idnits 2.17.1 draft-ietf-rtgwg-rlfa-node-protection-10.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The abstract seems to contain references ([RFC7490]), which it shouldn't. Please replace those with straight textual mentions of the documents in question. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (December 30, 2016) is 2671 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) No issues found here. Summary: 1 error (**), 0 flaws (~~), 1 warning (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Routing Area Working Group P. Sarkar, Ed. 3 Internet-Draft Individual Contributor 4 Intended status: Standards Track S. Hegde 5 Expires: July 3, 2017 C. Bowers 6 Juniper Networks, Inc. 7 H. Gredler 8 RtBrick, Inc. 9 S. Litkowski 10 Orange 11 December 30, 2016 13 Remote-LFA Node Protection and Manageability 14 draft-ietf-rtgwg-rlfa-node-protection-10 16 Abstract 18 The loop-free alternates computed following the current Remote-LFA 19 specification guarantees only link-protection. The resulting Remote- 20 LFA nexthops (also called PQ-nodes), may not guarantee node- 21 protection for all destinations being protected by it. 23 This document describes an extension to the Remote Loop-Free based IP 24 fast reroute mechanisms described in [RFC7490], that describes 25 procedures for determining if a given PQ-node provides node- 26 protection for a specific destination or not. The document also 27 shows how the same procedure can be uitilized for collection of 28 complete characteristics for alternate paths. Knowledge about the 29 characteristics of all alternate path is precursory to apply operator 30 defined policy for eliminating paths not fitting constraints. 32 Requirements Language 34 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 35 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 36 document are to be interpreted as described in RFC2119 [RFC2119]. 38 Status of This Memo 40 This Internet-Draft is submitted in full conformance with the 41 provisions of BCP 78 and BCP 79. 43 Internet-Drafts are working documents of the Internet Engineering 44 Task Force (IETF). Note that other groups may also distribute 45 working documents as Internet-Drafts. The list of current Internet- 46 Drafts is at http://datatracker.ietf.org/drafts/current/. 48 Internet-Drafts are draft documents valid for a maximum of six months 49 and may be updated, replaced, or obsoleted by other documents at any 50 time. It is inappropriate to use Internet-Drafts as reference 51 material or to cite them other than as "work in progress." 53 This Internet-Draft will expire on July 3, 2017. 55 Copyright Notice 57 Copyright (c) 2016 IETF Trust and the persons identified as the 58 document authors. All rights reserved. 60 This document is subject to BCP 78 and the IETF Trust's Legal 61 Provisions Relating to IETF Documents 62 (http://trustee.ietf.org/license-info) in effect on the date of 63 publication of this document. Please review these documents 64 carefully, as they describe your rights and restrictions with respect 65 to this document. Code Components extracted from this document must 66 include Simplified BSD License text as described in Section 4.e of 67 the Trust Legal Provisions and are provided without warranty as 68 described in the Simplified BSD License. 70 Table of Contents 72 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 73 2. Node Protection with Remote-LFA . . . . . . . . . . . . . . . 3 74 2.1. The Problem . . . . . . . . . . . . . . . . . . . . . . . 4 75 2.2. Additional Definitions . . . . . . . . . . . . . . . . . 6 76 2.2.1. Link-Protecting Extended P-Space . . . . . . . . . . 6 77 2.2.2. Node-Protecting Extended P-Space . . . . . . . . . . 6 78 2.2.3. Q-Space . . . . . . . . . . . . . . . . . . . . . . . 6 79 2.2.4. Link-Protecting PQ Space . . . . . . . . . . . . . . 6 80 2.2.5. Candidate Node-Protecting PQ Space . . . . . . . . . 7 81 2.2.6. Cost-Based Definitions . . . . . . . . . . . . . . . 7 82 2.2.6.1. Link-Protecting Extended P-Space . . . . . . . . 7 83 2.2.6.2. Node-Protecting Extended P-Space . . . . . . . . 7 84 2.2.6.3. Q-Space . . . . . . . . . . . . . . . . . . . . . 8 85 2.3. Computing Node-protecting R-LFA Path . . . . . . . . . . 9 86 2.3.1. Computing Candidate Node-protecting PQ-Nodes for 87 Primary nexthops . . . . . . . . . . . . . . . . . . 9 88 2.3.2. Computing node-protecting paths from PQ-nodes to 89 destinations . . . . . . . . . . . . . . . . . . . . 11 90 2.3.3. Computing Node-Protecting R-LFA Paths for 91 Destinations with ECMP primary nexthop nodes . . . . 13 92 2.3.4. Limiting extra computational overhead . . . . . . . . 17 93 3. Manageability of Remote-LFA Alternate Paths . . . . . . . . . 18 94 3.1. The Problem . . . . . . . . . . . . . . . . . . . . . . . 18 95 3.2. The Solution . . . . . . . . . . . . . . . . . . . . . . 18 97 4. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 19 98 5. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 19 99 6. Security Considerations . . . . . . . . . . . . . . . . . . . 19 100 7. References . . . . . . . . . . . . . . . . . . . . . . . . . 19 101 7.1. Normative References . . . . . . . . . . . . . . . . . . 19 102 7.2. Informative References . . . . . . . . . . . . . . . . . 20 103 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 20 105 1. Introduction 107 The Remote-LFA [RFC7490] specification provides loop-free alternates 108 that guarantee only link-protection. The resulting Remote-LFA 109 alternate nexthops (also referred to as the PQ-nodes) may not provide 110 node-protection for all destinations covered by the same, in case of 111 failure of the primary nexthop node. Neither does the specification 112 provide a means to determine the same. 114 Also, the LFA Manageability [RFC7916] document requires a computing 115 router to find all possible (including all possible Remote-LFA) 116 alternate nexthops, collect the complete set of path characteristics 117 for each alternate path, run an alternate-selection policy 118 (configured by the operator) and find the best alternate path. This 119 will require the Remote-LFA implementation to gather all the required 120 path characteristics along each link on the entire Remote-LFA 121 alternate path. 123 With current LFA [RFC5286] and Remote-LFA implementations, the 124 forward SPF (and reverse SPF) is run with the computing router and 125 its immediate 1-hop routers as the roots. While that enables 126 computation of path attributes (e.g. SRLG, Admin-groups) for first 127 alternate path segment from the computing router to the PQ-node, 128 there is no means for the computing router to gather any path 129 attributes for the path segment from the PQ-node to destination. 130 Consequently any policy-based selection of alternate paths will 131 consider only the path attributes from the computing router up until 132 the PQ-node. 134 This document describes a procedure for determining node-protection 135 with Remote-LFA. The same procedure is also extended for collection 136 of a complete set of path attributes, enabling more accurate policy- 137 based selection for alternate paths obtained with Remote-LFA. 139 2. Node Protection with Remote-LFA 141 Node-protection is required to provide protection of traffic on a 142 given forwarding node, against the failure of the first-hop node on 143 the primary forwarding path. Such protection becomes more critical 144 in the absence of mechanisms like non-stop-routing in the network. 146 Certain operators refrain from deploying non-stop-routing in their 147 network, due to the required complex state synchronization between 148 redundant control plane hardwares it requires, and the significant 149 additional performance complexities it hence introduces. In such 150 cases node-protection is essential to guarantee un-interrupted flow 151 of traffic, even in the case of an entire forwarding node going down. 153 The following sections discuss the node-protection problem in the 154 context of Remote-LFA and propose a solution. 156 2.1. The Problem 158 To better illustrate the problem and the solution proposed in this 159 document the following topology diagram from the Remote-LFA [RFC7490] 160 draft is being re-used with slight modification. 162 D1 163 / 164 S-x-E 165 / \ 166 N R3--D2 167 \ / 168 R1---R2 170 Figure 1: Topology 1 172 In the above topology, for all (non-ECMP) destinations reachable via 173 the S-E link there is no standard LFA alternate. As per the Remote- 174 LFA [RFC7490] alternate specifications node R2 being the only PQ-node 175 for the S-E link provides nexthop for all the above destinations. 176 Table 1 below, shows all possible primary and Remote-LFA alternate 177 paths for each destination. 179 +-------------+--------------+---------+-------------------------+ 180 | Destination | Primary Path | PQ-node | Remote-LFA Backup Path | 181 +-------------+--------------+---------+-------------------------+ 182 | R3 | S->E->R3 | R2 | S=>N=>R1=>R2->R3 | 183 | E | S->E | R2 | S=>N=>R1=>R2->R3->E | 184 | D1 | S->E->D1 | R2 | S=>N=>R1=>R2->R3->E->D1 | 185 | D2 | S->E->R3->D2 | R2 | S=>N=>R1=>R2->R3->D2 | 186 +-------------+--------------+---------+-------------------------+ 188 Table 1: Remote-LFA backup paths via PQ-node R2 190 A closer look at Table 1 shows that, while the PQ-node R2 provides 191 link-protection for all the destinations, it does not provide node- 192 protection for destinations E and D1. In the event of the node- 193 failure on primary nexthop E, the alternate path from Remote-LFA 194 nexthop R2 to E and D1 also becomes unavailable. So for a Remote-LFA 195 nexthop to provide node-protection for a given destination, it is 196 mandatory that, the shortest path from the given PQ-node to the given 197 destination MUST NOT traverse the primary nexthop. 199 In another extension of the topology in Figure 1 let us consider an 200 additional link between N and E with the same cost as the other 201 links. 203 D1 204 / 205 S-x-E 206 / / \ 207 N---+ R3--D2 208 \ / 209 R1---R2 211 Figure 2: Topology 2 213 In the above topology, the S-E link is no more on any of the shortest 214 paths from N to R3, E and D1. Hence R3, E and D1 are also included 215 in both the Extended-P space and Q space of E (w.r.t S-E link). 216 Table 2 below, shows all possible primary and R-LFA alternate paths 217 via PQ-node R3, for each destination reachable through the S-E link 218 in the above topology. The R-LFA alternate paths via PQ-node R2 219 remains same as in Table 1. 221 +-------------+--------------+---------+------------------------+ 222 | Destination | Primary Path | PQ-node | Remote-LFA Backup Path | 223 +-------------+--------------+---------+------------------------+ 224 | R3 | S->E->R3 | R3 | S=>N=>E=>R3 | 225 | E | S->E | R3 | S=>N=>E=>R3->E | 226 | D1 | S->E->D1 | R3 | S=>N=>E=>R3->E->D1 | 227 | D2 | S->E->R3->D2 | R3 | S=>N=>E=>R3->D2 | 228 +-------------+--------------+---------+------------------------+ 230 Table 2: Remote-LFA backup paths via PQ-node R3 232 Again a closer look at Table 2 shows that, unlike Table 1, where the 233 single PQ-node R2 provided node-protection for destinations R3 and 234 D2, if we choose R3 as the R-LFA nexthop, it does not provide node- 235 protection for R3 and D2 anymore. If S chooses R3 as the R-LFA 236 nexthop, in the event of the node-failure on primary nexthop E, on 237 the alternate path from S to R-LFA nexthop R3, one of parallel ECMP 238 path between N and R3 also becomes unavailable. So for a Remote-LFA 239 nexthop to provide node-protection for a given destination, it is 240 also mandatory that, the shortest paths from S to the chosen PQ-node 241 MUST NOT traverse the primary nexthop node. 243 2.2. Additional Definitions 245 This document adds and enhances the following definitions extending 246 the ones mentioned in Remote-LFA [RFC7490] specification. 248 2.2.1. Link-Protecting Extended P-Space 250 The Remote-LFA [RFC7490] specification already defines this. The 251 link-protecting extended P-space for a link S-E being protected is 252 the set of routers that are reachable from one or more direct 253 neighbors of S, except primary node E, without traversing the S-E 254 link on any of the shortest paths from the direct neighbor to the 255 router. This MUST exclude any direct neighbor for which there is at 256 least one ECMP path from the direct neighbor traversing the link(S-E) 257 being protected. 259 For a cost-based definition for Link-protecting Extended P-Space 260 refer to Section 2.2.6.1. 262 2.2.2. Node-Protecting Extended P-Space 264 The node-protecting extended P-space for a primary nexthop node E 265 being protected, is the set of routers that are reachable from one or 266 more direct neighbors of S, except primary node E, without traversing 267 the node E. This MUST exclude any direct neighbors for which there 268 is at least one ECMP path from the direct neighbor traversing the 269 node E being protected. 271 For a cost-based definition for Node-protecting Extended P-Space 272 refer to Section 2.2.6.2. 274 2.2.3. Q-Space 276 The Remote-LFA [RFC7490] draft already defines this. The Q-space for 277 a link S-E being protected is the set of nodes that can reach primary 278 node E, without traversing the S-E link on any of the shortest paths 279 from the node itself to primary nexthop E. This MUST exclude any 280 node for which there is at least one ECMP path from the node to the 281 primary nexthop E traversing the link(S-E) being protected. 283 For a cost-based definition for Q-Space refer to Section 2.2.6.3. 285 2.2.4. Link-Protecting PQ Space 287 A node Y is in link-protecting PQ space w.r.t the link (S-E) being 288 protected, if and only if, Y is present in both link-protecting 289 extended P-space and the Q-space for the link being protected. 291 2.2.5. Candidate Node-Protecting PQ Space 293 A node Y is in candidate node-protecting PQ space w.r.t the node (E) 294 being protected, if and only if, Y is present in both node-protecting 295 extended P-space and the Q-space for the link being protected. 297 Please note, that a node Y being in candidate node-protecting PQ- 298 space, does not guarantee that the R-LFA alternate path via the same, 299 in entirety, is unaffected in the event of a node failure of primary 300 nexthop node E. It only guarantees that the path segment from S to 301 PQ-node Y is unaffected by the same failure event. The PQ-nodes in 302 the candidate node-protecting PQ space may provide node protection 303 for only a subset of destinations that are reachable through the 304 corresponding primary link. 306 2.2.6. Cost-Based Definitions 308 This section provides cost-based definitions for some of the terms 309 introduced in Section 2.2 of this document. 311 2.2.6.1. Link-Protecting Extended P-Space 313 Please refer to Section 2.2.1 for a formal definition for Link- 314 protecting Extended P-Space. 316 A node Y is in link-protecting extended P-space w.r.t the link (S-E) 317 being protected, if and only if, there exists at least one direct 318 neighbor of S, Ni, other than primary nexthop E, that satisfies the 319 following condition. 321 D_opt(Ni,Y) < D_opt(Ni,S) + D_opt(S,Y) 323 Where, 324 D_opt(A,B) : Distance on most optimum path from A to B. 325 Ni : A direct neighbor of S other than primary 326 nexthop E. 327 Y : The node being evaluated for link-protecting 328 extended P-Space. 330 Figure 3: Link-Protecting Ext-P-Space Condition 332 2.2.6.2. Node-Protecting Extended P-Space 334 Please refer to Section 2.2.2 for a formal definition for Node- 335 protecting Extended P-Space. 337 A node Y is in node-protecting extended P-space w.r.t the node E 338 being protected, if and only if, there exists at least one direct 339 neighbor of S, Ni, other than primary nexthop E, that satisfies the 340 following condition. 342 D_opt(Ni,Y) < D_opt(Ni,E) + D_opt(E,Y) 344 Where, 345 D_opt(A,B) : Distance on most optimum path from A to B. 346 E : The primary nexthop on shortest path from S 347 to destination. 348 Ni : A direct neighbor of S other than primary 349 nexthop E. 350 Y : The node being evaluated for node-protecting 351 extended P-Space. 353 Figure 4: Node-Protecting Ext-P-Space Condition 355 Please note, that a node Y satisfying the condition in Figure 4 above 356 only guarantees that the R-LFA alternate path segment from S via 357 direct neighbor Ni to the node Y is not affected in the event of a 358 node failure of E. It does not yet guarantee that the path segment 359 from node Y to the destination is also unaffected by the same failure 360 event. 362 2.2.6.3. Q-Space 364 Please refer to Section 2.2.3 for a formal definition for Q-Space. 366 A node Y is in Q-space w.r.t the link (S-E) being protected, if and 367 only if, the following condition is satisfied. 369 D_opt(Y,E) < D_opt(S,E) + D_opt(Y,S) 371 Where, 372 D_opt(A,B) : Distance on most optimum path from A to B. 373 E : The primary nexthop on shortest path from S 374 to destination. 375 Y : The node being evaluated for Q-Space. 377 Figure 5: Q-Space Condition 379 2.3. Computing Node-protecting R-LFA Path 381 The R-LFA alternate path through a given PQ-node to a given 382 destination is comprised of two path segments as follows. 384 1. Path segment from the computing router to the PQ-node (Remote-LFA 385 alternate nexthop), and 387 2. Path segment from the PQ-node to the destination being protected. 389 So to ensure a R-LFA alternate path for a given destination provides 390 node-protection we need to ensure that none of the above path 391 segments are affected in the event of failure of the primary nexthop 392 node. Sections Section 2.3.1 and Section 2.3.2 show how this can be 393 ensured. 395 2.3.1. Computing Candidate Node-protecting PQ-Nodes for Primary 396 nexthops 398 To choose a node-protecting R-LFA nexthop for a destination R3, 399 router S needs to consider a PQ-node from the candidate node- 400 protecting PQ-space for the primary nexthop E on shortest path from S 401 to R3. As mentioned in Section 2.2.2, to consider a PQ-node as 402 candidate node-protecting PQ-node, there must be at least one direct 403 neighbor Ni of S, such that all shortest paths from Ni to the PQ-node 404 does not traverse primary nexthop node E. 406 Implementations SHOULD run the inequality in Section 2.2.2 Figure 4 407 for all direct neighbors, other than primary nexthop node E, to 408 determine whether a node Y is a candidate node-protecting PQ-node. 409 All of the metrics needed by this inequality would have been already 410 collected from the forward SPFs rooted at each of direct neighbor S, 411 computed as part of standard LFA [RFC5286] implementation. With 412 reference to the topology in Figure 2, Table 3 below shows how the 413 above condition can be used to determine the candidate node- 414 protecting PQ-space for S-E link (primary nexthop E). 416 +------------+----------+----------+----------+---------+-----------+ 417 | Candidate | Direct | D_opt | D_opt | D_opt | Condition | 418 | PQ-node | Nbr (Ni) | (Ni,Y) | (Ni,E) | (E,Y) | Met | 419 | (Y) | | | | | | 420 +------------+----------+----------+----------+---------+-----------+ 421 | R2 | N | 2 (N,R2) | 1 (N,E) | 2 | Yes | 422 | | | | | (E,R2) | | 423 | R3 | N | 2 (N,R3) | 1 (N,E) | 1 | No | 424 | | | | | (E,R3) | | 425 +------------+----------+----------+----------+---------+-----------+ 427 Table 3: Node-protection evaluation for R-LFA repair tunnel to PQ- 428 node 430 As seen in the above Table 3, R3 does not meet the node-protecting 431 extended-p-space inequality and so, while R2 is in candidate node- 432 protecting PQ space, R3 is not. 434 Some SPF implementations may also produce a list of links and nodes 435 traversed on the shortest path(s) from a given root to others. In 436 such implementations, router S may have executed a forward SPF with 437 each of its direct neighbors as the SPF root, executed as part of the 438 standard LFA [RFC5286] computations. So S may re-use the list of 439 links and nodes collected from the same SPF computations, to decide 440 whether a node Y is a candidate node-protecting PQ-node or not. A 441 node Y shall be considered as a node-protecting PQ-node, if and only 442 if, there is at least one direct neighbor of S, other than the 443 primary nexthop E, for which, the primary nexthop node E does not 444 exist on the list of nodes traversed on any of the shortest paths 445 from the direct neighbor to the PQ-node. Table 4 below is an 446 illustration of the mechanism with the topology in Figure 2. 448 +-----------+-------------------+-----------------+-----------------+ 449 | Candidate | Repair Tunnel | Link-Protection | Node-Protection | 450 | PQ-node | Path(Repairing | | | 451 | | router to PQ- | | | 452 | | node) | | | 453 +-----------+-------------------+-----------------+-----------------+ 454 | R2 | S->N->R1->R2 | Yes | Yes | 455 | R2 | S->E->R3->R2 | No | No | 456 | R3 | S->N->E->R3 | Yes | No | 457 +-----------+-------------------+-----------------+-----------------+ 459 Table 4: Protection of Remote-LFA tunnel to the PQ-node 461 As seen in the above Table 4 while R2 is candidate node-protecting 462 Remote-LFA nexthop for R3 and D2, it is not so for E and D1, since 463 the primary nexthop E is in the shortest path from R2 to E and D1. 465 2.3.2. Computing node-protecting paths from PQ-nodes to destinations 467 Once a computing router finds all the candidate node-protecting PQ- 468 nodes for a given directly attached primary link, it shall follow the 469 procedure as proposed in this section, to choose one or more node- 470 protecting R-LFA paths, for destinations reachable through the same 471 primary link in the primary SPF graph. 473 To find a node-protecting R-LFA path for a given destination, the 474 computing router needs to pick a subset of PQ-nodes from the 475 candidate node-protecting PQ-space for the corresponding primary 476 nexthop, such that all the path(s) from the PQ-node(s) to the given 477 destination remain unaffected in the event of a node failure of the 478 primary nexthop node. To determine whether a given PQ-node belongs 479 to such a subset of PQ-nodes, the computing router MUST ensure that 480 none of the primary nexthop node are found on any of the shortest 481 paths from the PQ-node to the given destination. 483 This document proposes an additional forward SPF computation for each 484 of the PQ-nodes, to discover all shortest paths from the PQ-nodes to 485 the destination. This will help determine, if a given primary 486 nexthop node is on the shortest paths from the PQ-node to the given 487 destination or not. To determine if a given candidate node- 488 protecting PQ-node provides node-protecting alternate for a given 489 destination, or not, all the shortest paths from the PQ-node to the 490 given destination has to be inspected, to check if the primary 491 nexthop node is found on any of these shortest paths. To compute all 492 the shortest paths from a candidate node-protecting PQ-node to one 493 (or more) destination, the computing router MUST run the forward SPF 494 on the candidate node-protecting PQ-node. Soon after running the 495 forward SPF, the computer router SHOULD run the inequality in 496 Figure 6 below, once for each destination. A PQ-node that does not 497 qualify the condition for a given destination, does not guarantee 498 node-protection for the path segment from the PQ-node to the specific 499 destination. 501 D_opt(Y,D) < D_opt(Y,E) + Distance_opt(E,D) 503 Where, 504 D_opt(A,B) : Distance on most optimum path from A to B. 505 D : The destination node. 506 E : The primary nexthop on shortest path from S 507 to destination. 508 Y : The node-protecting PQ-node being evaluated 510 Figure 6: Node-Protecting Condition for PQ-node to Destination 512 All of the above metric costs except D_opt(Y, D), can be obtained 513 with forward and reverse SPFs with E(the primary nexthop) as the 514 root, run as part of the regular LFA and Remote-LFA implementation. 515 The Distance_opt(Y, D) metric can only be determined by the 516 additional forward SPF run with PQ-node Y as the root. With 517 reference to the topology in Figure 2, Table 5 below shows how the 518 above condition can be used to determine node-protection with node- 519 protecting PQ-node R2. 521 +-------------+------------+---------+--------+---------+-----------+ 522 | Destination | Primary-NH | D_opt | D_opt | D_opt | Condition | 523 | (D) | (E) | (Y, D) | (Y, E) | (E, D) | Met | 524 +-------------+------------+---------+--------+---------+-----------+ 525 | R3 | E | 1 | 2 | 1 | Yes | 526 | | | (R2,R3) | (R2,E) | (E,R3) | | 527 | E | E | 2 | 2 | 0 (E,E) | No | 528 | | | (R2,E) | (R2,E) | | | 529 | D1 | E | 3 | 2 | 1 | No | 530 | | | (R2,D1) | (R2,E) | (E,D1) | | 531 | D2 | E | 2 | 2 | 1 | Yes | 532 | | | (R2,D2) | (R2,E) | (E,D2) | | 533 +-------------+------------+---------+--------+---------+-----------+ 535 Table 5: Node-protection evaluation for R-LFA path segment between 536 PQ-node and destination 538 As seen in the above example above, R2 does not meet the node- 539 protecting inequality for destination E, and D1. And so, once again, 540 while R2 is a node-protecting Remote-LFA nexthop for R3 and D2, it is 541 not so for E and D1. 543 In SPF implementations that also produce a list of links and nodes 544 traversed on the shortest path(s) from a given root to others, the 545 inequality in Figure 6 above need not be evaluated. Instead, to 546 determine whether a PQ-node provides node-protection for a given 547 destination or not, the list of nodes computed from forward SPF run 548 on the PQ-node, for the given destination, SHOULD be inspected. In 549 case the list contains the primary nexthop node, the PQ-node does not 550 provide node-protection. Else, the PQ-node guarantees node- 551 protecting alternate for the given destination. Below is an 552 illustration of the mechanism with candidate node-protecting PQ-node 553 R2 in the topology in Figure 2. 555 +-------------+-----------------+-----------------+-----------------+ 556 | Destination | Shortest Path | Link-Protection | Node-Protection | 557 | | (Repairing | | | 558 | | router to PQ- | | | 559 | | node) | | | 560 +-------------+-----------------+-----------------+-----------------+ 561 | R3 | R2->R3 | Yes | Yes | 562 | E | R2->R3->E | Yes | No | 563 | D1 | R2->R3->E->D1 | Yes | No | 564 | D2 | R2->R3->D2 | Yes | Yes | 565 +-------------+-----------------+-----------------+-----------------+ 567 Table 6: Protection of Remote-LFA path between PQ-node and 568 destination 570 As seen in the above example while R2 is candidate node-protecting 571 R-LFA nexthop for R3 and D2, it is not so for E and D1, since the 572 primary nexthop E is in the shortest path from R2 to E and D1. 574 The procedure described in this document helps no more than to 575 determine whether a given Remote-LFA alternate provides node- 576 protection for a given destination or not. It does not find out any 577 new Remote-LFA alternate nexthops, outside the ones already computed 578 by standard Remote-LFA procedure. However, in case of availability 579 of more than one PQ-node (Remote-LFA alternates) for a destination, 580 and node-protection is required for the given primary nexthop, this 581 procedure will eliminate the PQ-nodes that do not provide node- 582 protection and choose only the ones that does. 584 2.3.3. Computing Node-Protecting R-LFA Paths for Destinations with ECMP 585 primary nexthop nodes 587 In certain scenarios, when one or more destinations maybe reachable 588 via multiple ECMP (equal-cost-multi-path) nexthop nodes, and only 589 link-protection is required, there is no need to compute any 590 alternate paths for such destinations. In the event of failure of 591 one of the nexthop links, the remaining primary nexthops shall always 592 provide link-protection. However, if node-protection is required, 593 the rest of the primary nexthops may not guarantee node-protection. 594 Figure 7 below shows one such example topology. 596 D1 597 2 / 598 S---x---E1 599 / \ / \ 600 / x / \ 601 / \ / \ 602 N-------E2 R3--D2 603 \ 2 / 604 \ / 605 \ / 606 R1-------R2 607 2 609 Primary Nexthops: 610 Destination D1 = [{ S-E1, E1}, {S-E2, E2}] 611 Destination D2 = [{ S-E1, E1}, {S-E2, E2}] 613 Figure 7: Topology with multiple ECMP primary nexthops 615 In the above example topology, costs of all links are 1, except the 616 following links: 618 Link: S-E1, Cost: 2 620 Link: N-E2: Cost: 2 622 Link: R1-R2: Cost: 2 624 In the above topology, on computing router S, destinations D1 and D2 625 are reachable via two ECMP nexthop nodes E1 and E2. However the 626 primary paths via nexthop node E2 also traverses via the nexthop node 627 E1. So in the event of node failure of nexthop node E1, both primary 628 paths (via E1 and E2) becomes unavailable. Hence if node-protection 629 is desired for destinations D1 and D2, alternate paths that does not 630 traverse any of the primary nexthop nodes E1 and E2, need to be 631 computed. In the above topology the only alternate neighbor N does 632 not provide such a LFA alternate path. Hence one (or more) R-LFA 633 node-protecting alternate paths for destinations D1 and D2, needs to 634 be computed. 636 In the above topology, following are the link-protecting PQ-nodes. 638 Primary Nexthop: E1, Link-Protecting PQ-Node: { R2 } 640 Primary Nexthop: E2, Link-Protecting PQ-Node: { R2 } 642 To find one (or more) node-protecting R-LFA paths for destinations D1 643 and D2, one (or more) node-protecting PQ-node(s) needs to be 644 determined first. Inequalities specified in Section 2.2.6.2 and 645 Section 2.2.6.3 can be evaluated to compute the node-protecting PQ- 646 space for each of the nexthop nodes E1 and E2, as shown in Table 7 647 below. To select a PQ-node as node-protecting PQ-node for a 648 destination with multiple primary nexthop nodes, the PQ-node MUST 649 satisfy the inequality for all primary nexthop nodes. Any PQ-node 650 which is NOT node-protecting PQ-node for all the primary nexthop 651 nodes, MUST NOT be chosen as the node-protecting PQ-node for 652 destination. 654 +--------+----------+-------+--------+--------+---------+-----------+ 655 | Primar | Candidat | Direc | D_opt | D_opt | D_opt | Condition | 656 | y Next | e PQ- | t Nbr | (Ni,Y) | (Ni,E) | (E,Y) | Met | 657 | hop | node (Y) | (Ni) | | | | | 658 | (E) | | | | | | | 659 +--------+----------+-------+--------+--------+---------+-----------+ 660 | E1 | R2 | N | 3 | 3 | 2 | Yes | 661 | | | | (N,R2) | (N,E1) | (E1,R2) | | 662 | E2 | R2 | N | 3 | 2 | 3 | Yes | 663 | | | | (N,R2) | (N,E2) | (E2,R2) | | 664 +--------+----------+-------+--------+--------+---------+-----------+ 666 Table 7: Computing Node-protected PQ-nodes for nexthop E1 and E2 668 In SPF implementations that also produce a list of links and nodes 669 traversed on the shortest path(s) from a given root to others, the 670 tunnel-repair paths from the computing router to candidate PQ-node 671 can be examined to ensure that none of the primary nexthop nodes is 672 traversed. PQ-nodes that provide one (or more) Tunnel-repair 673 paths(s) that does not traverse any of the primary nexthop nodes, are 674 to be considered as node-protecting PQ-nodes. Table 8 below shows 675 the possible tunnel-repair paths to PQ-node R2. 677 +--------------+------------+-------------------+-------------------+ 678 | Primary-NH | PQ-Node | Tunnel-Repair | Exclude All | 679 | (E) | (Y) | Paths | Primary-NH | 680 +--------------+------------+-------------------+-------------------+ 681 | E1, E2 | R2 | S==>N==>R1==>R2 | Yes | 682 +--------------+------------+-------------------+-------------------+ 684 Table 8: Tunnel-Repair paths to PQ-node R2 686 From Table 7 and Table 8, in the above example, R2 being node- 687 protecting PQ-node for both primary nexthops E1 and E2, should be 688 chosen as the node-protecting PQ-node for destinations D1 and D2 that 689 are both reachable via primary nexthop nodes E1 and E2. 691 Next, to find a node-protecting R-LFA path from node-protecting PQ- 692 node to destinations D1 and D2, inequalities specified in Figure 6 693 should be evaluated, to ensure if R2 provides a node-protecting R-LFA 694 path for each of these destinations, as shown below in Table 9. For 695 a R-LFA path to qualify as node-protecting R-LFA path for a 696 destination with multiple ECMP primary nexthop nodes, the R-LFA path 697 from the PQ-node to the destination MUST satisfy the inequality for 698 all primary nexthop nodes. 700 +----------+----------+-------+--------+--------+--------+----------+ 701 | Destinat | Primary- | PQ- | D_opt | D_opt | D_opt | Conditio | 702 | ion (D) | NH (E) | Node | (Y, D) | (Y, E) | (E, D) | n Met | 703 | | | (Y) | | | | | 704 +----------+----------+-------+--------+--------+--------+----------+ 705 | D1 | E1 | R2 | 3 (R2, | 2 (R2, | 1 (E1, | No | 706 | | | | D1) | E1) | D1) | | 707 | D1 | E2 | R2 | 3 (R2, | 3 (R2, | 2 (E2, | Yes | 708 | | | | D1) | E2) | D1) | | 709 | D2 | E1 | R2 | 2 (R2, | 2 (R2, | 2 (E1, | Yes | 710 | | | | D2) | E1) | D2) | | 711 | D2 | E2 | R2 | 2 (R2, | 2 (R2, | 3 (E2, | Yes | 712 | | | | D2) | E2) | D2) | | 713 +----------+----------+-------+--------+--------+--------+----------+ 715 Table 9: Finding node-protecting R-LFA path for destinations D1 and 716 D2 718 In SPF implementations that also produce a list of links and nodes 719 traversed on the shortest path(s) from a given root to others, the 720 R-LFA paths via node-protecting PQ-node to final destination can be 721 examined to ensure that none of the primary nexthop nodes is 722 traversed. R-LFA path(s) that does not traverse any of the primary 723 nexthop nodes, guarantees node-protection in the event of failure of 724 any of the primary nexthop nodes. Table 10 below shows the possible 725 R-LFA-paths for destinations D1 and D2 via the node-protecting PQ- 726 node R2. 728 +-------------+------------+---------+-----------------+------------+ 729 | Destination | Primary-NH | PQ-Node | R-LFA Paths | Exclude | 730 | (D) | (E) | (Y) | | All | 731 | | | | | Primary-NH | 732 +-------------+------------+---------+-----------------+------------+ 733 | D1 | E1, E2 | R2 | S==>N==>R1==>R2 | No | 734 | | | | -->R3-->E1-->D1 | | 735 | | | | | | 736 | D2 | E1, E2 | R2 | S==>N==>R1==>R2 | Yes | 737 | | | | -->R3-->D2 | | 738 +-------------+------------+---------+-----------------+------------+ 740 Table 10: R-LFA paths for destinations D1 and D2 742 From Table 9 and Table 10, in the example above, the R-LFA path from 743 R2 does not meet the node-protecting inequality for destination D1, 744 while it does meet the same inequality for destination D2. And so, 745 while R2 provides node-protecting R-LFA alternate for D2, it fails to 746 provide node-protection for destination D1. Finally, while it is 747 possible to get a node-protecting R-LFA path for D2, no such node- 748 protecting R-LFA path can be found for D1. 750 2.3.4. Limiting extra computational overhead 752 In addition to the extra reverse SPF computations suggested by the 753 Remote-LFA [RFC7490] draft (one reverse SPF for each of the directly 754 connected neighbors), this document proposes a forward SPF 755 computations for each PQ-node discovered in the network. Since the 756 average number of PQ-nodes found in any network is considerably more 757 than the number of direct neighbors of the computing router, the 758 proposal of running one forward SPF per PQ-node may add considerably 759 to the overall SPF computation time. 761 To limit the computational overhead of the approach proposed, this 762 document proposes that implementations MUST choose a subset from the 763 entire set of PQ-nodes computed in the network, with a finite limit 764 on the number of PQ-nodes in the subset. Implementations MUST choose 765 a default value for this limit and may provide user with a 766 configuration knob to override the default limit. Implementations 767 MUST also evaluate some default preference criteria while considering 768 a PQ-node in this subset. Finally, implementations MAY also allow 769 the user to override the default preference criteria, by providing a 770 policy configuration for the same. 772 This document proposes that implementations SHOULD use a default 773 preference criteria for PQ-node selection which will put a score on 774 each PQ-node, proportional to the number of primary interfaces for 775 which it provides coverage, its distance from the computing router, 776 and its router-id (or system-id in case of IS-IS). PQ-nodes that 777 cover more primary interfaces SHOULD be preferred over PQ-nodes that 778 cover fewer primary interfaces. When two or more PQ-nodes cover the 779 same number of primary interfaces, PQ-nodes which are closer (based 780 on metric) to the computing router SHOULD be preferred over PQ-nodes 781 farther away from it. For PQ-nodes that cover the same number of 782 primary interfaces and are the same distance from the computing 783 router, the PQ-node with smaller router-id (or system-id in case of 784 IS-IS) SHOULD be preferred. 786 Once a subset of PQ-nodes is found, computing router shall run a 787 forward SPF on each of the PQ-nodes in the subset to continue with 788 procedures proposed in Section 2.3.2. 790 3. Manageability of Remote-LFA Alternate Paths 792 3.1. The Problem 794 With the regular Remote-LFA [RFC7490] functionality the computing 795 router may compute more than one PQ-node as usable Remote-LFA 796 alternate nexthops. Additionally an alternate selection policy may 797 be configured to enable the network operator to choose one of them as 798 the most appropriate Remote-LFA alternate. For such policy-based 799 alternate selection to run, all the relevant path characteristics for 800 each the alternate paths (one through each of the PQ-nodes), needs to 801 be collected. As mentioned before in Section 2.3 the R-LFA alternate 802 path through a given PQ-node to a given destination is comprised of 803 two path segments. 805 The first path segment (i.e. from the computing router to the PQ- 806 node) can be calculated from the regular forward SPF done as part of 807 standard and remote LFA computations. However without the mechanism 808 proposed in section Section 2.3.2 of this document, there is no way 809 to determine the path characteristics for the second path segment 810 (i.e. from the PQ-node to the destination). In the absence of the 811 path characteristics for the second path segment, two Remote-LFA 812 alternate paths may be equally preferred based on the first path 813 segments characteristics only, although the second path segment 814 attributes may be different. 816 3.2. The Solution 818 The additional forward SPF computation proposed in Section 2.3.2 819 document shall also collect links, nodes and path characteristics 820 along the second path segment. This shall enable collection of 821 complete path characteristics for a given Remote-LFA alternate path 822 to a given destination. The complete alternate path characteristics 823 shall then facilitate more accurate alternate path selection while 824 running the alternate selection policy. 826 As already specified in Section 2.3.4 to limit the computational 827 overhead of the proposed approach, forward SPF computations MUST be 828 run on a selected subset from the entire set of PQ-nodes computed in 829 the network, with a finite limit on the number of PQ-nodes in the 830 subset. The detailed suggestion on how to select this subset is 831 specified in the same section. While this limits the number of 832 possible alternate paths provided to the alternate-selection policy, 833 this is needed to keep the computational complexity within affordable 834 limits. However if the alternate-selection policy is very 835 restrictive this may leave few destinations in the entire topology 836 without protection. Yet this limitation provides a necessary 837 tradeoff between extensive coverage and immense computational 838 overhead. 840 4. Acknowledgements 842 Many thanks to Bruno Decraene for providing his useful comments. We 843 would also like to thank Uma Chunduri for reviewing this document and 844 providing valuable feedback. Also, many thanks to Harish Raghuveer 845 for his review and comments on the initial versions of this document. 847 5. IANA Considerations 849 N/A. - No protocol changes are proposed in this document. 851 6. Security Considerations 853 This document does not introduce any change in any of the protocol 854 specifications. It simply proposes to run an extra SPF rooted on 855 each PQ-node discovered in the whole network. 857 7. References 859 7.1. Normative References 861 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 862 Requirement Levels", BCP 14, RFC 2119, 863 DOI 10.17487/RFC2119, March 1997, 864 . 866 [RFC5286] Atlas, A., Ed. and A. Zinin, Ed., "Basic Specification for 867 IP Fast Reroute: Loop-Free Alternates", RFC 5286, 868 DOI 10.17487/RFC5286, September 2008, 869 . 871 [RFC7490] Bryant, S., Filsfils, C., Previdi, S., Shand, M., and N. 872 So, "Remote Loop-Free Alternate (LFA) Fast Reroute (FRR)", 873 RFC 7490, DOI 10.17487/RFC7490, April 2015, 874 . 876 7.2. Informative References 878 [RFC7916] Litkowski, S., Ed., Decraene, B., Filsfils, C., Raza, K., 879 Horneffer, M., and P. Sarkar, "Operational Management of 880 Loop-Free Alternates", RFC 7916, DOI 10.17487/RFC7916, 881 July 2016, . 883 Authors' Addresses 885 Pushpasis Sarkar (editor) 886 Individual Contributor 888 Email: pushpasis.ietf@gmail.com 890 Shraddha Hegde 891 Juniper Networks, Inc. 892 Electra, Exora Business Park 893 Bangalore, KA 560103 894 India 896 Email: shraddha@juniper.net 898 Chris Bowers 899 Juniper Networks, Inc. 900 1194 N. Mathilda Ave. 901 Sunnyvale, CA 94089 902 US 904 Email: cbowers@juniper.net 906 Hannes Gredler 907 RtBrick, Inc. 909 Email: hannes@rtbrick.com 911 Stephane Litkowski 912 Orange 914 Email: stephane.litkowski@orange.com