LFA no further action is required for this link. 1086 If the receipt of this information is delayed, the correct operation 1087 of the protocol is not compromised provided that the necessity to 1088 perform a not-via computation is re-evaluated whenever new 1089 information arrives. 1091 This optimization is not particularly beneficial to nodes close to 1092 the repair since, as has been observed above, the computation for 1093 nodes on the LFA path is trivial. However, for nodes upstream of the 1094 link SP for which S-P is in the path to P, there is a significant 1095 reduction in the computation required. 1097 7. Multicast 1099 Multicast traffic can be repaired in a similar way to unicast. The 1100 multicast forwarder is able to use the not-via address to which the 1101 multicast packet was addressed as an indication of the expected 1102 receive interface and hence to correctly run the required RPF check. 1104 In some cases, all the destinations, including the repair endpoint, 1105 are repairable by an LFA. In this case, all unicast traffic may be 1106 repaired without encapsulation. Multicast traffic still requires 1107 encapsulation, but for the nodes on the LFA repair path the 1108 computation of the not-via forwarding entry is unnecessary since, by 1109 definition, their normal path to the repair endpoint is not via the 1110 failure. 1112 A more complete description of multicast operation is for further 1113 study. 1115 8. Fast Reroute in an MPLS LDP Network. 1117 Not-via addresses are IP addresses and LDP [RFC5036] will distribute 1118 labels for them in the usual way. The not-via repair mechanism may 1119 therefore be used to provide fast re-route in an MPLS network by 1120 first pushing the label which the repair endpoint uses to forward the 1121 packet, and then pushing the label corresponding to the not-via 1122 address needed to effect the repair. Referring once again to 1123 Figure 1, if S has a packet destined for D that it must reach via P 1124 and B, S first pushes B's label for D. S then pushes the label that 1125 its next hop to Bp needs to reach Bp. 1127 Note that in an MPLS LDP network it is necessary for S to have the 1128 repair endpoint's label for the destination. When S is effecting a 1129 link repair it already has this. In the case of a node repair, S 1130 either needs to set up a directed LDP session with each of its 1131 neighbor's neighbors, or it needs to use the next-next hop label 1132 distribution mechanism proposed in [I-D.shen-mpls-ldp-nnhop-label]. 1134 9. Encapsulation 1136 Any IETF specified IP in IP encapsulation may be used to carry a not- 1137 via repair. IP in IP [RFC2003], GRE [RFC1701] and L2TPv3 [RFC3931], 1138 all have the necessary and sufficient properties. The requirement is 1139 that both the encapsulating router and the router to which the 1140 encapsulated packet is addressed have a common ability to process the 1141 chosen encapsulation type. When an MPLS LDP network is being 1142 protected, the encapsulation would normally be an additional MPLS 1143 label. In an MPLS enabled IP network an MPLS label may be used in 1144 place of an IP in IP encapsulation in the case above. 1146 10. Routing Extensions 1148 IPFRR requires IGP extensions. Each IPFRR router that is directly 1149 connected to a protected network component must advertise a not-via 1150 address for that component. This must be advertised in such a way 1151 that the association between the protected component (link, router or 1152 SRLG) and the not-via address can be determined by the other routers 1153 in the network. 1155 It is necessary that not-via capable routers advertise in the IGP 1156 that they will calculate not-via routes. 1158 It is necessary for routers to advertise the type of encapsulation 1159 that they support (MPLS, GRE, L2TPv3 etc). However, the deployment 1160 of mixed IP encapsulation types within a network is discouraged. 1162 11. Incremental Deployment 1164 Incremental deployment is supported by excluding routers that are not 1165 calculating not-via routes (as indicated by their capability 1166 information flooded with their link state information) from the base 1167 topology used for the computation of repair paths. In that way 1168 repairs may be steered around islands of routers that are not IPFRR 1169 capable. Routers that are protecting a network component need to 1170 have the capability to encapsulate and decapsulate packets. However, 1171 routers that are on the repair path only need to be capable of 1172 calculating not-via paths and including the not-via addresses in 1173 their FIB i.e. these routers do not need any changes to their 1174 forwarding mechanism. 1176 12. IANA Considerations 1178 There are no IANA considerations that arise from this draft. 1180 13. Security Considerations 1182 The repair endpoints present vulnerability in that they might be used 1183 as a method of disguising the delivery of a packet to a point in the 1184 network. The primary method of protection should be through the use 1185 of a private address space for the not-via addresses. These 1186 addresses must not be advertised outside the area, and should be 1187 filtered at the network entry points. In addition, a mechanism might 1188 be developed that allowed the use of the mild security available 1189 through the use of a key [RFC1701] [RFC3931]. With the deployment of 1190 such mechanisms, the repair endpoints would not increase the security 1191 risk beyond that of existing IP tunnel mechanisms. An attacker may 1192 attempt to overload a router by addressing an excessive traffic load 1193 to the de-capsulation endpoint. Typically, routers take a 50% 1194 performance penalty in decapsulating a packet. The attacker could 1195 not be certain that the router would be impacted, and the extremely 1196 high volume of traffic needed, would easily be detected as an 1197 anomaly. If an attacker were able to influence the availability of a 1198 link, they could cause the network to invoke the not-via repair 1199 mechanism. A network protected by not-via IPFRR is less vulnerable 1200 to such an attack than a network that undertook a full convergence in 1201 response to a link up/down event. 1203 14. Acknowledgements 1205 The authors would like to acknowledge contributions made by Alia 1206 Atlas and John Harper. 1208 15. Informative References 1210 [I-D.ietf-rtgwg-ordered-fib] 1211 Shand, M., Bryant, S., Previdi, S., and C. Filsfils, 1212 "Loop-free convergence using oFIB", 1213 draft-ietf-rtgwg-ordered-fib-05 (work in progress), 1214 April 2011. 1216 [I-D.shen-mpls-ldp-nnhop-label] 1217 Shen, N., "Discovering LDP Next-Nexthop Labels", 1218 draft-shen-mpls-ldp-nnhop-label-02 (work in progress), 1219 May 2005. 1221 [ISPF] McQuillan, J., Richer, I., and E. Rosen, "ARPANET Routing 1222 Algorithm Improvements"", BBN Technical Report 3803, 1978. 1224 [RFC1701] Hanks, S., Li, T., Farinacci, D., and P. Traina, "Generic 1225 Routing Encapsulation (GRE)", RFC 1701, October 1994. 1227 [RFC2003] Perkins, C., "IP Encapsulation within IP", RFC 2003, 1228 October 1996. 1230 [RFC3931] Lau, J., Townsley, M., and I. Goyret, "Layer Two Tunneling 1231 Protocol - Version 3 (L2TPv3)", RFC 3931, March 2005. 1233 [RFC5036] Andersson, L., Minei, I., and B. Thomas, "LDP 1234 Specification", RFC 5036, October 2007. 1236 [RFC5286] Atlas, A. and A. Zinin, "Basic Specification for IP Fast 1237 Reroute: Loop-Free Alternates", RFC 5286, September 2008. 1239 [RFC5714] Shand, M. and S. Bryant, "IP Fast Reroute Framework", 1240 RFC 5714, January 2010. 1242 [RFC5880] Katz, D. and D. Ward, "Bidirectional Forwarding Detection 1243 (BFD)", RFC 5880, June 2010. 1245 Authors' Addresses 1247 Mike Shand 1248 Cisco Systems 1249 250, Longwater Avenue. 1250 Reading, Berks RG2 6GB 1251 UK 1253 Email: mshand@cisco.com 1255 Stewart Bryant 1256 Cisco Systems 1257 250, Longwater Avenue. 1258 Reading, Berks RG2 6GB 1259 UK 1261 Email: stbryant@cisco.com 1263 Stefano Previdi 1264 Cisco Systems 1265 Via Del Serafico, 200 1266 00142 Rome, 1267 Italy 1269 Email: sprevidi@cisco.com