idnits 2.17.1 draft-ietf-pce-state-sync-00.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 -- The document date (June 28, 2021) is 1031 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(See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Outdated reference: A later version (-14) exists of draft-knodel-terminology-05 -- Obsolete informational reference (is this intentional?): RFC 7525 (Obsoleted by RFC 9325) -- Obsolete informational reference (is this intentional?): RFC 7752 (Obsoleted by RFC 9552) Summary: 0 errors (**), 0 flaws (~~), 2 warnings (==), 3 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 PCE Working Group S. Litkowski 3 Internet-Draft Cisco 4 Intended status: Standards Track S. Sivabalan 5 Expires: December 30, 2021 Ciena Corporation 6 C. Li 7 H. Zheng 8 Huawei Technologies 9 June 28, 2021 11 Inter Stateful Path Computation Element (PCE) Communication Procedures. 12 draft-ietf-pce-state-sync-00 14 Abstract 16 The Path Computation Element Communication Protocol (PCEP) provides 17 mechanisms for Path Computation Elements (PCEs) to perform path 18 computation in response to a Path Computation Client (PCC) request. 19 The Stateful PCE extensions allow stateful control of Multi-Protocol 20 Label Switching (MPLS) Traffic Engineering (TE) Label Switched Paths 21 (LSPs) using PCEP. 23 A Path Computation Client (PCC) can synchronize an LSP state 24 information to a Stateful Path Computation Element (PCE). The 25 stateful PCE extension allows a redundancy scenario where a PCC can 26 have redundant PCEP sessions towards multiple PCEs. In such a case, 27 a PCC gives control of a LSP to only a single PCE, and only one PCE 28 is responsible for path computation for this delegated LSP. 30 There are some use cases, where an inter-PCE stateful communication 31 can bring additional resiliency in the design, for instance when some 32 PCC-PCE session fails. The inter-PCE stateful communication may also 33 provide a faster update of the LSP states when such an event occurs. 34 Finally, when, in a redundant PCE scenario, there is a need to 35 compute a set of paths that are part of a group (so there is a 36 dependency between the paths), there may be some cases where the 37 computation of all paths in the group is not handled by the same PCE: 38 this situation is called a split-brain. This split-brain scenario 39 may lead to computation loops between PCEs or suboptimal path 40 computation. 42 This document describes the procedures to allow a stateful 43 communication between PCEs for various use-cases and also the 44 procedures to prevent computations loops. 46 Requirements Language 48 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 49 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 50 "OPTIONAL" in this document are to be interpreted as described in BCP 51 14 [RFC2119] [RFC8174] when, and only when, they appear in all 52 capitals, as shown here. 54 Status of This Memo 56 This Internet-Draft is submitted in full conformance with the 57 provisions of BCP 78 and BCP 79. 59 Internet-Drafts are working documents of the Internet Engineering 60 Task Force (IETF). Note that other groups may also distribute 61 working documents as Internet-Drafts. The list of current Internet- 62 Drafts is at https://datatracker.ietf.org/drafts/current/. 64 Internet-Drafts are draft documents valid for a maximum of six months 65 and may be updated, replaced, or obsoleted by other documents at any 66 time. It is inappropriate to use Internet-Drafts as reference 67 material or to cite them other than as "work in progress." 69 This Internet-Draft will expire on December 30, 2021. 71 Copyright Notice 73 Copyright (c) 2021 IETF Trust and the persons identified as the 74 document authors. All rights reserved. 76 This document is subject to BCP 78 and the IETF Trust's Legal 77 Provisions Relating to IETF Documents 78 (https://trustee.ietf.org/license-info) in effect on the date of 79 publication of this document. Please review these documents 80 carefully, as they describe your rights and restrictions with respect 81 to this document. Code Components extracted from this document must 82 include Simplified BSD License text as described in Section 4.e of 83 the Trust Legal Provisions and are provided without warranty as 84 described in the Simplified BSD License. 86 Table of Contents 88 1. Introduction and Problem Statement . . . . . . . . . . . . . 3 89 1.1. Reporting LSP Changes . . . . . . . . . . . . . . . . . . 4 90 1.2. Split-Brain . . . . . . . . . . . . . . . . . . . . . . . 5 91 1.3. Applicability to H-PCE . . . . . . . . . . . . . . . . . 12 92 2. Proposed solution . . . . . . . . . . . . . . . . . . . . . . 12 93 2.1. State-sync session . . . . . . . . . . . . . . . . . . . 12 94 2.2. Primary/Secondary relationship between PCE . . . . . . . 14 95 3. Procedures and Protocol Extensions . . . . . . . . . . . . . 14 96 3.1. Opening a state-sync session . . . . . . . . . . . . . . 14 97 3.1.1. Capability Advertisement . . . . . . . . . . . . . . 14 98 3.2. State synchronization . . . . . . . . . . . . . . . . . . 15 99 3.3. Incremental updates and report forwarding rules . . . . . 16 100 3.4. Maintaining LSP states from different sources . . . . . . 17 101 3.5. Computation priority between PCEs and sub-delegation . . 18 102 3.6. Passive stateful procedures . . . . . . . . . . . . . . . 19 103 3.7. PCE initiation procedures . . . . . . . . . . . . . . . . 20 104 4. Examples . . . . . . . . . . . . . . . . . . . . . . . . . . 20 105 4.1. Example 1 . . . . . . . . . . . . . . . . . . . . . . . . 20 106 4.2. Example 2 . . . . . . . . . . . . . . . . . . . . . . . . 22 107 4.3. Example 3 . . . . . . . . . . . . . . . . . . . . . . . . 24 108 5. Using Primary/Secondary Computation and State-sync Sessions 109 to increase Scaling . . . . . . . . . . . . . . . . . . . . . 25 110 6. PCEP-PATH-VECTOR TLV . . . . . . . . . . . . . . . . . . . . 27 111 7. Security Considerations . . . . . . . . . . . . . . . . . . . 28 112 8. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 28 113 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 28 114 9.1. PCEP-Error Object . . . . . . . . . . . . . . . . . . . . 28 115 9.2. PCEP TLV Type Indicators . . . . . . . . . . . . . . . . 29 116 9.3. STATEFUL-PCE-CAPABILITY TLV . . . . . . . . . . . . . . . 29 117 10. References . . . . . . . . . . . . . . . . . . . . . . . . . 29 118 10.1. Normative References . . . . . . . . . . . . . . . . . . 29 119 10.2. Informative References . . . . . . . . . . . . . . . . . 30 120 Appendix A. Contributors . . . . . . . . . . . . . . . . . . . . 31 121 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 31 123 1. Introduction and Problem Statement 125 The Path Computation Element communication Protocol (PCEP) [RFC5440] 126 provides mechanisms for Path Computation Elements (PCEs) to perform 127 path computations in response to Path Computation Clients' (PCCs) 128 requests. 130 A stateful PCE [RFC8231] is capable of considering, for the purposes 131 of path computation, not only the network state in terms of links and 132 nodes (referred to as the Traffic Engineering Database or TED) but 133 also the status of active services (previously computed paths, and 134 currently reserved resources, stored in the Label Switched Paths 135 Database (LSP-DB). 137 [RFC8051] describes general considerations for a stateful PCE 138 deployment and examines its applicability and benefits, as well as 139 its challenges and limitations through a number of use cases. 141 The examples in this section are for illustrative purpose to showcase 142 the need for inter-PCE stateful PCEP sessions. 144 1.1. Reporting LSP Changes 146 When using a stateful PCE ([RFC8231]), a PCC can synchronize an LSP 147 state information to the stateful PCE. If the PCC grants the control 148 of the LSP to the PCE (called delegation [RFC8231]), the PCE can 149 update the LSP parameters at any time. 151 In a multi PCE deployment (redundancy, loadbalancing...), with the 152 current specification defined in [RFC8231], when a PCE makes an 153 update, it is the PCC that is in charge of reporting the LSP status 154 to all PCEs with LSP parameter change which brings additional hops 155 and delays in notifying the overall network of the LSP parameter 156 change. 158 This delay may affect the reaction time of the other PCEs if they 159 need to take action after being notified of the LSP parameter change. 161 Apart from the synchronization from the PCC, it is also useful if 162 there is a synchronization mechanism between the stateful PCEs. As 163 stateful PCE make changes to its delegated LSPs, these changes 164 (pending LSPs and the sticky resources [RFC7399]) can be synchronized 165 immediately to the other PCEs. 167 +----------+ 168 | PCC1 | LSP1 169 +----------+ 170 / \ 171 / \ 172 +---------+ +---------+ 173 | PCE1 | | PCE2 | 174 +---------+ +---------+ 175 \ / 176 \ / 177 +----------+ 178 | PCC2 | LSP2 179 +----------+ 181 In the figure above, we consider a load-balanced PCE architecture, so 182 PCE1 is responsible to compute paths for PCC1 and PCE2 is responsible 183 to compute paths for PCC2. When PCE1 triggers an LSP update for 184 LSP1, it sends a PCUpd message to PCC1 containing the new parameters 185 for LSP1. PCC1 will take the parameters into account and will send a 186 PCRpt message to PCE1 and PCE2 reflecting the changes. PCE2 will so 187 be notified of the change only after receiving the PCRpt message from 188 PCC1. 190 Let's consider that the LSP1 parameters changed in such a way that 191 LSP1 will take over resources from LSP2 with a higher priority. 192 After receiving the report from PCC1, PCE2 will therefore try to find 193 a new path for LSP2. If we consider that there is a round trip delay 194 of about 150 milliseconds (ms) between the PCEs and PCC1 and a round 195 trip delay of 10 ms between the two PCEs if will take more than 150 196 ms for PCE2 to be notified of the change. 198 Adding a PCEP session between PCE1 and PCE2 may allow to reduce the 199 synchronization time, so PCE2 can react more quickly by taking the 200 pending LSPs and attached resources into account during path 201 computation and re-optimization. 203 1.2. Split-Brain 205 In a resiliency case, a PCC has redundant PCEP sessions towards 206 multiple PCEs. In such a case, a PCC gives control on an LSP to a 207 single PCE only, and only this PCE is responsible for the path 208 computation for the delegated LSP: the PCC achieves this by setting 209 the D flag only towards the active PCE [RFC8231] selected for 210 delegation. The election of the active PCE to delegate an LSP is 211 controlled by each PCC. The PCC usually elects the active PCE by a 212 local configured policy (by setting a priority). Upon PCEP session 213 failure, or active PCE failure, PCC may decide to elect a new active 214 PCE by sending new PCRpt message with D flag set to this new active 215 PCE. When the failed PCE or PCEP session comes back online, it will 216 be up to the implementation to do preemption. Doing preemption may 217 lead to some disruption on the existing path if path results from 218 both PCEs are not exactly the same. By considering a network with 219 multiple PCCs and implementing multiple stateful PCEs for redundancy 220 purpose, there is no guarantee that at any time all the PCCs delegate 221 their LSPs to the same PCE. 223 +----------+ 224 | PCC1 | LSP1 225 +----------+ 226 / \ 227 / \ 228 +---------+ +---------+ 229 | PCE1 | | PCE2 | 230 +---------+ +---------+ 231 \ / 232 *fail* \ / 233 +----------+ 234 | PCC2 | LSP2 235 +----------+ 237 In the example above, we consider that by configuration, both PCCs 238 will firstly delegate their LSPs to PCE1. So, PCE1 is responsible 239 for computing a path for both LSP1 and LSP2. If the PCEP session 240 between PCC2 and PCE1 fails, PCC2 will delegate LSP2 to PCE2. So 241 PCE1 becomes responsible only for LSP1 path computation while PCE2 is 242 responsible for the path computation of LSP2. When the PCC2-PCE1 243 session is back online, PCC2 will keep using PCE2 as active PCE 244 (consider no preemption in this example). So the result is a 245 permanent situation where each PCE is responsible for a subset of 246 path computation. 248 This situation is called a split-brain scenario, as there are 249 multiple computation brains running at the same time while a central 250 computation unit was required in some deployments/use cases. 252 Further, there are use cases where a particular LSP path computation 253 is linked to another LSP path computation: the most common use case 254 is path disjointness (see [RFC8800]). The set of LSPs that are 255 dependent to each other may start from a different head-end. 257 _________________________________________ 258 / \ 259 / +------+ +------+ \ 260 | | PCE1 | | PCE2 | | 261 | +------+ +------+ | 262 | | 263 | +------+ +------+ | 264 | | PCC1 | ----------------------> | PCC2 | | 265 | +------+ +------+ | 266 | | 267 | | 268 | +------+ +------+ | 269 | | PCC3 | ----------------------> | PCC4 | | 270 | +------+ +------+ | 271 | | 272 \ / 273 \_________________________________________/ 275 _________________________________________ 276 / \ 277 / +------+ +------+ \ 278 | | PCE1 | | PCE2 | | 279 | +------+ +------+ | 280 | | 281 | +------+ 10 +------+ | 282 | | PCC1 | ----- R1 ---- R2 ------- | PCC2 | | 283 | +------+ | | +------+ | 284 | | | | 285 | | | | 286 | +------+ | | +------+ | 287 | | PCC3 | ----- R3 ---- R4 ------- | PCC4 | | 288 | +------+ +------+ | 289 | | 290 \ / 291 \_________________________________________/ 293 In the figure above, the requirement is to create two link-disjoint 294 LSPs: PCC1->PCC2 and PCC3->PCC4. In the topology, all links cost 295 metric is set to 1 except for the link 'R1-R2' which has a metric of 296 10. The PCEs are responsible for the path computation and PCE1 is 297 the active primary PCE for all PCCs in the nominal case. 299 Scenario 1: 301 In the normal case (PCE1 as active primary PCE), consider that 302 PCC1->PCC2 LSP is configured first with the link disjointness 303 constraint, PCE1 sends a PCUpd message to PCC1 with the ERO: 304 R1->R3->R4->R2->PCC2 (shortest path). PCC1 signals and installs the 305 path. When PCC3->PCC4 is configured, the PCEs already knows the path 306 of PCC1->PCC2 and can compute a link-disjoint path: the solution 307 requires to move PCC1->PCC2 onto a new path to let room for the new 308 LSP. PCE1 sends a PCUpd message to PCC1 with the new ERO: 309 R1->R2->PCC2 and a PCUpd to PCC3 with the following ERO: 310 R3->R4->PCC4. In the normal case, there is no issue for PCE1 to 311 compute a link-disjoint path. 313 Scenario 2: 315 Consider that PCC1 lost its PCEP session with PCE1 (all other PCEP 316 sessions are UP). PCC1 delegates its LSP to PCE2. 318 +----------+ 319 | PCC1 | LSP: PCC1->PCC2 320 +----------+ 321 \ 322 \ D=1 323 +---------+ +---------+ 324 | PCE1 | | PCE2 | 325 +---------+ +---------+ 326 D=1 \ / D=0 327 \ / 328 +----------+ 329 | PCC3 | LSP: PCC3->PCC4 330 +----------+ 332 Consider that the PCC1->PCC2 LSP is configured first with the link 333 disjointness constraint, PCE2 (which is the new active primary PCE 334 for PCC1) sends a PCUpd message to PCC1 with the ERO: 335 R1->R3->R4->R2->PCC2 (shortest path). When PCC3->PCC4 is configured, 336 PCE1 is not aware of LSPs from PCC1 any more, so it cannot compute a 337 disjoint path for PCC3->PCC4 and will send a PCUpd message to PCC3 338 with the shortest path ERO: R3->R4->PCC4. When PCC3->PCC4 LSP will 339 be reported to PCE2 by PCC3, PCE2 will ensure disjointness 340 computation and will correctly move PCC1->PCC2 (as it owns delegation 341 for this LSP) on the following path: R1->R2->PCC2. With this 342 sequence of event and these PCEP sessions, disjointness is ensured. 344 Scenario 3: 346 +----------+ 347 | PCC1 | LSP: PCC1->PCC2 348 +----------+ 349 / \ 350 D=1 / \ D=0 351 +---------+ +---------+ 352 | PCE1 | | PCE2 | 353 +---------+ +---------+ 354 / D=1 355 / 356 +----------+ 357 | PCC3 | LSP: PCC3->PCC4 358 +----------+ 360 Consider the above PCEP sessions and the PCC1->PCC2 LSP is configured 361 first with the link disjointness constraint, PCE1 computes the 362 shortest path as it is the only LSP in the disjoint association group 363 that it is aware of: R1->R3->R4->R2->PCC2 (shortest path). When 364 PCC3->PCC4 is configured, PCE2 must compute a disjoint path for this 365 LSP. The only solution found is to move PCC1->PCC2 LSP on another 366 path, but PCE2 cannot do it as it does not have delegation for this 367 LSP. In this set-up, PCEs are not able to find a disjoint path. 369 Scenario 4: 371 +----------+ 372 | PCC1 | LSP: PCC1->PCC2 373 +----------+ 374 / \ 375 D=1 / \ D=0 376 +---------+ +---------+ 377 | PCE1 | | PCE2 | 378 +---------+ +---------+ 379 D=0 \ / D=1 380 \ / 381 +----------+ 382 | PCC3 | LSP: PCC3->PCC4 383 +----------+ 385 Consider the above PCEP sessions and that PCEs are configured to 386 fall-back to the shortest path if disjointness cannot be found as 387 described in [RFC8800]. The PCC1->PCC2 LSP is configured first, PCE1 388 computes the shortest path as it is the only LSP in the disjoint 389 association group that it is aware of: R1->R3->R4->R2->PCC2 (shortest 390 path). When PCC3->PCC4 is configured, PCE2 must compute a disjoint 391 path for this LSP. The only solution found is to move PCC1->PCC2 LSP 392 on another path, but PCE2 cannot do it as it does not have delegation 393 for this LSP. PCE2 then provides the shortest path for PCC3->PCC4: 394 R3->R4->PCC4. When PCC3 receives the ERO, it reports it back to both 395 PCEs. When PCE1 becomes aware of the PCC3->PCC4 path, it recomputes 396 the constrained shortest path first (CSPF) algorithm and provides a 397 new path for PCC1->PCC2: R1->R2->PCC2. The new path is reported back 398 to all PCEs by PCC1. PCE2 recomputes also CSPF to take into account 399 the new reported path. The new computation does not lead to any path 400 update. 402 Scenario 5: 404 _____________________________________ 405 / \ 406 / +------+ +------+ \ 407 | | PCE1 | | PCE2 | | 408 | +------+ +------+ | 409 | | 410 | +------+ 100 +------+ | 411 | | | -------------------- | | | 412 | | PCC1 | ----- R1 ----------- | PCC2 | | 413 | +------+ | +------+ | 414 | | | | | 415 | 6 | | 2 | 2 | 416 | | | | | 417 | +------+ | +------+ | 418 | | PCC3 | ----- R3 ----------- | PCC4 | | 419 | +------+ 10 +------+ | 420 | | 421 \ / 422 \_____________________________________/ 424 Now, consider a new network topology with the same PCEP sessions as 425 the previous example. Suppose that both LSPs are configured almost 426 at the same time. PCE1 will compute a path for PCC1->PCC2 while PCE2 427 will compute a path for PCC3->PCC4. As each PCE is not aware of the 428 path of the second LSP in the association group (not reported yet), 429 each PCE is computing the shortest path for the LSP. PCE1 computes 430 ERO: R1->PCC2 for PCC1->PCC2 and PCE2 computes ERO: 431 R3->R1->PCC2->PCC4 for PCC3->PCC4. When these shortest paths will be 432 reported to each PCE. Each PCE will recompute disjointness. PCE1 433 will provide a new path for PCC1->PCC2 with ERO: PCC1->PCC2. PCE2 434 will provide also a new path for PCC3->PCC4 with ERO: R3->PCC4. When 435 those new paths will be reported to both PCEs, this will trigger CSPF 436 again. PCE1 will provide a new more optimal path for PCC1->PCC2 with 437 ERO: R1->PCC2 and PCE2 will also provide a more optimal path for 438 PCC3->PCC4 with ERO: R3->R1->PCC2->PCC4. So we come back to the 439 initial state. When those paths will be reported to both PCEs, this 440 will trigger CSPF again. An infinite loop of CSPF computation is 441 then happening with a permanent flap of paths because of the split- 442 brain situation. 444 This permanent computation loop comes from the inconsistency between 445 the state of the LSPs as seen by each PCE due to the split-brain: 446 each PCE is trying to modify at the same time its delegated path 447 based on the last received path information which de facto 448 invalidates this received path information. 450 Scenario 6: multi-domain 452 Domain/Area 1 Domain/Area 2 453 ________________ ________________ 454 / \ / \ 455 / +------+ | | +------+ \ 456 | | PCE1 | | | | PCE3 | | 457 | +------+ | | +------+ | 458 | | | | 459 | +------+ | | +------+ | 460 | | PCE2 | | | | PCE4 | | 461 | +------+ | | +------+ | 462 | | | | 463 | +------+ | | +------+ | 464 | | PCC1 | | | | PCC2 | | 465 | +------+ | | +------+ | 466 | | | | 467 | | | | 468 | +------+ | | +------+ | 469 | | PCC3 | | | | PCC4 | | 470 | +------+ | | +------+ | 471 \ | | | 472 \_______________/ \________________/ 474 In the example above, suppose that the disjoint LSPs from PCC1 to 475 PCC2 and from PCC4 to PCC3 are created. All the PCEs have the 476 knowledge of both domain topologies (e.g. using BGP-LS [RFC7752]). 477 For operation/management reasons, each domain uses its own group of 478 redundant PCEs. PCE1/PCE2 in domain 1 have PCEP sessions with PCC1 479 and PCC3 while PCE3/PCE4 in domain 2 have PCEP sessions with PCC2 and 480 PCC4. As PCE1/2 does not know about LSPs from PCC2/4 and PCE3/4 do 481 not know about LSPs from PCC1/3, there is no possibility to compute 482 the disjointness constraint. This scenario can also be seen as a 483 split-brain scenario. This multi-domain architecture (with multiple 484 groups of PCEs) can also be used in a single domain, where an 485 operator wants to limit the failure domain by creating multiple 486 groups of PCEs maintaining a subset of PCCs. As for the multi-domain 487 example, there will be no possibility to compute the disjoint path 488 starting from head-ends managed by different PCE groups. 490 In this document, we propose a solution that addresses the 491 possibility to compute LSP association based constraints (like 492 disjointness) in split-brain scenarios while preventing computation 493 loops. 495 1.3. Applicability to H-PCE 497 [RFC8751] describes general considerations and use cases for the 498 deployment of Stateful PCE(s) using the Hierarchical PCE [RFC6805] 499 architecture. In this architecture, there is a clear need to 500 communicate between a child stateful PCE and a parent stateful PCE. 501 The procedures and extensions as described in Section 3 are equally 502 applicable to the H-PCE scenario. 504 2. Proposed solution 506 Our solution is based on : 508 o The creation of the inter-PCE stateful PCEP session with specific 509 procedures. 511 o A Primary/Secondary relationship between PCEs. 513 2.1. State-sync session 515 This document proposes to set-up a PCEP session between the stateful 516 PCEs. Creating such a session is already authorized by multiple 517 scenarios like the one described in [RFC4655] (multiple PCEs that are 518 handling part of the path computation) and [RFC6805] (hierarchical 519 PCE) but was only focused on the stateless PCEP sessions. As 520 stateful PCE brings additional features (LSP state synchronization, 521 path update, delegation, ...), thus some new behaviors need to be 522 defined. 524 This inter-PCE PCEP session will allow the exchange of LSP states 525 between PCEs that would help some scenarios where PCEP sessions are 526 lost between PCC and PCE. This inter-PCE PCEP session is henceforth 527 called a state-sync session. 529 For example, in the scenario below, there is no possibility to 530 compute disjointness as there is no PCE that is aware of both LSPs. 532 +----------+ 533 | PCC1 | LSP: PCC1->PCC2 534 +----------+ 535 / 536 D=1 / 537 +---------+ +---------+ 538 | PCE1 | | PCE2 | 539 +---------+ +---------+ 540 / D=1 541 / 542 +----------+ 543 | PCC3 | LSP: PCC3->PCC4 544 +----------+ 546 If we add a state-sync session, PCE1 will be able to do state 547 synchronization via PCRpt messages for its LSP to PCE2 and PCE2 will 548 do the same. All the PCEs will be aware of all LSPs even if a 549 PCC->PCE session is down. PCEs will then be able to compute disjoint 550 paths. 552 +----------+ 553 | PCC1 | LSP : PCC1->PCC2 554 +----------+ 555 / 556 D=1 / 557 +---------+ PCEP +---------+ 558 | PCE1 | ----- | PCE2 | 559 +---------+ +---------+ 560 / D=1 561 / 562 +----------+ 563 | PCC3 | LSP : PCC3->PCC4 564 +----------+ 566 The procedures associated with this state-sync session are defined in 567 Section 3. 569 By just adding this state-sync session, it does not ensure that a 570 path with LSP association based constraints can always be computed 571 and does not prevent the computation loop, but it increases 572 resiliency and ensures that PCEs will have the state information for 573 all LSPs. Also, this session will allow for a PCE to update the 574 other PCEs providing a faster synchronization mechanism than relying 575 on PCCs only. 577 2.2. Primary/Secondary relationship between PCE 579 As seen in Section 1, performing a path computation in a split-brain 580 scenario (multiple PCEs responsible for computation) may provide a 581 non-optimal LSP placement, no path, or computation loops. To provide 582 the best efficiency, an LSP association constraint-based computation 583 requires that a single PCE performs the path computation for all LSPs 584 in the association group. Note that, it could be all LSPs belonging 585 to a particular association group, or all LSPs from a particular PCC, 586 or all LSPs in the network that need to be delegated to a single PCE 587 based on the deployment scenarios. 589 This document proposes to add a priority mechanism between PCEs to 590 elect a single computing PCE. Using this priority mechanism, PCEs 591 can agree on the PCE that will be responsible for the computation for 592 a particular association group, or set of LSPs. The priority could 593 be set per association, per PCC, or for all LSPs. How this priority 594 is set or advertised is out of the scope of this document. The rest 595 of the text considers the association group as an example. 597 When a single PCE is performing the computation for a particular 598 association group, no computation loop can happen and an optimal 599 placement will be provided. The other PCEs will only act as state 600 collectors and forwarders. 602 In the scenario described in Section 2.1, PCE1 and PCE2 will decide 603 that PCE1 will be responsible for the path computation of both LSPs. 604 If we first configure PCC1->PCC2, PCE1 computes the shortest path at 605 it is the only LSP in the disjoint-group that it is aware of: 606 R1->R3->R4->R2->PCC2 (shortest path). When PCC3->PCC4 is configured, 607 PCE2 will not perform computation even if it has delegation but 608 forwards the delegation via PCRpt message to PCE1 through the state- 609 sync session. PCE1 will then perform disjointness computation and 610 will move PCC1->PCC2 onto R1->R2->PCC2 and provides an ERO to PCE2 611 for PCC3->PCC4: R3->R4->PCC4. The PCE2 will further update the PCC3 612 with the new path. 614 3. Procedures and Protocol Extensions 616 3.1. Opening a state-sync session 618 3.1.1. Capability Advertisement 620 A PCE indicates its support of state-sync procedures during the PCEP 621 Initialization phase [RFC5440]. The OPEN object in the Open message 622 MUST contains the "Stateful PCE Capability" TLV defined in [RFC8231]. 623 A new P (INTER-PCE-CAPABILITY) flag is introduced to indicate the 624 support of state-sync. 626 This document adds a new bit in the Flags field with : 628 P (INTER-PCE-CAPABILITY - 1 bit - TBD4): If set to 1 by a PCEP 629 Speaker, the PCEP speaker indicates that the session MUST follow 630 the state-sync procedures as described in this document. The P 631 bit MUST be set by both speakers: if a PCEP Speaker receives a 632 STATEFUL-PCE-CAPABILITY TLV with P=0 while it advertised P=1 or if 633 both set P flag to 0, the session SHOULD be set-up but the state- 634 sync procedures MUST NOT be applied on this session. 636 The U flag [RFC8231] MUST be set when sending the STATEFUL-PCE- 637 CAPABILITY TLV with the P flag set. In case the U flag is not set 638 along with the P flag, the state sync capability is not enabled and 639 it is considered as if the P flag is not set. The S flag MAY be set 640 if optimized synchronization is required as per [RFC8232]. 642 3.2. State synchronization 644 When the state sync capability has been negotiated between stateful 645 PCEs, each PCEP speaker will behave as a PCE and as a PCC at the same 646 time regarding the state synchronization as defined in [RFC8231]. 647 This means that each PCEP Speaker: 649 o MUST send a PCRpt message towards its neighbor with S flag set for 650 each LSP in its LSP database learned from a PCC. (PCC role) 652 o MUST send the End Of Synchronization Marker towards its neighbor 653 when all LSPs have been reported. (PCC role) 655 o MUST wait for the LSP synchronization from its neighbor to end 656 (receiving an End Of Synchronization Marker). (PCE role) 658 The process of synchronization runs in parallel on each PCE (with no 659 defined order). 661 The optimized state synchronization procedures MAY be used, as 662 defined in [RFC8232]. 664 When a PCEP Speaker sends a PCRpt on a state-sync session, it MUST 665 add the SPEAKER-IDENTITY-TLV (defined in [RFC8232]) in the LSP 666 Object, the value used will refer to the 'owner' PCC of the LSP. If 667 a PCEP Speaker receives a PCRpt on a state-sync session without this 668 TLV, it MUST discard the PCRpt message and it MUST reply with a PCErr 669 message using error-type=6 (Mandatory Object missing) and error- 670 value=TBD1 (SPEAKER-IDENTITY-TLV missing). 672 3.3. Incremental updates and report forwarding rules 674 During the life of an LSP, its state may change (path, constraints, 675 operational state...) and a PCC will advertise a new PCRpt to the PCE 676 for each such change. 678 When propagating LSP state changes from a PCE to other PCEs, it is 679 mandatory to ensure that a PCE always uses the freshest state coming 680 from the PCC. 682 When a PCE receives a new PCRpt from a PCC with the LSP-DB-VERSION, 683 the PCE MUST forward the PCRpt to all its state-sync sessions and 684 MUST add the appropriate SPEAKER-IDENTITY-TLV in the PCRpt. In 685 addition, it MUST add a new ORIGINAL-LSP-DB-VERSION TLV (described 686 below). The ORIGINAL-LSP-DB-VERSION contains the LSP-DB-VERSION 687 coming from the PCC. 689 When a PCE receives a new PCRpt from a PCC without the LSP-DB- 690 VERSION, it SHOULD NOT forward the PCRpt on any state-sync sessions 691 and log such an event on the first occurrence. 693 When a PCE receives a new PCRpt from a PCC with the R flag (Remove) 694 set and an LSP-DB-VERSION TLV, the PCE MUST forward the PCRpt to all 695 its state-sync sessions keeping the R flag set (Remove) and MUST add 696 the appropriate SPEAKER-IDENTITY-TLV and ORIGINAL-LSP-DB-VERSION TLV 697 in the PCRpt message. 699 When a PCE receives a PCRpt from a state-sync session, it MUST NOT 700 forward the PCRpt to other state-sync sessions. This helps to 701 prevent message loops between PCEs. As a consequence, a full mesh of 702 PCEP sessions between PCEs are REQUIRED. 704 When a PCRpt is forwarded, all the original objects and values are 705 kept. As an example, the PLSP-ID used in the forwarded PCRpt will be 706 the same as the original one used by the PCC. Thus an implementation 707 supporting this document MUST consider SPEAKER-IDENTITY-TLV and PLSP- 708 ID together to uniquely identify an LSP on the state-sync session. 710 The ORIGINAL-LSP-DB-VERSION TLV is encoded as follows and MUST always 711 contain the LSP-DB-VERSION received from the owner PCC of the LSP: 713 0 1 2 3 714 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 715 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 716 | Type=TBD2 | Length=8 | 717 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 718 | LSP State DB Version Number | 719 | | 720 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 722 Using the ORIGINAL-LSP-DB-VERSION TLV allows a PCE to keep using 723 optimized synchronization ([RFC8232]) with another PCE. In such a 724 case, the PCE will send a PCRpt to another PCE with both ORIGINAL- 725 LSP-DB-VERSION TLV and LSP-DB-VERSION TLV. The ORIGINAL-LSP-DB- 726 VERSION TLV will contain the version number as allocated by the PCC 727 while the LSP-DB-VERSION will contain the version number allocated by 728 the local PCE. 730 3.4. Maintaining LSP states from different sources 732 When a PCE receives a PCRpt on a state-sync session, it stores the 733 LSP information into the original PCC address context (as the LSP 734 belongs to the PCC). A PCE SHOULD maintain a single state for a 735 particular LSP and SHOULD maintain the list of sources it learned a 736 particular state from. 738 A PCEP speaker may receive state information for a particular LSP 739 from different sources: the PCC that owns the LSP (through a regular 740 PCEP session) and some PCEs (through PCEP state-sync sessions). A 741 PCEP speaker MUST always keep the freshest state in its LSP database, 742 overriding the previously received information. 744 A PCE, receiving a PCRpt from a PCC, updates the state of the LSP in 745 its LSP-DB with the newly received information. When receiving a 746 PCRpt from another PCE, a PCE SHOULD update the LSP state only if the 747 ORIGINAL-LSP-DB-VERSION present in the PCRpt is greater than the 748 current ORIGINAL-LSP-DB-VERSION of the stored LSP state. This 749 ensures that a PCE never tries to update its stored LSP state with an 750 old information. Each time a PCE updates an LSP state in its LSP-DB, 751 it SHOULD reset the source list associated with the LSP state and 752 SHOULD add the source speaker address in the source list. When a PCE 753 receives a PCRpt which has an ORIGINAL-LSP-DB-VERSION (if coming from 754 a PCE) or an LSP-DB-VERSION (if coming from the PCC) equals to the 755 current ORIGINAL-LSP-DB-VERSION of the stored LSP state, it SHOULD 756 add the source speaker address in the source list. 758 When a PCE receives a PCRpt requesting an LSP deletion from a 759 particular source, it SHOULD remove this particular source from the 760 list of sources associated with this LSP. 762 When the list of sources becomes empty for a particular LSP, the LSP 763 state MUST be removed. This means that all the sources must send a 764 PCRpt with R=1 for an LSP to make the PCE remove the LSP state. 766 3.5. Computation priority between PCEs and sub-delegation 768 A computation priority is necessary to ensure that a single PCE will 769 perform the computation for all the LSPs in an association group: 770 this will allow for a more optimized LSP placement and will prevent 771 computation loops. 773 All PCEs in the network that are handling LSPs in a common LSP 774 association group SHOULD be aware of each other including the 775 computation priority of each PCE. Note that there is no need for PCC 776 to be aware of this. The computation priority is a number and the 777 PCE having the highest priority SHOULD be responsible for the 778 computation. If several PCEs have the same priority value, their IP 779 address SHOULD be used as a tie-breaker to provide a rank: the 780 highest IP address has more priority. How PCEs are aware of the 781 priority of each other is out of the scope of this document, but as 782 example learning priorities could be done through PCE discovery or 783 local configuration. 785 The definition of the priority could be global so the highest 786 priority PCE will handle all path computations or more granular, so a 787 PCE may have the highest priority for only a subset of LSPs or 788 association-groups. 790 A PCEP Speaker receiving a PCRpt from a PCC with the D flag set that 791 does not have the highest computation priority, SHOULD forward the 792 PCRpt on all state-sync sessions (as per Section 3.3) and SHOULD set 793 D flag on the state-sync session towards the highest priority PCE, D 794 flag will be unset to all other state-sync sessions. This behavior 795 is similar to the delegation behavior handled at the PCC side and is 796 called a sub-delegation (the PCE sub-delegates the control of the LSP 797 to another PCE). When a PCEP Speaker sub-delegates an LSP to another 798 PCE, it loose control of the LSP and cannot update it anymore by its 799 own decision. When a PCE receives a PCRpt with D flag set on a 800 state-sync session, as a regular PCE, it is granted control over the 801 LSP. 803 If the highest priority PCE is failing or if the state-sync session 804 between the local PCE and the highest priority PCE failed, the local 805 PCE MAY decide to delegate the LSP to the next highest priority PCE 806 or to take back control of the LSP. It is a local policy decision. 808 When a PCE has the delegation for an LSP and needs to update this 809 LSP, it MUST send a PCUpd message to all state-sync sessions and to 810 the PCC session on which it received the delegation. The D-Flag 811 would be unset in the PCUpd for state-sync sessions whereas the 812 D-Flag would be set for the PCC. In the case of sub-delegation, the 813 computing PCE will send the PCUpd only to all state-sync sessions (as 814 it has no direct delegation from a PCC). The D-Flag would be set for 815 the state-sync session to the PCE that sub-delegated this LSP and the 816 D-Flag would be unset for other state-sync sessions. 818 The PCUpd sent over a state-sync session MUST contain the SPEAKER- 819 IDENTITY-TLV in the LSP Object (the value used must identify the 820 target PCC). The PLSP-ID used is the original PLSP-ID generated by 821 the PCC and learned from the forwarded PCRpt. If a PCE receives a 822 PCUpd on a state-sync session without the SPEAKER-IDENTITY-TLV, it 823 MUST discard the PCUpd and MUST reply with a PCErr message using 824 error-type=6 (Mandatory Object missing) and error-value=TBD1 825 (SPEAKER-IDENTITY-TLV missing). 827 When a PCE receives a valid PCUpd on a state-sync session, it SHOULD 828 forward the PCUpd to the appropriate PCC (identified based on the 829 SPEAKER-IDENTITY-TLV value) that delegated the LSP originally and 830 SHOULD remove the SPEAKER-IDENTITY-TLV from the LSP Object. The 831 acknowledgment of the PCUpd is done through a cascaded mechanism, and 832 the PCC is the only responsible for triggering the acknowledgment: 833 when the PCC receives the PCUpd from the local PCE, it acknowledges 834 it with a PCRpt as per [RFC8231]. When receiving the new PCRpt from 835 the PCC, the local PCE uses the defined forwarding rules on the 836 state-sync session so the acknowledgment is relayed to the computing 837 PCE. 839 A PCE SHOULD NOT compute a path using an association-group constraint 840 if it has delegation for only a subset of LSPs in the group. In this 841 case, an implementation MAY use a local policy on PCE to decide if 842 PCE does not compute path at all for this set of LSP or if it can 843 compute a path by relaxing the association-group constraint. 845 3.6. Passive stateful procedures 847 In the passive stateful PCE architecture, the PCC is responsible for 848 triggering a path computation request using a PCReq message to its 849 PCE. Similarly to PCRpt Message, which remains unchanged for passive 850 mode, if a PCE receives a PCReq for an LSP and if this PCE finds that 851 it does not have the highest computation priority of this LSP, or 852 groups..., it MUST forward the PCReq message to the highest priority 853 PCE over the state-sync session. When the highest priority PCE 854 receives the PCReq, it computes the path and generates a PCRep 855 message towards the PCE that made the request. This PCE will then 856 forward the PCRep to the requesting PCC. The handling of LSP object 857 and the SPEAKER-IDENTITY-TLV in PCReq and PCRep is similar to PCRpt/ 858 PCUpd messages. 860 3.7. PCE initiation procedures 862 It is possible that a PCE does not have a PCEP session with the 863 headend to initiate a LSP as per [RFC8281]. A PCE could send the 864 PCInitiate message on the state-sync sessions to other PCE to request 865 it to create a PCE-Initiated LSP on its behalf. If the PCE is able 866 to initiate the LSP it would report it on the state-sync session via 867 PCRpt message. If the PCE does not have a session to the headend, it 868 MUST send a PCErr message with Error-type=24 (PCE instantiation 869 error) and Error-value=TBD5 (No PCEP session with the headend). PCE 870 could try to initiate via another state-sync PCE if available. 872 4. Examples 874 The examples in this section are for illustrative purpose to show how 875 the behavior of the state sync inter-PCE sessions. 877 4.1. Example 1 878 _________________________________________ 879 / \ 880 / +------+ +------+ \ 881 | | PCE1 | | PCE2 | | 882 | +------+ +------+ | 883 | | 884 | +------+ 10 +------+ | 885 | | PCC1 | ----- R1 ---- R2 ------- | PCC2 | | 886 | +------+ | | +------+ | 887 | | | | 888 | | | | 889 | +------+ | | +------+ | 890 | | PCC3 | ----- R3 ---- R4 ------- | PCC4 | | 891 | +------+ +------+ | 892 | | 893 \ / 894 \_________________________________________/ 896 +----------+ 897 | PCC1 | LSP : PCC1->PCC2 898 +----------+ 899 / 900 D=1 / 901 +---------+ +---------+ 902 | PCE1 |----| PCE2 | 903 +---------+ +---------+ 904 / D=1 905 / 906 +----------+ 907 | PCC3 | LSP : PCC3->PCC4 908 +----------+ 910 PCE1 computation priority 100 911 PCE2 computation priority 200 913 Consider the PCEP sessions as shown above, where computation priority 914 is global for all the LSPs and link disjoint between LSPs PCC1->PCC2 915 and PCC3->PCC4 is required. 917 Consider the PCC1->PCC2 is configured first and PCC1 delegates the 918 LSP to PCE1, but as PCE1 does not have the highest computation 919 priority, it sub-delegates the LSP to PCE2 by sending a PCRpt with 920 D=1 and including the SPEAKER-IDENTITY-TLV over the state-sync 921 session. PCE2 receives the PCRpt and as it has delegation for this 922 LSP, it computes the shortest path: R1->R3->R4->R2->PCC2. It then 923 sends a PCUpd to PCE1 (including the SPEAKER-IDENTITY-TLV) with the 924 computed ERO. PCE1 forwards the PCUpd to PCC1 (removing the SPEAKER- 925 IDENTITY-TLV). PCC1 acknowledges the PCUpd by a PCRpt to PCE1. PCE1 926 forwards the PCRpt to PCE2. 928 When PCC3->PCC4 is configured, PCC3 delegates the LSP to PCE2, PCE2 929 can compute a disjoint path as it has knowledge of both LSPs and has 930 delegation also for both. The only solution found is to move 931 PCC1->PCC2 LSP on another path, PCE2 can move PCC1->PCC2 as it has 932 sub-delegation for it. It creates a new PCUpd with a new ERO: 933 R1->R2-PCC2 towards PCE1 which forwards to PCC1. PCE2 sends a PCUpd 934 to PCC3 with the path: R3->R4->PCC4. 936 In this set-up, PCEs are able to find a disjoint path while without 937 state-sync and computation priority they could not. 939 4.2. Example 2 940 _____________________________________ 941 / \ 942 / +------+ +------+ \ 943 | | PCE1 | | PCE2 | | 944 | +------+ +------+ | 945 | | 946 | +------+ 100 +------+ | 947 | | | -------------------- | | | 948 | | PCC1 | ----- R1 ----------- | PCC2 | | 949 | +------+ | +------+ | 950 | | | | | 951 | 6 | | 2 | 2 | 952 | | | | | 953 | +------+ | +------+ | 954 | | PCC3 | ----- R3 ----------- | PCC4 | | 955 | +------+ 10 +------+ | 956 | | 957 \ / 958 \_____________________________________/ 960 +----------+ 961 | PCC1 | LSP : PCC1->PCC2 962 +----------+ 963 / \ 964 D=1 / \ D=0 965 +---------+ +---------+ 966 | PCE1 |----| PCE2 | 967 +---------+ +---------+ 968 D=0 \ / D=1 969 \ / 970 +----------+ 971 | PCC3 | LSP : PCC3->PCC4 972 +----------+ 974 PCE1 computation priority 200 975 PCE2 computation priority 100 977 In this example, suppose both LSPs are configured almost at the same 978 time. PCE1 sub-delegates PCC1->PCC2 to PCE2 while PCE2 keeps 979 delegation for PCC3->PCC4, PCE2 computes a path for PCC1->PCC2 and 980 PCC3->PCC4 and can achieve disjointness computation easily. No 981 computation loop happens in this case. 983 4.3. Example 3 985 _________________________________________ 986 / \ 987 / +------+ +------+ \ 988 | | PCE1 | | PCE2 | | 989 | +------+ +------+ | 990 | | 991 | +------+ 10 +------+ | 992 | | PCC1 | ----- R1 ---- R2 ------- | PCC2 | | 993 | +------+ | | +------+ | 994 | | | | 995 | | | | 996 | +------+ | | +------+ | 997 | | PCC3 | ----- R3 ---- R4 ------- | PCC4 | | 998 | +------+ +------+ | 999 | | 1000 \ / 1001 \_________________________________________/ 1003 +----------+ 1004 | PCC1 | LSP : PCC1->PCC2 1005 +----------+ 1006 / 1007 D=1 / 1008 +---------+ +---------+ +---------+ 1009 | PCE1 |----| PCE2 |----| PCE3 | 1010 +---------+ +---------+ +---------+ 1011 / D=1 1012 / 1013 +----------+ 1014 | PCC3 | LSP : PCC3->PCC4 1015 +----------+ 1017 PCE1 computation priority 100 1018 PCE2 computation priority 200 1019 PCE3 computation priority 300 1021 With the PCEP sessions as shown above, consider the need to have link 1022 disjoint LSPs PCC1->PCC2 and PCC3->PCC4. 1024 Suppose PCC1->PCC2 is configured first, PCC1 delegates the LSP to 1025 PCE1, but as PCE1 does not have the highest computation priority, it 1026 will sub-delegate the LSP to PCE2 (as it not aware of PCE3 and has no 1027 way to reach it). PCE2 cannot compute a path for PCC1->PCC2 as it 1028 does not have the highest priority and is not allowed to sub-delegate 1029 the LSP again towards PCE3 as per Section 3. 1031 When PCC3->PCC4 is configured, PCC3 delegates the LSP to PCE2 that 1032 performs sub-delegation to PCE3. As PCE3 will have knowledge of only 1033 one LSP in the group, it cannot compute disjointness and can decide 1034 to fall-back to a less constrained computation to provide a path for 1035 PCC3->PCC4. In this case, it will send a PCUpd to PCE2 that will be 1036 forwarded to PCC3. 1038 Disjointness cannot be achieved in this scenario because of lack of 1039 state-sync session between PCE1 and PCE3, but no computation loop 1040 happens. Thus it is advised for all PCEs that support state-sync to 1041 have a full mesh sessions between each other. 1043 5. Using Primary/Secondary Computation and State-sync Sessions to 1044 increase Scaling 1046 The Primary/Secondary computation and state-sync sessions 1047 architecture can be used to increase the scaling of the PCE 1048 architecture. If the number of PCCs is really high, it may be too 1049 resource consuming for a single PCE to maintain all the PCEP sessions 1050 while at the same time performing all path computations. Using 1051 primary/secondary computation and state-sync sessions may allow to 1052 create groups of PCEs that manage a subset of the PCCs and perform 1053 some or no path computations. Decoupling PCEP session maintenance 1054 and computation will allow increasing scaling of the PCE 1055 architecture. 1057 +----------+ 1058 | PCC500 | 1059 +----------+-+ 1060 | PCC1 | 1061 +----------+ 1062 / \ 1063 / \ 1064 +---------+ +---------+ 1065 | PCE1 |---| PCE2 | 1066 +---------+ +---------+ 1067 | \ / | 1068 | \/ | 1069 | /\ | 1070 | / \ | 1071 +---------+ +---------+ 1072 | PCE3 |---| PCE4 | 1073 +---------+ +---------+ 1074 \ / 1075 \ / 1076 +----------+ 1077 | PCC501 | 1078 +----------+-+ 1079 | PCC1000 | 1080 +----------+ 1082 In the figure above, two groups of PCEs are created: PCE1/2 maintain 1083 PCEP sessions with PCC1 up to PCC500, while PCE3/4 maintain PCEP 1084 sessions with PCC501 up to PCC1000. A granular primary/secondary 1085 policy is set-up as follows to load-share computation between PCEs: 1087 o PCE1 has priority 200 for association ID 1 up to 300, association 1088 source 0.0.0.0. All other PCEs have a decreasing priority for 1089 those associations. 1091 o PCE3 has priority 200 for association ID 301 up to 500, 1092 association source 0.0.0.0. All other PCEs have a decreasing 1093 priority for those associations. 1095 If some PCCs delegate LSPs with association ID 1 up to 300 and 1096 association source 0.0.0.0, the receiving PCE (if not PCE1) will sub- 1097 delegate the LSPs to PCE1. PCE1 becomes responsible for the 1098 computation of these LSP associations while PCE3 is responsible for 1099 the computation of another set of associations. 1101 The procedures described in this document could help greatly in load- 1102 sharing between a group of stateful PCEs. 1104 6. PCEP-PATH-VECTOR TLV 1106 This document allows PCEP messages to be propagated among PCEP 1107 speaker. It may be useful to track information about the propagation 1108 of the messages. One of the use cases is a message loop detection 1109 mechanism, but other use cases like hop by hop information recording 1110 may also be implemented. 1112 This document introduces the PCEP-PATH-VECTOR TLV (type TBD3) with 1113 the following format: 1115 0 1 2 3 1116 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 1117 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1118 | Type=TBD3 | Length | 1119 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1120 | PCEP-SPEAKER-INFORMATION#1 | 1121 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1122 | ... | 1123 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1124 | ... | 1125 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1126 | PCEP-SPEAKER-INFORMATION#n | 1127 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1129 The TLV format and padding rules are as per [RFC5440]. 1131 The PCEP-SPEAKER-INFORMATION field has the following format: 1133 0 1 2 3 1134 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 1135 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1136 | Length | ID Length | 1137 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1138 // Speaker Entity identity (variable) // 1139 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1140 // SubTLVs (optional) // 1141 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1143 Length: defines the total length of the PCEP-SPEAKER-INFORMATION 1144 field. 1146 ID Length: defines the length of the Speaker identity actual field 1147 (non-padded). 1149 Speaker Entity identity: same possible values as the SPEAKER- 1150 IDENTIFIER-TLV. Padded with trailing zeros to a 4-byte boundary. 1152 The PCEP-SPEAKER-INFORMATION may also carry some optional subTLVs 1153 so each PCEP speaker can add local information that could be 1154 recorded. This document does not define any sub-TLV. 1156 The PCEP-PATH-VECTOR TLV MAY be carried in the LSP Object. Its usage 1157 is purely optional. 1159 The list of speakers within the PCEP-PATH-VECTOR TLV MUST be ordered. 1160 When sending a PCEP message (PCRpt, PCUpd, or PCInitiate), a PCEP 1161 Speaker MAY add the PCEP-PATH-VECTOR TLV with a PCEP-SPEAKER- 1162 INFORMATION containing its own information. If the PCEP message sent 1163 is the result of a previously received PCEP message, and if the PCEP- 1164 PATH-VECTOR TLV was already present in the initial message, the PCEP 1165 speaker MAY append a new PCEP-SPEAKER-INFORMATION containing its own 1166 information. 1168 7. Security Considerations 1170 The security considerations described in [RFC8231] and [RFC5440] 1171 apply to the extensions described in this document as well. 1172 Additional considerations related to state synchronization and sub- 1173 delegation between stateful PCEs are introduced, as it could be 1174 spoofed and could be used as an attack vector. An attacker could 1175 attempt to create too much state in an attempt to load the PCEP peer. 1176 The PCEP peer responds with a PCErr message as described in 1177 [RFC8231]. An attacker could impact LSP operations by creating bogus 1178 state. Further, state synchronization between stateful PCEs could 1179 provide an adversary with the opportunity to eavesdrop on the 1180 network. Thus, securing the PCEP session using Transport Layer 1181 Security (TLS) [RFC8253], as per the recommendations and best current 1182 practices in [RFC7525], is RECOMMENDED. 1184 8. Acknowledgements 1186 Thanks to [I-D.knodel-terminology] urging for better use of terms. 1188 9. IANA Considerations 1190 This document requests IANA actions to allocate code points for the 1191 protocol elements defined in this document. 1193 9.1. PCEP-Error Object 1195 IANA is requested to allocate a new Error Value for the Error Type 9. 1197 Error-Type Meaning Reference 1198 6 Mandatory Object Missing [RFC5440] 1199 Error-value=TBD1: SPEAKER-IDENTITY-TLV This 1200 missing document 1201 24 LSP instantiation error [RFC8281] 1202 Error-value=TBD5: No PCEP session with the This 1203 headend document 1205 9.2. PCEP TLV Type Indicators 1207 IANA is requested to allocate new TLV Type Indicator values within 1208 the "PCEP TLV Type Indicators" sub-registry of the PCEP Numbers 1209 registry, as follows: 1211 Value Meaning Reference 1212 TBD2 ORIGINAL-LSP-DB-VERSION TLV This document 1213 TBD3 PCEP-PATH-VECTOR TLV This document 1215 9.3. STATEFUL-PCE-CAPABILITY TLV 1217 IANA is requested to allocate a new bit value in the STATEFUL-PCE- 1218 CAPABILITY TLV Flag Field sub-registry. 1220 Bit Description Reference 1221 TBD4 INTER-PCE-CAPABILITY This document 1223 10. References 1225 10.1. Normative References 1227 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1228 Requirement Levels", BCP 14, RFC 2119, 1229 DOI 10.17487/RFC2119, March 1997, 1230 . 1232 [RFC5440] Vasseur, JP., Ed. and JL. Le Roux, Ed., "Path Computation 1233 Element (PCE) Communication Protocol (PCEP)", RFC 5440, 1234 DOI 10.17487/RFC5440, March 2009, 1235 . 1237 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 1238 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 1239 May 2017, . 1241 [RFC8231] Crabbe, E., Minei, I., Medved, J., and R. Varga, "Path 1242 Computation Element Communication Protocol (PCEP) 1243 Extensions for Stateful PCE", RFC 8231, 1244 DOI 10.17487/RFC8231, September 2017, 1245 . 1247 [RFC8232] Crabbe, E., Minei, I., Medved, J., Varga, R., Zhang, X., 1248 and D. Dhody, "Optimizations of Label Switched Path State 1249 Synchronization Procedures for a Stateful PCE", RFC 8232, 1250 DOI 10.17487/RFC8232, September 2017, 1251 . 1253 [RFC8253] Lopez, D., Gonzalez de Dios, O., Wu, Q., and D. Dhody, 1254 "PCEPS: Usage of TLS to Provide a Secure Transport for the 1255 Path Computation Element Communication Protocol (PCEP)", 1256 RFC 8253, DOI 10.17487/RFC8253, October 2017, 1257 . 1259 10.2. Informative References 1261 [I-D.knodel-terminology] 1262 Knodel, M. and N. T. Oever, "Terminology, Power, and 1263 Inclusive Language in Internet-Drafts and RFCs", draft- 1264 knodel-terminology-05 (work in progress), February 2021. 1266 [RFC4655] Farrel, A., Vasseur, J., and J. Ash, "A Path Computation 1267 Element (PCE)-Based Architecture", RFC 4655, 1268 DOI 10.17487/RFC4655, August 2006, 1269 . 1271 [RFC6805] King, D., Ed. and A. Farrel, Ed., "The Application of the 1272 Path Computation Element Architecture to the Determination 1273 of a Sequence of Domains in MPLS and GMPLS", RFC 6805, 1274 DOI 10.17487/RFC6805, November 2012, 1275 . 1277 [RFC7399] Farrel, A. and D. King, "Unanswered Questions in the Path 1278 Computation Element Architecture", RFC 7399, 1279 DOI 10.17487/RFC7399, October 2014, 1280 . 1282 [RFC7525] Sheffer, Y., Holz, R., and P. Saint-Andre, 1283 "Recommendations for Secure Use of Transport Layer 1284 Security (TLS) and Datagram Transport Layer Security 1285 (DTLS)", BCP 195, RFC 7525, DOI 10.17487/RFC7525, May 1286 2015, . 1288 [RFC7752] Gredler, H., Ed., Medved, J., Previdi, S., Farrel, A., and 1289 S. Ray, "North-Bound Distribution of Link-State and 1290 Traffic Engineering (TE) Information Using BGP", RFC 7752, 1291 DOI 10.17487/RFC7752, March 2016, 1292 . 1294 [RFC8051] Zhang, X., Ed. and I. Minei, Ed., "Applicability of a 1295 Stateful Path Computation Element (PCE)", RFC 8051, 1296 DOI 10.17487/RFC8051, January 2017, 1297 . 1299 [RFC8281] Crabbe, E., Minei, I., Sivabalan, S., and R. Varga, "Path 1300 Computation Element Communication Protocol (PCEP) 1301 Extensions for PCE-Initiated LSP Setup in a Stateful PCE 1302 Model", RFC 8281, DOI 10.17487/RFC8281, December 2017, 1303 . 1305 [RFC8751] Dhody, D., Lee, Y., Ceccarelli, D., Shin, J., and D. King, 1306 "Hierarchical Stateful Path Computation Element (PCE)", 1307 RFC 8751, DOI 10.17487/RFC8751, March 2020, 1308 . 1310 [RFC8800] Litkowski, S., Sivabalan, S., Barth, C., and M. Negi, 1311 "Path Computation Element Communication Protocol (PCEP) 1312 Extension for Label Switched Path (LSP) Diversity 1313 Constraint Signaling", RFC 8800, DOI 10.17487/RFC8800, 1314 July 2020, . 1316 Appendix A. Contributors 1318 Dhruv Dhody 1319 Huawei Technologies 1320 Divyashree Techno Park, Whitefield 1321 Bangalore, Karnataka 560066 1322 India 1324 Email: dhruv.ietf@gmail.com 1326 Authors' Addresses 1328 Stephane Litkowski 1329 Cisco 1331 Email: slitkows.ietf@gmail.com 1332 Siva Sivabalan 1333 Ciena Corporation 1335 Email: msiva282@gmail.com 1337 Cheng Li 1338 Huawei Technologies 1339 Huawei Campus, No. 156 Beiqing Rd. 1340 Beijing 100095 1341 China 1343 Email: c.l@huawei.com 1345 Haomian Zheng 1346 Huawei Technologies 1347 H1, Huawei Xiliu Beipo Village, Songshan Lake 1348 Dongguan, Guangdong 523808 1349 China 1351 Email: zhenghaomian@huawei.com