idnits 2.17.1 draft-bryant-shand-lf-conv-frmwk-03.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- ** It looks like you're using RFC 3978 boilerplate. You should update this to the boilerplate described in the IETF Trust License Policy document (see https://trustee.ietf.org/license-info), which is required now. -- Found old boilerplate from RFC 3978, Section 5.1 on line 15. -- Found old boilerplate from RFC 3978, Section 5.5 on line 891. -- Found old boilerplate from RFC 3979, Section 5, paragraph 1 on line 867. -- Found old boilerplate from RFC 3979, Section 5, paragraph 2 on line 874. -- Found old boilerplate from RFC 3979, Section 5, paragraph 3 on line 880. ** This document has an original RFC 3978 Section 5.4 Copyright Line, instead of the newer IETF Trust Copyright according to RFC 4748. ** This document has an original RFC 3978 Section 5.5 Disclaimer, instead of the newer disclaimer which includes the IETF Trust according to RFC 4748. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- == No 'Intended status' indicated for this document; assuming Proposed Standard Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the RFC 3978 Section 5.4 Copyright Line does not match the current year -- 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 (Oct 2006) is 6396 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Missing Reference: 'RFC3036' is mentioned on line 106, but not defined ** Obsolete undefined reference: RFC 3036 (Obsoleted by RFC 5036) == Missing Reference: 'MPLS-TE' is mentioned on line 121, but not defined -- Obsolete informational reference (is this intentional?): RFC 3036 (ref. 'LDP') (Obsoleted by RFC 5036) -- Obsolete informational reference (is this intentional?): RFC 1305 (ref. 'NTP') (Obsoleted by RFC 5905) Summary: 4 errors (**), 0 flaws (~~), 4 warnings (==), 9 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 1 Network Working Group S. Bryant 2 Internet Draft M. Shand 3 Expiration Date: Sept 2006 Cisco Systems 5 Oct 2006 7 A Framework for Loop-free Convergence 8 10 Status of this Memo 12 By submitting this Internet-Draft, each author represents that any 13 applicable patent or other IPR claims of which he or she is aware 14 have been or will be disclosed, and any of which he or she becomes 15 aware will be disclosed, in accordance with Section 6 of BCP 79. 17 Internet-Drafts are working documents of the Internet Engineering 18 Task Force (IETF), its areas, and its working groups. Note that 19 other groups may also distribute working documents as 20 Internet-Drafts. 22 Internet-Drafts are draft documents valid for a maximum of six 23 months and may be updated, replaced, or obsoleted by other 24 documents at any time. It is inappropriate to use Internet-Drafts 25 as reference material or to cite them other than as "work in 26 progress." 28 The list of current Internet-Drafts can be accessed at 29 http://www.ietf.org/1id-abstracts.html 30 The list of Internet-Draft Shadow Directories can be accessed at 31 http://www.ietf.org/shadow.html 33 Abstract 34 This draft describes mechanisms that may be used to prevent or to 35 suppress the formation of micro-loops when an IP or MPLS network 36 undergoes topology change due to failure, repair or management 37 action. 39 Conventions used in this document 41 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 42 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in 43 this document are to be interpreted as described in RFC 2119 44 [RFC2119]. 46 Table of Contents 47 1. Introduction........................................................4 49 2. The Nature of Micro-loops...........................................5 51 3. Applicability.......................................................6 53 4. Micro-loop Control Strategies.......................................6 55 5. Loop mitigation.....................................................7 57 6. Micro-loop Prevention...............................................9 58 6.1. Incremental Cost Advertisement...................................9 59 6.2. Nearside Tunneling..............................................11 60 6.3. Farside Tunnels.................................................12 61 6.4. Distributed Tunnels.............................................13 62 6.5. Packet Marking..................................................13 63 6.6. MPLS New Labels.................................................13 64 6.7. Ordered FIB Update..............................................15 65 6.8. Synchronised FIB Update.........................................16 66 7. Using PLSN In Conjunction With Other Methods.......................17 68 8. Loop Suppression...................................................18 70 9. Compatibility Issues...............................................19 72 10. Comparison of Loop-free Convergence Methods.......................19 74 11. IANA considerations...............................................20 76 12. Security Considerations...........................................20 78 13. Intellectual Property Statement...................................20 80 14. Disclaimer of Validity............................................21 82 15. copyright Statement...............................................21 84 16. Normative References..............................................21 86 17. Informative References............................................21 88 18. Authors' Addresses................................................22 89 1. Introduction 91 When there is a change to the network topology (due to the failure 92 or restoration of a link or router, or as a result of management 93 action) the routers need to converge on a common view of the new 94 topology and the paths to be used for forwarding traffic to each 95 destination. During this process, referred to as a routing 96 transition, packet delivery between certain source/destination 97 pairs may be disrupted. This occurs due to the time it takes for 98 the topology change to be propagated around the network together 99 with the time it takes each individual router to determine and then 100 update the forwarding information base (FIB) for the affected 101 destinations. During this transition, packets may be lost due to 102 the continuing attempts to use the failed component, and due to 103 forwarding loops. Forwarding loops arise due to the inconsistent 104 FIBs that occur as a result of the difference in time taken by 105 routers to execute the transition process. This is a problem that 106 occurs in both IP networks and MPLS networks that use LDP [RFC3036] 107 as the label switched path (LSP) signaling protocol. 109 The service failures caused by routing transitions are largely 110 hidden by higher-level protocols that retransmit the lost data. 111 However new Internet services are emerging which are more sensitive 112 to the packet disruption that occurs during a transition. To make 113 the transition transparent to their users, these services require a 114 short routing transition. Ideally, routing transitions would be 115 completed in zero time with no packet loss. 117 Regardless of how optimally the mechanisms involved have been 118 designed and implemented, it is inevitable that a routing 119 transition will take some minimum interval that is greater than 120 zero. This has led to the development of a TE fast-reroute 121 mechanism for MPLS [MPLS-TE]. Alternative mechanisms that might be 122 deployed in an MPLS network and mechanisms that may be used in an 123 IP network are work in progress in the IETF [IPFRR]. Any repair 124 mechanism may however be disrupted by the formation of micro-loops 125 during the period between the time when the failure is announced, 126 and the time when all FIBs have been updated to reflect the new 127 topology. 129 There is, however, little point in introducing new mechanisms into 130 an IP network to provide fast re-route, without also deploying 131 mechanisms that prevent the disruptive effects of micro-loops which 132 may starve the repair or cause congestion loss as a result of 133 looping packets. 135 The disruptive effect of micro-loops is not confined to periods 136 when there is a component failure. Micro-loops can, for example, 137 form when a component is put back into service following repair. 138 Micro-loops can also form as a result of a network maintenance 139 action such as adding a new network component, removing a network 140 component or modifying a link cost. 142 This framework provides a summary of the mechanisms that have been 143 proposed to address the micro-loop issue. 145 2. The Nature of Micro-loops 147 Micro-loops may form during the periods when a network is re- 148 converging following ANY topology change, and are caused by 149 inconsistent FIBs in the routers. During the transition, micro- 150 loops may occur over a single link between a pair of routers that 151 temporarily use each other as the next hop for a prefix. Micro- 152 loops may also form when a cycle of routers have the next router in 153 the cycle as a next hop for a prefix. Cyclic micro-loops always 154 include at least one link with an asymmetric cost, and/or at least 155 two symmetric cost link cost changes within the convergence time. 157 Micro-loops have two undesirable side-effects; congestion and 158 repair starvation. A looping packet consumes bandwidth until it 159 either escapes as a result of the re-synchronization of the FIBs, 160 or its TTL expires. This transiently increases the traffic over a 161 link by as much as 128 times, and may cause the link to congest. 162 This congestion reduces the bandwidth available to other traffic 163 (which is not otherwise affected by the topology change). As a 164 result the "innocent" traffic using the link experiences increased 165 latency, and is liable to congestive packet loss. 167 In cases where the link or node failure has been protected by a 168 fast re-route repair, the inconsistency in the FIBs prevents some 169 traffic from reaching the failure and hence being repaired. The 170 repair may thus become starved of traffic and hence become 171 ineffective. Thus in addition to the congestive damage, the repair 172 is rendered ineffective by the micro-loop. Similarly, if the 173 topology change is the result of management action the link could 174 have been retained in service throughout the transition (i.e. the 175 link acts as its own repair path), however, if micro-loops form, 176 they prevent productive forwarding during the transition. 178 Unless otherwise controlled, micro-loops may form in any part of 179 the network that forwards (or in the case of a new link, will 180 forward) packets over a path that includes the affected topology 181 change. The time taken to propagate the topology change through the 182 network, and the non-uniform time taken by each router to calculate 183 the new shortest path tree (SPT) and update its FIB may 184 significantly extend the duration of the packet disruption caused 185 by the micro-loops. In some cases a packet may be subject to 186 disruption from micro-loops which occur sequentially at links along 187 the path, thus further extending the period of disruption beyond 188 that required to resolve a single loop. 190 3. Applicability 192 Loop free convergence techniques are applicable [APPL] to any 193 situation in which micro-loops may form. For example the 194 convergence of a network following: 196 1) Component failure. 198 2) Component repair. 200 3) Management withdrawal of a component. 202 4) Management insertion or a component. 204 5) Management change of link cost (either positive or negative). 206 6) External cost change, for example change of external gateway as 207 a result of a BGP change. 209 7) A Shared risk link group failure. 211 In each case, a component may be a link or a router. 212 Loop free convergence techniques are applicable to both IP networks 213 and MPLS enabled networks that use LDP, including LDP networks that 214 use the single-hop tunnel fast-reroute mechanism. 216 4. Micro-loop Control Strategies. 218 Micro-loop control strategies fall into three basic classes: 220 1. Micro-loop mitigation 222 2. Micro-loop prevention 224 3. Micro-loop suppression 226 A micro-loop mitigation scheme works by re-converging the network 227 in such a way that it reduces, but does not eliminate, the 228 formation of micro-loops. Such schemes cannot guarantee the 229 productive forwarding of packets during the transition. 231 A micro-loop prevention mechanism controls the re-convergence of 232 network in such a way that no micro-loops form. Such a micro-loop 233 prevention mechanism allows the continued use of any fast repair 234 method until the network has converged on its new topology, and 235 prevents the collateral damage that occurs to other traffic for the 236 duration of each micro-loop. 238 A micro-loop suppression mechanism attempts to eliminate the 239 collateral damage done by micro-loops to other traffic. This may be 240 achieved by, for example, using a packet monitoring method, which 241 detects that a packet is looping and drops it. Such schemes make no 242 attempt to productively forward the packet throughout the network 243 transition. 245 Note that all known micro-loop mitigation and micro-loop prevention 246 mechanisms extend the duration of the re-convergence process. When 247 the failed component is protected by a fast re-route repair this 248 implies that the converging network requires the repair to remain 249 in place for longer than would otherwise be the case. The extended 250 convergence time means any traffic which is NOT repaired by an 251 imperfect repair experiences a significantly longer outage than it 252 would experience with conventional convergence. 254 When a component is returned to service, or when a network 255 management action has taken place, this additional delay does not 256 cause traffic disruption, because there is no repair involved. 257 However the extended delay is undesirable, because it increases the 258 time that the network takes to be ready for another failure, and 259 hence leaves it vulnerable to multiple failures. 261 5. Loop mitigation 263 The only known loop mitigation approach is the Path Locking with 264 safe-neighbors (PLSN) method described in [ZININ]. In this method, 265 a micro-loop free next-hop safety condition is defined as follows: 266 In a symmetric cost network, it is safe for router X to change to 267 the use of neighbor Y as its next-hop for a specific destination if 268 the path through Y to that destination satisfies both of the 269 following criteria: 271 1. X considers Y as its loop-free neighbor based on the 272 topology before the change AND 274 2. X considers Y as its downstream neighbor based on the 275 topology after the change. 277 In an asymmetric cost network, a stricter safety condition is 278 needed, and the criterion is that: 280 X considers Y as its downstream neighbor based on the 281 topology both before and after the change. 283 Based on these criteria, destinations are classified by each router 284 into three classes: 286 Type A destinations: Destinations unaffected by the change and also 287 destinations whose next hop after the change satisfies the safety 288 criteria. 290 Type B destinations: Destinations that cannot be sent via the new 291 primary next-hop because the safety criteria are not satisfied, but 292 which can be sent via another next-hop that does satisfy the safety 293 criteria. 295 Type C destinations: All other destinations. 297 Following a topology change, Type A destinations are immediately 298 changed to go via the new topology. Type B destinations are 299 immediately changed to go via the next hop that satisfies the 300 safety criteria, even though this is not the shortest path. Type B 301 destinations continue to go via this path until all routers have 302 changed their Type C destinations over to the new next hop. Routers 303 must not change their Type C destinations until all routers have 304 changed their Type A2 and Type B destinations to the new or 305 intermediate (safe) next hop. 307 Simulations indicate that this approach produces a significant 308 reduction in the number of links that are subject to micro-looping. 309 However unlike all of the micro-loop prevention methods it is only 310 a partial solution. In particular, micro-loops may form on any link 311 joining a pair of type C routers. 313 Because routers delay updating their Type C destination FIB 314 entries, they will continue to route towards the failure during the 315 time when the routers are changing their Type A and B destinations, 316 and hence will continue to productively forward packets provided 317 that viable repair paths exist. 319 A backwards compatibility issue arises with PLSN. If a router is 320 not capable of micro-loop control, it will not correctly delay its 321 FIB update. If all such routers had only type A destinations this 322 loop mitigation mechanism would work as it was designed. 323 Alternatively, if all such incapable routers had only type C 324 destinations, the "covert" announcement mechanism used to trigger 325 the tunnel based schemes could be used to cause the Type A and Type 326 B destinations to be changed, with the incapable routers and 327 routers having type C destinations delaying until they received the 328 "real" announcement. Unfortunately, these two approaches are 329 mutually incompatible. 331 Note that simulations indicate that in most topologies treating 332 type B destinations as type C results in only a small degradation 333 in loop prevention. Also note that simulation results indicate that 334 in production networks where some, but not all, links have 335 asymmetric costs, using the stricter asymmetric cost criterion 336 actually REDUCES the number of loop free destinations, because 337 fewer destinations can be classified as type A or B. 339 This mechanism operates identically for both "bad-news" events, 340 "good-news" events and SRLG failure. 342 6. Micro-loop Prevention 344 Eight micro-loop prevention methods have been proposed: 346 1. Incremental cost advertisement 348 2. Nearside tunneling 350 3. Farside tunneling 352 4. Distributed tunnels 354 5. Packet marking 356 6. New MPLS labels 358 7. Ordered FIB update 360 8. Synchronized FIB update 362 6.1. Incremental Cost Advertisement 364 When a link fails, the cost of the link is normally changed from 365 its assigned metric to "infinity" in one step. However, it can be 366 proved that no micro-loops will form if the link cost is increased 367 in suitable increments, and the network is allowed to stabilize 368 before the next cost increment is advertised. Once the link cost 369 has been increased to a value greater than that of the lowest 370 alternative cost around the link, the link may be disabled without 371 causing a micro-loop. 373 The criterion for a link cost change to be safe is that any link 374 which is subjected to a cost change of x can only cause loops in a 375 part of the network that has a cyclic cost less than or equal to x. 376 Because there may exist links which have a cost of one in each 377 direction, resulting in a cyclic cost of two, this can result in 378 the link cost having to be raised in increments of one. However the 379 increment can be larger where the minimum cost permits. Determining 380 the minimum link cost in the network is trivial, but unfortunately, 381 calculating the optimum increment at each step is thought to be a 382 costly calculation. 384 This approach has the advantage that it requires no change to the 385 routing protocol. It will work in any network that uses a link- 386 state IGP because it does not require any co-operation from the 387 other routers in the network. However the method can be extremely 388 slow, particularly if large metrics are used. For the duration of 389 the transition some parts of the network continue to use the old 390 forwarding path, and hence use any repair mechanism for an extended 391 period. In the case of a failure that cannot be fully repaired, 392 some destinations may become unreachable for an extended period. 394 Where the micro-loop prevention mechanism was being used to support 395 a fast re-route repair the network may be vulnerable to a second 396 failure for the duration of the controlled re-convergence. 398 Where the micro-loop prevention mechanism was being used to support 399 a reconfiguration of the network the extended time is less of an 400 issue. In this case, because the real forwarding path is available 401 throughout the whole transition, there is no conflict between 402 concurrent change actions throughout the network. 404 It will be appreciated that when a link is returned to service, its 405 cost is reduced in small steps from "infinity" to its final cost, 406 thereby providing similar micro-loop prevention during a "good- 407 news" event. Note that the link cost may be decreased from 408 "infinity" to any value greater than that of the lowest alternative 409 cost around the link in one step without causing a micro-loop. 410 When the failure is an SRLG the link cost increments must be 411 coordinated across all members of the SRLG. This may be achieved by 412 completing the transition of one link before starting the next, or 413 by interleaving the changes. This can be achieved without the need 414 for any protocol extensions, by for example, using existing 415 identifiers to establish the ordering and the arrival of LSP/LSAs 416 to trigger the generation of the next increment. 418 6.2. Nearside Tunneling 420 This mechanism works by creating an overlay network using tunnels 421 whose path is not effected by the topology change and carrying the 422 traffic affected by the change in that new network. When all the 423 traffic is in the new, tunnel based, network, the real network is 424 allowed to converge on the new topology. Because all the traffic 425 that would be affected by the change is carried in the overlay 426 network no micro-loops form. 428 When a failure is detected (or a link is withdrawn from service), 429 the router adjacent to the failure issues a new ("covert") routing 430 message announcing the topology change. This message is propagated 431 through the network by all routers, but is only understood by 432 routers capable of using one of the tunnel based micro-loop 433 prevention mechanisms. 435 Each of the micro-loop preventing routers builds a tunnel to the 436 closest router adjacent to the failure. They then determine which 437 of their traffic would transit the failure and place that traffic 438 in the tunnel. When all of these tunnels are in place, the failure 439 is then announced as normal. Because these tunnels will be 440 unaffected by the transition, and because the routers protecting 441 the link will continue the repair (or forward across the link being 442 withdrawn), no traffic will be disrupted by the failure. When the 443 network has converged these tunnels are withdrawn, allowing traffic 444 to be forwarded along its new "natural" path. The order of tunnel 445 insertion and withdrawal is not important, provided that the 446 tunnels are all in place before the normal announcement is issued. 448 This method completes in bounded time, and is much faster than the 449 incremental cost method. Depending on the exact design, it 450 completes in two or three flood-SPF-FIB update cycles. 452 At the time at which the failure is announced as normal, micro- 453 loops may form within isolated islands of non-micro-loop preventing 454 routers. However, only traffic entering the network via such 455 routers can micro-loop. All traffic entering the network via a 456 micro-loop preventing router will be tunneled correctly to the 457 nearest repairing router, including, if necessary being tunneled 458 via a non-micro-loop preventing router, and will not micro-loop. 460 Where there is no requirement to prevent the formation of micro- 461 loops involving non-micro-loop preventing routers, a single, 462 "normal" announcement may be made, and a local timer used to 463 determine the time at which transition from tunneled forwarding to 464 normal forwarding over the new topology may commence. 466 This technique has the disadvantage that it requires traffic to be 467 tunneled during the transition. This is an issue in IP networks 468 because not all router designs are capable of high performance IP 469 tunneling. It is also an issue in MPLS networks because the 470 encapsulating router has to know the labels set that the 471 decapsulating router is distributing. 473 A further disadvantage of this method is that it requires co- 474 operation from all the routers within the routing domain to fully 475 protect the network against micro-loops. 477 When a new link is added, the mechanism is run in "reverse". When 478 the "covert" announcement is heard, routers determine which traffic 479 they will send over the new link, and tunnel that traffic to the 480 router on the near side of that link. This path will not be 481 affected by the presence of the new link. When the "normal" 482 announcement is heard, they then update their FIB to send the 483 traffic normally according to the new topology. Any traffic 484 encountering a router that has not yet updated its FIB will be 485 tunneled to the near side of the link, and will therefore not loop. 487 When a management change to the topology is required, again exactly 488 the same mechanism protects against micro-looping of packets by the 489 micro-loop preventing routers. 491 When the failure is an SRLG, the required strategy is to classify 492 traffic according the first member of the SRLG that it will 493 traverse on its way to the destination, and to tunnel that traffic 494 to the router that is closest to that SRLG member. This will 495 require multiple tunnel destinations, in the limiting case, one per 496 SRLG member. 498 6.3. Farside Tunnels 500 Farside tunneling loop prevention requires the loop preventing 501 routers to place all of the traffic that would traverse the failure 502 in one or more tunnels terminating at the router (or in the case of 503 node failure routers) at the far side of the failure. The 504 properties of this method are a more uniform distribution of repair 505 traffic than is a achieved using the nearside tunnel method, and in 506 the case of node failure, a reduction in the decapsulation load on 507 any single router. 509 Unlike the nearside tunnel method (which uses normal routing to the 510 repairing router), this method requires the use of a repair path to 511 the farside router. This may be provided by the not-via mechanism, 512 in which case no further computation is needed. 514 The mode of operation is otherwise identical to the nearside 515 tunneling loop prevention method (Section 6.2). 517 6.4. Distributed Tunnels 519 In the distributed tunnels loop prevention method, each router 520 calculates its own repair and forwards traffic affected by the 521 failure using that repair. Unlike the FRR case, the actual failure 522 is known at the time of the calculation. The objective of the loop 523 preventing routers is to get the packets that would have gone via 524 the failure into G-space [TUNNEL] using routers that are in F- 525 space. Because packets are decapsulated on entry to G-space, rather 526 than being forced to go to the farside of the failure, more optimum 527 routing may be achieved. This method is subject to the same 528 reachability constraints described in [TUNNEL]. 530 The mode of operation is otherwise identical to the nearside 531 tunneling loop prevention method (Section 6.2). 533 6.5. Packet Marking 535 If packets could be marked in some way, this information could be 536 used to assign them to one of: the new topology, the old topology 537 or a transition topology. They would then be correctly forwarded 538 during the transition. This could, for example, be achieved by 539 allocating a Type of Service bit to the task [RFC791]. This 540 mechanism works identically for both "bad-news" and "good-news" 541 events. It also works identically for SRLG failure. There are three 542 problems with this solution: 544 1) The packet marking bit may not available. 546 2) The mechanism would introduce a non-standard forwarding 547 procedure. 549 3) Packet marking using either the old or the new topology would 550 double the size of the FIB, however some optimizations may be 551 possible. 553 6.6. MPLS New Labels 555 In an MPLS network that is using LDP [LDP] for label distribution, 556 loop free convergence can be achieved through the use of new labels 557 when the path that a prefix will take through the network changes. 559 As described in Section 6.2, the repairing routers issue a covert 560 announcement to start the loop free convergence process. All loop 561 preventing routers calculate the new topology and determine whether 562 their FIB needs to be changed. If there is no change in the FIB 563 they take no part in the following process. 565 The routers that need to make a change to their FIB consider each 566 change and check the new next hop to determine whether it will use 567 a path in the OLD topology which reaches the destination without 568 traversing the failure (i.e. the next hop is in F-space with 569 respect to the failure [TUNNEL]). If so the FIB entry can be 570 immediately updated. For all of the remaining FIB entries, the 571 router issues a new label to each of its neighbors. This new label 572 is used to lock the path during the transition in a similar manner 573 to the previously described loop-free convergence with tunnels 574 method (Section 6.2). Routers receiving a new label install it in 575 their FIB, for MPLS label translation, but do not yet remove the 576 old label and do not yet use this new label to forward IP packets. 577 i.e. they prepare to forward using the new label on the new path, 578 but do not use it yet. Any packets received continue to be 579 forwarded the old way, using the old labels, towards the repair. 581 At some time after the covert announcement, an overt announcement 582 of the failure is issued. This announcement MUST NOT be issued 583 until such time as all routers have carried out all of their covert 584 announcement activities. On receipt of the overt announcement all 585 routers that were delaying convergence move to their new path for 586 both the new and the old labels. This involves changing the IP 587 address entries to use the new labels, AND changing the old labels 588 to forward using the new labels. 590 Because the new label path was installed during the covert phase, 591 packets reach their destinations as follows: 593 o If they do not go via any router using a new label they go 594 via the repairing router and the repair. 596 o If they meet any router that is using the new labels they 597 get marked with the new labels and reach their destination 598 using the new path, back-tracking if necessary. 600 When all routers have changed to the new path the network is 601 converged. At some time later, when it can be assumed that all 602 routers have moved to using the new path, the FIB can be cleaned up 603 to remove the, now redundant, old labels. 605 As with other method methods this new labels may be modified to 606 provide loop prevention for "good news". There are also a number of 607 optimizations of this method. Further details will be provided in a 608 forthcoming draft. 610 6.7. Ordered FIB Update 612 The Ordered FIB loop prevention method is described in [OFIB]. 613 Micro-loops occur following a failure or a cost increase, when a 614 router closer to the failed component revises its routes to take 615 account of the failure before a router which is further away. By 616 analyzing the reverse spanning tree over which traffic is directed 617 to the failed component in the old topology, it is possible to 618 determine a strict ordering which ensures that nodes closer to the 619 root always process the failure after any nodes further away, and 620 hence micro-loops are prevented. 622 When the failure has been announced, each router waits a multiple 623 of the convergence timer [TIMER]. The multiple is determined by the 624 node's position in the reverse spanning tree, and the delay value 625 is chosen to guarantee that a node can complete its processing 626 within this time. The convergence time may be reduced by employing 627 a signaling mechanism to notify the parent when all the children 628 have completed their processing, and hence when it was safe for the 629 parent to instantiate its new routes. 631 The property of this approach is therefore that it imposes a delay 632 which is bounded by the network diameter although in many cases it 633 will be much less. 635 When a link is returned to service the convergence process above is 636 reversed. A router first determines its distance (in hops) from the 637 new link in the NEW topology. Before updating its FIB, it then 638 waits a time equal to the value of that distance multiplied by the 639 convergence timer. 641 It will be seen that network management actions can similarly be 642 undertaken by treating a cost increase in a manner similar to a 643 failure and a cost decrease similar to a restoration. 645 The ordered FIB mechanism requires all nodes in the domain to 646 operate according to these procedures, and the presence of non 647 co-operating nodes can give rise to loops for any traffic which 648 traverses them (not just traffic which is originated through them). 649 Without additional mechanisms these loops could remain in place for 650 a significant time. 652 It should be noted that this method requires per router ordering, 653 but not per prefix ordering. A router must wait its turn to update 654 its FIB, but it should then update its entire FIB. 656 When an SRLG failure occurs a router must classify traffic into the 657 classes that pass over each member of the SRLG. Each router is then 658 independently assigned a ranking with respect to each SRLG member 659 for which they have a traffic class. These rankings may be 660 different for each traffic class. The prefixes of each class are 661 then changed in the FIB according to the ordering of their specific 662 ranking. Again, as for the single failure case, signaling may be 663 used to speed up the convergence process. 665 Note that the special SRLG case of a full or partial node failure, 666 can be deal with without using per prefix ordering, by running a 667 single reverse SPF rooted at the failed node (or common point of 668 the subset of failing links in the partial case). 670 There are two classes of signaling optimization that can be applied 671 to the ordered FIB loop-prevention method: 673 1. When the router makes NO change, it can signal 674 immediately. This significantly reduces the time taken by 675 the network to process long chains of routers that have no 676 change to make to their FIB. 678 2. When a router HAS changed, it can signal that it has 679 completed. This is more problematic since this may be 680 difficult to determine, particularly in a distributed 681 architecture, and the optimization obtained is the difference 682 between the actual time taken to make the FIB change and the 683 worst case timer value. This saving could be of the order of 684 one second per hop. 686 There is another method of executing ordered FIB which is based on 687 pure signaling [OB]. Methods that use signaling as an optimization 688 are safe because eventually they fall back on the established IGP 689 mechanisms which ensure that networks converge under conditions of 690 packet loss. However a mechanism that relies on signaling in order 691 to converge requires a reliable signaling mechanism which must be 692 proven to recover from any failure circumstance. 694 6.8. Synchronised FIB Update 696 Micro-loops form because of the asynchronous nature of the FIB 697 update process during a network transition. In many router 698 architectures it is the time taken to update the FIB itself that is 699 the dominant term. One approach would be to have two FIBs and, in a 700 synchronized action throughout the network, to switch from the old 701 to the new. One way to achieve this synchronized change would be to 702 signal or otherwise determine the wall clock time of the change, 703 and then execute the change at that time, using NTP [NTP] to 704 synchronize the wall clocks in the routers. 706 This approach has a number of major issues. Firstly two complete 707 FIBs are needed which may create a scaling issue and secondly a 708 suitable network wide synchronization method is needed. However, 709 neither of these are insurmountable problems. 711 Since the FIB change synchronization will not be perfect there may 712 be some interval during which micro-loops form. Whether this scheme 713 is classified as a micro-loop prevention mechanism or a micro-loop 714 mitigation mechanism within this taxonomy is therefore dependent on 715 the degree of synchronization achieved. 717 This mechanism works identically for both "bad-news" and "good- 718 news" events. It also works identically for SRLG failure. 719 Further consideration needs to be given to interoperating with 720 routers that do not support this mechanism. Without a suitable 721 interoperating mechanism, loops may form for the duration of the 722 synchronization delay. 724 7. Using PLSN In Conjunction With Other Methods 726 All of the tunnel methods and packet marking can be combined with 727 PLSN [ZININ] to reduce the traffic that needs to be protected by 728 the advanced method. Specifically all traffic could use PLSN except 729 traffic between a pair of routers both of which consider the 730 destination to be type C. The type C to type C traffic would be 731 protected from micro-looping through the use of a loop prevention 732 method. 734 However, determining whether the new next hop router considers a 735 destination to be type C may be computationally intensive. An 736 alternative approach would be to use a loop prevention method for 737 all local type C destinations. This would not require any 738 additional computation, but would require the additional loop 739 prevention method to be used in cases which would not have 740 generated loops (i.e. when the new next-hop router considered this 741 to be a type A or B destination). 743 The amount of traffic that would use PLSN is highly dependent on 744 the network topology and the specific change, but would be expected 745 to be in the region %70 to %90 in typical networks. 747 However, PLSN cannot be combined safely with Ordered FIB. Consider 748 the network fragment shown below: 750 R 751 /|\ 752 / | \ 753 1/ 2| \3 754 / | \ cost S->T = 10 755 Y-----X----S----T cost T->S = 1 756 | 1 2 | 757 |1 | 758 D---------------+ 759 20 761 On failure of link XY, according to PLSN, S will regard R as a safe 762 neighbor for traffic to D. However the ordered FIB rank of both R 763 and T will be zero and hence these can change their FIBs during the 764 same time interval. If R changes before T, then a loop will form 765 around R, T and S. This can be prevented by using a stronger safety 766 condition than PLSN currently specifies, at the cost of introducing 767 more type C routers, and hence reducing the PLSN coverage. 769 8. Loop Suppression 771 A micro-loop suppression mechanism recognizes that a packet is 772 looping and drops it. One such approach would be for a router to 773 recognize, by some means, that it had seen the same packet before. 774 It is difficult to see how sufficiently reliable discrimination 775 could be achieved without some form of per-router signature such as 776 route recording. A packet recognizing approach therefore seems 777 infeasible. 779 An alternative approach would be to recognize that a packet was 780 looping by recognizing that it was being sent back to the place 781 that it had just come from. This would work for the types of loop 782 that form in symmetric cost networks, but would not suppress the 783 cyclic loops that form in asymmetric networks. 785 This mechanism operates identically for both "bad-news" events, 786 "good-news" events and SRLG failure. 788 The problem with this class of micro-loop control strategies is 789 that whilst they prevent collateral damage they do nothing to 790 enhance the productive forwarding of packets during the network 791 transition. 793 9. Compatibility Issues 795 Deployment of any micro-loop control mechanism is a major change to 796 a network. Full consideration must be given to interoperation 797 between routers that are capable of micro-loop control, and those 798 that are not. Additionally there may be a desire to limit the 799 complexity of micro-loop control by choosing a method based purely 800 on its simplicity. Any such decision must take into account that if 801 a more capable scheme is needed in the future, its deployment will 802 be complicated by interaction with the scheme previously deployed. 804 10. Comparison of Loop-free Convergence Methods 806 PLSN [ZININ] is an efficient mechanism to prevent the formation of 807 micro-loops, but is only a partial solution. It is a useful adjunct 808 to some of the complete solutions, but may need modification. 810 Incremental cost advertisement is impractical as a general solution 811 because it takes too long to complete. However, it is universally 812 available, and hence may find use in certain network 813 reconfiguration operations. 815 Packet Marking is probably impractical because of the need to find 816 the marking bit and to change the forwarding behavior. 818 Of the remaining methods distributed tunnels is significantly more 819 complex than nearside or farside tunnels, and should only be 820 considered if there is a requirement to distribute the tunnel 821 decapsulation load. 823 Synchronised FIBs is a fast method, but has the issue that a 824 suitable synchronization mechanism needs to be defined. One method 825 would be to use NTP [NTP], however the coupling of routing 826 convergence to a protocol that uses the network may be a problem. 827 During the transition there will be some micro-looping for a short 828 interval because it is not possible to achieve complete 829 synchronization of the FIB changeover. 831 The ordered FIB mechanism has the major advantage that it is a 832 control plane only solution. However, SRLGs require a per- 833 destination calculation, and the convergence delay is high, bounded 834 by the network diameter. The use of signaling as an accelerator 835 will reduce the number of destinations that experience the full 836 delay, and hence reduce the total re-convergence time to an 837 acceptable period. 839 The nearside and farside tunnel methods deal relatively easily with 840 SRLGs and uncorrelated changes. The convergence delay would be 841 small. However these methods require the use of tunneled forwarding 842 which is not supported on all router hardware, and raises issues of 843 forwarding performance. When used with PLSN, the amount of traffic 844 that was tunneled would be significantly reduced, thus reducing the 845 forwarding performance concerns. If the selected repair mechanism 846 requires the use of tunnels, then a tunnel based loop prevention 847 scheme may be acceptable. 849 11. IANA considerations 851 There are no IANA considerations that arise from this draft. 853 12. Security Considerations 855 All micro-loop control mechanisms raise significant security issues 856 which must be addressed in their detailed technical description. 858 13. Intellectual Property Statement 860 The IETF takes no position regarding the validity or scope of any 861 Intellectual Property Rights or other rights that might be claimed 862 to pertain to the implementation or use of the technology described 863 in this document or the extent to which any license under such 864 rights might or might not be available; nor does it represent that 865 it has made any independent effort to identify any such rights. 866 Information on the procedures with respect to rights in RFC 867 documents can be found in BCP 78 and BCP 79. 869 Copies of IPR disclosures made to the IETF Secretariat and any 870 assurances of licenses to be made available, or the result of an 871 attempt made to obtain a general license or permission for the use 872 of such proprietary rights by implementers or users of this 873 specification can be obtained from the IETF on-line IPR repository 874 at http://www.ietf.org/ipr. 876 The IETF invites any interested party to bring to its attention any 877 copyrights, patents or patent applications, or other proprietary 878 rights that may cover technology that may be required to implement 879 this standard. Please address the information to the IETF at 880 ietf-ipr@ietf.org. 882 14. Disclaimer of Validity 884 This document and the information contained herein are provided on 885 an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE 886 REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND 887 THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, 888 EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT 889 THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR 890 ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A 891 PARTICULAR PURPOSE. 893 15. copyright Statement 895 Copyright (C) The Internet Society (2006). 897 This document is subject to the rights, licenses and restrictions 898 contained in BCP 78, and except as set forth therein, the authors 899 retain all their rights. 901 16. Normative References 903 There are no normative references. 905 17. Informative References 907 Internet-drafts are works in progress available from 908 910 [APPL] Bryant, S., Shand, M., "Applicability of Loop- 911 free Convergence", , October 2006, (work in 913 progress). 915 [IPFRR] Shand, M., "IP Fast-reroute Framework", 916 , 917 October 2006, (work in progress). 919 [LDP] Andersson, L., Doolan, P., Feldman, N., 920 Fredette, A. and B. Thomas, "LDP 921 Specification", RFC3036, 922 January 2001. 924 [NTP] RFC1305 Network Time Protocol (Version 3) 925 Specification, Implementation and Analysis. D. 926 Mills. March 1992. 928 [OB] Avoiding transient loops during IGP convergence 929 P. Francois, O. Bonaventure 930 IEEE INFOCOM 2005, March 2005, Miami, Fl., USA 932 [OFIB] Francois et. al., "Loop-free convergence using 933 ordered FIB updates", , October 2006 (work in progress). 936 [RFC2119] Bradner, S., "Key words for use in RFCs to 937 Indicate Requirement Levels", BCP 14, RFC2119, 938 March 1997. 940 [RFC791] RFC791, Internet Protocol Protocol 941 Specification, September 1981 943 [TIMER] S. Bryant, et. al. , "Synchronisation of Loop 944 Free Timer Values", , October 2006 946 [TUNNEL] Bryant, S., Shand, M., "IP Fast Reroute using 947 tunnels", , 948 Apr 2005 (work in progress). 950 [ZININ] Zinin, A., "Analysis and Minimization of 951 Microloops in Link-state Routing Protocols", 952 , 953 February 2006 (work in progress). 955 18. Authors' Addresses 957 Mike Shand 958 Cisco Systems, 959 250, Longwater Ave, 960 Green Park, 961 Reading, RG2 6GB, 962 United Kingdom. Email: mshand@cisco.com 964 Stewart Bryant 965 Cisco Systems, 966 250, Longwater Ave, 967 Green Park, 968 Reading, RG2 6GB, 969 United Kingdom. Email: stbryant@cisco.com