idnits 2.17.1 draft-ietf-ccamp-inter-domain-rsvp-te-06.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- ** It looks like you're using RFC 3978 boilerplate. You should update this to the boilerplate described in the IETF Trust License Policy document (see https://trustee.ietf.org/license-info), which is required now. -- Found old boilerplate from RFC 3978, Section 5.1 on line 22. -- Found old boilerplate from RFC 3978, Section 5.5, updated by RFC 4748 on line 1020. -- Found old boilerplate from RFC 3979, Section 5, paragraph 1 on line 997. -- Found old boilerplate from RFC 3979, Section 5, paragraph 2 on line 1004. -- Found old boilerplate from RFC 3979, Section 5, paragraph 3 on line 1010. 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 'Updates: ' line in the draft header should list only the _numbers_ of the RFCs which will be updated by this document (if approved); it should not include the word 'RFC' in the list. -- The draft header indicates that this document updates RFC3473, but the abstract doesn't seem to mention this, which it should. -- The draft header indicates that this document updates RFC3209, but the abstract doesn't seem to mention this, which it should. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust Copyright Line does not match the current year == Line 826 has weird spacing: '...message mess...' (Using the creation date from RFC3209, updated by this document, for RFC5378 checks: 1998-11-20) -- The document seems to lack a disclaimer for pre-RFC5378 work, but may have content which was first submitted before 10 November 2008. If you have contacted all the original authors and they are all willing to grant the BCP78 rights to the IETF Trust, then this is fine, and you can ignore this comment. If not, you may need to add the pre-RFC5378 disclaimer. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (April 2007) is 6220 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) ** Obsolete normative reference: RFC 4420 (Obsoleted by RFC 5420) -- Obsolete informational reference (is this intentional?): RFC 4379 (Obsoleted by RFC 8029) Summary: 2 errors (**), 0 flaws (~~), 3 warnings (==), 10 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 1 INTERNET-DRAFT A. Farrel (Editor) 2 Network Working Group Old Dog Consulting 3 Intended Status: Standards Track 4 Updates: RFC 3209, RFC 3473 A. Ayyangar 5 Expires: October 2007 Nuova Systems 7 JP. Vasseur 8 Cisco Systems, Inc. 10 April 2007 12 Inter domain Multiprotocol Label Switching (MPLS) and 13 Generalized MPLS (GMPLS) Traffic Engineering - RSVP-TE extensions 15 draft-ietf-ccamp-inter-domain-rsvp-te-06.txt 17 Status of this Memo 19 By submitting this Internet-Draft, each author represents that any 20 applicable patent or other IPR claims of which he or she is aware 21 have been or will be disclosed, and any of which he or she becomes 22 aware will be disclosed, in accordance with Section 6 of BCP 79. 24 Internet-Drafts are working documents of the Internet Engineering 25 Task Force (IETF), its areas, and its working groups. Note that 26 other groups may also distribute working documents as Internet- 27 Drafts. 29 Internet-Drafts are draft documents valid for a maximum of six months 30 and may be updated, replaced, or obsoleted by other documents at any 31 time. It is inappropriate to use Internet-Drafts as reference 32 material or to cite them other than as "work in progress." 34 The list of current Internet-Drafts can be accessed at 35 http://www.ietf.org/ietf/1id-abstracts.txt. 37 The list of Internet-Draft Shadow Directories can be accessed at 38 http://www.ietf.org/shadow.html. 40 Abstract 42 This document describes procedures and protocol extensions for the 43 use of Resource ReserVation Protocol Traffic Engineering (RSVP-TE) 44 signaling in Multiprotocol Label Switching Traffic Engineering 45 (MPLS-TE) packet networks and Generalized MPLS (GMPLS) packet and 46 non-packet networks to support the establishment and maintenance of 47 Label Switched Paths that cross domain boundaries. 49 For the purpose of this document, a domain is considered to be any 50 collection of network elements within a common realm of address space 51 or path computation responsibility. Examples of such domains include 52 Autonomous Systems, IGP routing areas, and GMPLS overlay networks. 54 Table of Contents 56 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 2 57 1.1 Conventions Used In This Document . . . . . . . . . . . . 3 58 1.2 Terminology . . . . . . . . . . . . . . . . . . . . . . . 3 59 2. Signaling Overview . . . . . . . . . . . . . . . . . . . . . 4 60 2.1 Signaling Options . . . . . . . . . . . . . . . . . . . . 4 61 3. Procedures on the Domain Border Node . . . . . . . . . . . . . 5 62 3.1 Rules on ERO Processing . . . . . . . . . . . . . . . . . 6 63 3.2 LSP Setup Failure and Crankback . . . . . . . . . . . . . 8 64 3.3 RRO Processing Across Domains . . . . . . . . . . . . . . 9 65 3.4 Notify Message Processing . . . . . . . . . . . . . . . . 9 66 4. RSVP-TE Signaling Extensions . . . . . . . . . . . . . . . . . 10 67 4.1 Control of Downstream Choice of Signaling Method . . . . . 10 68 5. Protection and Recovery of Inter-Domain TE LSPs . . . . . . . 11 69 5.1 Fast Recovery Support Using MPLS-TE Fast Reroute . . . . . 11 70 5.1.1 Failure Within a Domain (Link or Node Failure) . . . . 12 71 5.1.2 Failure of Link at Domain Borders . . . . . . . . . . 12 72 5.1.3 Failure of a Border Node . . . . . . . . . . . . . . . 12 73 5.2 Protection and Recovery of GMPLS LSPs . . . . . . . . . . 13 74 6. Re-Optimization of Inter-Domain TE LSPs . . . . . . . . . . . 13 75 7. Backward Compatibility . . . . . . . . . . . . . . . . . . . . 5 76 8. Security Considerations . . . . . . . . . . . . . . . . . . . 15 77 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 17 78 9.1 Attribute Flags for LSP_Attributes Object . . . . . . . . 17 79 9.2 New Error Codes . . . . . . . . . . . . . . . . . . . . . 17 80 10. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 18 81 11. References . . . . . . . . . . . . . . . . . . . . . . . . . 18 82 11.1 Normative References . . . . . . . . . . . . . . . . . . 18 83 11.2 Informative References . . . . . . . . . . . . . . . . . 18 84 12. Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . 20 86 1. Introduction 88 The requirements for inter-area and inter-AS Multiprotocol Label 89 Switching (MPLS) Traffic Engineering (TE) are stated in [RFC4105] and 90 [RFC4216] respectively. Many of these requirements also apply to 91 Generalized MPLS (GMPLS) networks. The framework for inter-domain 92 MPLS-TE is provided in [RFC4726]. 94 This document presents procedures and extensions to Resource 95 Reservation Protocol Traffic Engineering (RSVP-TE) signaling for the 96 setup and maintenance of traffic engineered Label Switched Paths (TE 97 LSPs) that span multiple domains in MPLS-TE or GMPLS networks. The 98 signaling procedures described in this document are applicable to 99 MPLS-TE packet LSPs established using RSVP-TE ([RFC3209]) and all 100 LSPs (packet and non-packet) that use RSVP-TE GMPLS extensions as 101 described in [RFC3473]. 103 Three different signaling methods for inter-domain RSVP-TE signaling 104 are identified in [RFC4726]. Contiguous LSPs are 105 achieved using the procedures of [RFC3209] and [RFC3473] to create a 106 single end-to-end LSP that spans all domains. Nested LSPs are 107 established using the techniques described in [RFC4206] to carry the 108 end-to-end LSP in a separate tunnel across each domain. Stitched LSPs 109 are established using the procedures of [LSP-STITCHING] to construct 110 an end-to-end LSP from the concatenation of separate LSPs each 111 spanning a domain. 113 This document defines the RSVP-TE protocol extensions necessary to 114 control and select which of the three signaling mechanisms is used 115 for any one end-to-end inter-domain TE LSP. 117 For the purpose of this document, a domain is considered to be any 118 collection of network elements within a common realm of address space 119 or path computation responsibility. Examples of such domains include 120 Autonomous Systems, IGP areas, and GMPLS overlay networks 121 ([RFC4208]). 123 1.1. Conventions Used In This Document 125 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 126 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 127 document are to be interpreted as described in RFC 2119 [RFC2119]. 129 1.2. Terminology 131 AS: Autonomous System. 133 ASBR: routers used to connect together ASes of a different or the 134 same Service Provider via one or more Inter-AS links. 136 Bypass Tunnel: an LSP that is used to protect a set of LSPs passing 137 over a common facility. 139 ERO: Explicit Route Object. 141 FA: Forwarding Adjacency. 143 LSR: Label Switching Router. 145 MP: Merge Point. The node where bypass tunnels meet the protected 146 LSP. 148 NHOP bypass tunnel: Next-Hop Bypass Tunnel. A backup tunnel, which 149 bypasses a single link of the protected LSP. 151 NNHOP bypass tunnel: Next-Next-Hop Bypass Tunnel. A backup tunnel, 152 which bypasses a single node of the protected LSP. 154 PLR: Point of Local Repair. The ingress of a bypass tunnel. 156 RRO: Record Route Object. 158 TE link: Traffic Engineering link. 160 2. Signaling Overview 162 The RSVP-TE signaling of a TE LSP within a single domain is described 163 in [RFC3209] and [RFC3473]. Inter-domain TE LSPs can be supported by 164 one of three options as described in [RFC4726] and set out in the 165 next section: 167 - contiguous LSPs 168 - nested LSPs 169 - stitched LSPs. 171 In fact, as pointed out in [RFC4726], any combination of these three 172 options may be used in the course of an end-to-end inter-domain LSP. 174 This document describes the RSVP-TE signaling extensions required to 175 control and select which of the three signaling mechanisms is used 176 for any one end-to-end inter-domain TE LSP. 178 The specific protocol extensions required to signal each LSP type are 179 described in other documents and are out of scope for this document. 180 Similarly, the routing extensions and path computation techniques 181 necessary for the establishment of inter-domain LSPs are out of 182 scope. 184 2.1. Signaling Options 186 There are three ways in which an RSVP-TE LSP could be signaled across 187 multiple domains: 189 Contiguous 190 A contiguous TE LSP is a single end-to-end TE LSP that is set up 191 across multiple domains using RSVP-TE signaling procedures 192 described in [RFC3209] and [RFC3473]. No additional TE LSPs are 193 required to create a contiguous TE LSP and the same RSVP-TE 194 information for the TE LSP is maintained along the entire LSP 195 path. In particular, the TE LSP has the same RSVP-TE session and 196 LSP ID at every LSR along its path. 198 Nesting 199 One or more TE LSPs may be nested within another TE LSP as 200 described in [RFC4206]. This technique can be used to nest one 201 or more inter-domain TE LSPs into an intra-domain hierarchical LSP 202 (H-LSP). The label stacking construct is used to achieve nesting 203 in packet networks. In the rest of this document, the term H-LSP 204 is used to refer to an LSP that allows other LSPs to be nested 205 within it. An H-LSP may be advertised as a TE link within the same 206 instance of the routing protocol as was used to advertise the TE 207 links from which it was created, in which case it is a Forwarding 208 Adjacency (FA) [RFC4206]. 210 Stitching 211 The concept of LSP stitching as well as the required signaling 212 procedures is described in [LSP-STITCHING]. This technique can be 213 used to stitch together shorter LSPs (LSP segments) to create a 214 single, longer LSP. The LSP segments of an inter-domain LSP may be 215 intra-domain LSPs or inter-domain LSPs. 217 The process of stitching in the data plane results in a single, 218 end-to-end contiguous LSP. But in the control plane each segment is 219 signaled as a separate LSP (with distinct RSVP sessions) and the 220 end-to-end LSP is signaled as yet another LSP with its own RSVP 221 session. Thus the control plane operation for LSP stitching is very 222 similar to that for nesting. 224 An end-to-end inter-domain TE LSP may be achieved using one or more 225 of the signaling techniques described. The choice is a matter of 226 policy for the node requesting LSP setup (the ingress) and policy for 227 each successive domain border node. On receipt of an LSP setup 228 request (RSVP-TE Path message) for an inter-domain TE LSP, the 229 decision of whether to signal the LSP contiguously or whether to nest 230 or stitch it to another TE LSP, depends on the parameters signaled 231 from the ingress node and on the configuration of the local node. 233 The stitching segment LSP or H-LSP used to cross a domain may be 234 pre-established or signaled dynamically based on the demand caused by 235 the arrival of the inter-domain TE LSP setup request. 237 3. Procedures on the Domain Border Node 239 Whether an inter-domain TE LSP is contiguous, nested, or stitched is 240 limited by the signaling methods supported by or configured on the 241 intermediate nodes, and it is usually the domain border nodes where 242 this restriction applies since other transit nodes are oblivious to 243 the mechanism in use. The ingress of the LSP may further restrict the 244 choice by setting parameters in the Path message when it is signaled. 246 When a domain border node receives the RSVP Path message for an 247 inter-domain TE LSP setup, it MUST carry out the following procedures 248 before it can forward the Path message to the next node along the 249 path: 251 1. Apply policies for the domain and the domain border node. These 252 policies may restrict the establishment of inter-domain TE LSPs. 253 In case of a policy failure, the node SHOULD fail the setup and 254 send a PathErr message with error code "Policy control failure"/ 255 "Inter-domain policy failure". 257 2. Determine the signaling method to be used to cross the domain. 258 If the ingress node of the inter-domain TE LSP has specified 259 restrictions on the methods to be used, these MUST be adhered 260 to. Within the freedom allowed by the ingress node, the domain 261 border node MAY choose any method according to local 262 configuration and policies. If no resultant signaling method is 263 available or allowed, the domain border node MUST send a PathErr 264 message with an error code as described in Section 4.1. 266 3. Carry out ERO procedures as described in Section 3 in addition 267 to the procedures in [RFC3209] and [RFC3473]. 269 4. Perform any path computations as required to determine the path 270 across the domain and potentially to select the exit point from 271 the domain. 273 The path computation procedure is outside the scope of this 274 document. A path computation option is specified in 275 [INTER-DOMAIN-PD-PATH-COMP], and another option is to use a 276 Path Computation Element (PCE) [RFC4655]. 278 4a. In the case of nesting or stitching, either find an existing 279 intra-domain TE LSP to carry the inter-domain TE LSP or 280 signal a new one, depending on local policy. 282 In event of a path computation failure, a PathErr message SHOULD 283 be sent with error code "Routing Problem" using an error value 284 selected according to the reason for computation failure. 286 In the event of the receipt of a PathErr message reporting 287 signaling failure from within the domain or reported from a 288 downstream domain, the domain border node MAY apply crankback 289 procedures as described in Section 3.2. If crankback is not 290 applied, or is exhausted, the border node MUST continue with 291 PathErr processing as described in [RFC3209] and [RFC3473]. 293 In the event of successful processing of a Path or Resv message, 294 the domain border node MUST carry out RRO procedures as described 295 in Section 3.3. 297 3.1. Rules on ERO Processing 299 The ERO that a domain border node receives in the Path message was 300 supplied by the ingress node of the TE LSP and may have been updated 301 by other nodes (for example, other domain border nodes) as the Path 302 message was propagated. The content of the ERO depends on several 303 factors including: 305 - the path computation techniques used 306 - the degree of TE visibility available to the nodes performing 307 path computation 308 - policy at the nodes creating/modifying the ERO. 310 In general, H-LSPs and LSP segments are used between domain border 311 nodes, but there is no restriction on the use of such LSPs to span 312 multiple hops entirely within a domain. Therefore, the discussion 313 that follows may be equally applied to any node within a domain 314 although the term 'domain border node' continues to be used for 315 clarity. 317 When a Path message reaches the domain border node, the following 318 rules SHOULD be used for ERO processing and for further signaling. 320 1. If there are any policies related to ERO processing for the 321 LSP, they SHOULD be applied and corresponding actions taken. For 322 example, there might be a policy to reject EROs that identify 323 nodes within the domain. In case of inter-domain LSP setup 324 failures due to policy failures related to ERO processing, the 325 node SHOULD issue a PathErr with error code "Policy control 326 failure"/"Inter-domain explicit route rejected". 328 2. Section 8.2 of [RFC4206] describes how a node at the edge of a 329 region processes the ERO in the incoming Path message and uses 330 this ERO, to either find an existing H-LSP or signal a new H-LSP 331 using the ERO hops. This process includes adjusting the ERO 332 before sending the Path message to the next hop. These 333 procedures SHOULD be followed for nesting or stitching of 334 inter-domain TE LSPs. 336 3. If an ERO subobject identifies a TE link formed by the 337 advertisement of an H-LSP or LSP segment (whether numbered or 338 unnumbered), contiguous signaling MUST NOT be used. The node MUST 339 either use nesting or stitching according to the capabilities of 340 the LSP that forms the TE link, the parameters signaled in the 341 Path message, and local policy. If there is a conflict between the 342 capabilities of the LSP that forms the TE link indicated in the 343 ERO and the parameters on the Path message, the domain border node 344 SHOULD send a PathErr with error code "Routing Problem"/"ERO 345 conflicts with inter-domain signaling method". 347 4. An ERO in a Path message received by a domain border node may 348 have a loose hop as the next hop. This may be an IP address or 349 an AS number. In such cases, the ERO MUST be expanded to 350 determine the path to the next hop using some form of path 351 computation that may, itself, generate loose hops. 353 5. In the absence of any ERO subobjects beyond the local domain 354 border node, the LSP egress (the destination encoded in the RSVP 355 Session object) MUST be considered as the next loose hop and 356 rule 4 applied. 358 6. In the event of any other failures processing the ERO, a PathErr 359 message SHOULD be sent as described in [RFC3209] or [RFC3473]. 361 3.2. LSP Setup Failure and Crankback 363 When an error occurs during LSP setup a PathErr message is sent back 364 towards the LSP ingress node to report the problem. If the LSP 365 traverses multiple domains, this PathErr will be seen successively by 366 each domain border node. 368 Domain border nodes MAY apply local policies to restrict the 369 propagation of information about the contents of the domain. For 370 example, a domain border node MAY replace the information in a 371 PathErr message that indicates a specific failure at a specific node 372 with information that reports a more general error with the entire 373 domain. These procedures are similar to those described for the 374 borders of overlay networks in [RFC4208]. 376 However: 378 - A domain border node MUST NOT suppress the propagation of a PathErr 379 message. 381 - Nodes other than domain border nodes SHOULD NOT modify the contents 382 of a PathErr message. 384 - Domain border nodes SHOULD NOT modify the contents of a PathErr 385 message unless domain confidentiality is a specific requirement. 387 Domain border nodes provide an opportunity for crankback rerouting 388 [CRANKBACK]. On receipt of a PathErr message generated because of an 389 LSP setup failure, a domain border node MAY hold the PathErr and make 390 further attempts to establish the LSP if allowed by local policy and 391 by the parameters signaled on the Path message for the LSP. Such 392 attempts might involve the computation of alternate routes through 393 the domain, or the selection of different downstream domains. If a 394 subsequent attempt is successful, the domain border router MUST 395 discard the held PathErr message, but if all subsequent attempts are 396 unsuccessful, the domain border router MUST send the PathErr upstream 397 toward the ingress node. In this latter case, the domain border 398 router MAY change the information in the PathErr message to provide 399 further crankback details and information aggregation as described 400 in [CRANKBACK]. 402 Crankback rerouting MAY also be used to handle the failure of LSPs 403 after they have been established [CRANKBACK]. 405 3.3. RRO Processing Across Domains 407 [RFC3209] defines the RRO as an optional used for loop detection and 408 for providing information about the hops traversed by LSPs. 410 As described for overlay networks in [RFC4208], a domain border node 411 MAY filter or modify the information provided in an RRO for 412 confidentiality reasons according to local policy. For example, a 413 series of identifiers of hops within a domain MAY be replaced with 414 the domain identifier (such as the AS number) or be removed entirely 415 leaving just the domain border nodes. 417 Note that a domain border router MUST NOT mask its own presence, and 418 MUST include itself in the RRO. 420 Such filtering of RRO information does not hamper the working of the 421 signaling protocol, but the subsequent information loss may render 422 management diagnostic procedures inoperable or at least make them 423 more complicated requiring the coordination of administrators of 424 multiple domains. 426 Similarly protocol procedures that depend on the presence of RRO 427 information may become inefficient. For example, the fast reroute 428 procedures defined in [RFC4090] use information in the RRO to 429 determine the labels to use and the downstream MP. 431 3.4. Notify Message Processing 433 Notify messages are introduced in [RFC3473]. They may be sent direct 434 rather than hop-by-hop, and so may speed the propagation of error 435 information. If a domain border router is interested in seeing 436 such messages (for example, to enable it to provide protection 437 switching), it is RECOMMENDED that the domain border router update 438 the Notify Request objects in the Path and Resv messages to show its 439 own address following the procedures of [RFC3473]. 441 Note that the replacement of a Notify Recipient in the Notify Request 442 object means that some Notify messages (for example, those intended 443 for delivery to the ingress LSR) may need to be examined, processed 444 and forwarded at domain borders. This is an obvious trade-off issue 445 as the ability to handle notifiable events locally (i.e. within the 446 domain) may or may not outweigh the cost of processing and forwarding 447 Notify messages beyond the domain. Observe that the cost increases 448 linearly with the number of domains in use. 450 Also note that, as described in section 8, a domain administrator may 451 wish to filter or modify Notify messages that are generated within a 452 domain in order to preserve security or confidentiality of network 453 information. This is most easily achieved if the Notify messages are 454 sent via the domain borders. 456 4. RSVP-TE Signaling Extensions 458 The following RSVP-TE signaling extensions are defined to enable 459 inter-domain LSP setup. 461 4.1. Control of Choice of Signaling Method 463 In many network environments there may be a network-wide policy that 464 determines which one of the three inter-domain LSP techniques is 465 used. In these cases, no protocol extensions are required. 467 However, in environments that support more than one technique, an 468 ingress node may wish to constrain the choice made by domain border 469 nodes for each inter-domain TE LSP that it originates. 471 [RFC4420] defines the LSP_Attributes object that can be used to 472 signal required attributes of an LSP. The Attributes Flags TLV 473 includes Boolean flags that define individual attributes. 475 This document defines a new bit in the TLV that can be set by the 476 ingress node of an inter-domain TE LSP to restrict the intermediate 477 nodes to using contiguous signaling. 479 Contiguous LSP bit (bit number assignment in Section 9.1). 481 This flag is set by the ingress node that originates a Path message 482 to set up an inter-domain TE LSP if it requires that the contiguous 483 LSP technique is used. This flag bit is only to be used in the 484 Attributes Flags TLV. 486 When a domain border LSR receives a Path message containing this bit 487 set (one), the node MUST NOT perform stitching or nesting in support 488 of the inter-domain TE LSP being set up. When this bit is clear 489 (zero), a domain border LSR MAY perform stitching or nesting 490 according to local policy. 492 This bit MUST NOT be modified by any transit node. 494 An intermediate node that supports the LSP_Attributes object and the 495 Attributes Flags TLV, and also recognizes the "Contiguous LSP" bit, 496 but cannot support contiguous TE LSPs, MUST send a Path Error message 497 with an error code "Routing Problem"/"Contiguous LSP type not 498 supported" if it receives a Path message with this bit set. 500 If an intermediate node receiving a Path message with the "Contiguous 501 LSP" bit set in the Flags field of the LSP_Attributes, recognizes the 502 object, the TLV, and the bit and also supports the desired contiguous 503 LSP behavior, then it MUST signal a contiguous LSP. If the node is a 504 domain border node, or if the node expands a loose hop in the ERO, it 505 MUST include an RRO Attributes subobject in the RRO of the 506 corresponding Resv message (if such an object is present) with the 507 "Contiguous LSP" bit set to report its behavior. 509 Domain border LSRs MUST support and act on the setting of the 510 "Contiguous LSP" flag. 512 However, if the intermediate node supports the LSP_Attributes object 513 but does not recognize the Attributes Flags TLV, or supports the TLV 514 but does not recognize this "Contiguous LSP" bit, then it MUST 515 forward the object unmodified. 517 The choice of action by an ingress node that receives a PathErr when 518 requesting the use of a contiguous LSP is out of the scope of this 519 document, but may include the computation of an alternate path. 521 5. Protection and Recovery of Inter-Domain TE LSPs 523 The procedures described in Sections 3 and 4 MUST be applied to all 524 inter-domain TE LSPs, including bypass tunnels, detour LSPs 525 [RFC4090], and segment recovery LSPs [SEGMENT-PROTECTION]. This means 526 that these LSPs will also be subjected to ERO processing, policies, 527 path computation, etc. 529 Note also that the paths for these backup LSPs needs to be either 530 pre-configured, computed and signaled with the protected LSP, or 531 computed on-demand at the PLR. Just as with any inter-domain TE LSP, 532 the ERO may comprise of strict or loose hops, and will depend on the 533 TE visibility of the computation point into the subsequent domain. 535 If loose hops are required created in the path of the backup LSP, ERO 536 expansion will be required at some point along the path: probably at 537 a domain border node. In order that the backup path remains disjoint 538 from the protected LSP(s) the node performing the ERO expansion must 539 be provided with the path of the protected LSPs between the PLR and 540 the MP. This information can be gathered from the RROs of the 541 protected LSPs and is signaled in the DETOUR object for Fast Reroute 542 [RFC4090], and using route exclusion [RFC4874] for other protection 543 schemes. 545 5.1. Fast Recovery Support Using MPLS-TE Fast Reroute (FRR) 547 [RFC4090] describes two methods for local protection for a 548 packet TE LSP in case of link, SRLG, or node failure. This section 549 describes how these mechanisms work with the proposed signaling 550 solutions for inter-domain TE LSP setup. 552 5.1.1. Failure Within a Domain (Link or Node Failure) 554 The mode of operation of MPLS-TE Fast Reroute to protect a 555 contiguous, stitched or nested TE LSP within a domain is identical to 556 the existing procedures described in [RFC4090]. Note that, in the 557 case of nesting or stitching, the end-to-end LSP is automatically 558 protected by the protection operation performed on the H-LSP or 559 stitching segment LSP. 561 No protocol extensions are required. 563 5.1.2. Failure of a Link at a Domain Borders 565 This cases arises where two domains are connected by a TE link. In 566 this case each domain has its own domain border node, and these two 567 nodes are connected by the TE link. An example of this case is where 568 the ASBRs of two ASes are connected by a TE link. 570 A contiguous LSP can be backed up using any PLR and MP, but if the 571 LSP uses stitching or nesting in either of the connected domains, the 572 PLR and MP MUST be domain border nodes for those domains. It will be 573 usual to attempt to use the local (connected by the failed link) 574 domain border nodes as the PLR and MP. 576 To protect an inter-domain link with MPLS-TE Fast Reroute, a set of 577 backup tunnels must be configured or dynamically computed between the 578 PLR and MP such that they are diversely routed from the protected 579 inter-domain link and the protected inter-domain LSPs. 581 Each protected inter-domain LSP using the protected inter-domain TE 582 link must be assigned to an NHOP bypass tunnel that is diverse from 583 the protected LSP. Such an NHOP bypass tunnel can be selected by 584 analyzing the RROs in the Resv messages of the available bypass 585 tunnels and the protected TE LSP. It may be helpful to this process 586 if the extensions defined in [RFC4561] are used to clearly 587 distinguish nodes and links in the RROs. 589 5.1.3. Failure of a Border Node 591 Two border node failure cases exist. If the domain border falls on a 592 link as described in the previous section, the border node at either 593 end of the link may fail. Alternatively, if the border falls on a 594 border node (as is the case with IGP areas) that single border node 595 may fail. 597 It can be seen that if stitching or nesting are used, the failed node 598 will be the start or end (or both) or a stitching segment LSP or 599 H-LSP in which case, protection must be provided to the far end of 600 stitching segment or H-LSP. Thus, where one of these two techniques 601 is in use, the PLR will be the upstream domain entry point in the 602 case of the failure of the domain exit point, and the MP will be the 603 downstream domain exit point in the case of the failure of the 604 domain entry point. Where the domain border falls at a single domain 605 border node, both cases will apply. 607 If the contiguous LSP mechanism is in use, normal selection of the 608 PLR and MP can be applied and any node within the domains may be used 609 to fill these roles. 611 As before, selection of a suitable backup tunnel (in this case an 612 NNHOP backup) must consider the paths of the backed up LSPs and the 613 available NNHOP tunnels by examination of their RROs. 615 Note that where the PLR is not immediately upstream of the failed 616 node, error propagation time may be delayed unless some mechanism 617 such as [BFD-MPLS] is implemented, or unless direct reporting, such 618 as through the GMPLS Notify message [RFC3473] is employed. 620 5.2. Protection and Recovery of GMPLS LSPs 622 [SEGMENT-PROTECTION] describes GMPLS based segment recovery. This 623 allows protection against a span failure, a node failure, or failure 624 over any particular portion of a network used by an LSP. 626 The domain border failure cases described in Section 5.1 may also 627 occur in GMPLS networks (including packet networks) and can be 628 protected against using segment protection without any additional 629 protocol extensions. 631 Note that if loose hops are used in the construction of the working 632 and protection paths signaled for segment protection then care is 633 required to keep these paths disjoint. If the paths are signaled 634 incrementally then route exclusion [RFC4874] may be used to ensure 635 that the paths are disjoint. Otherwise a coordinated path computation 636 technique such as that offered by cooperating Path Computation 637 Elements [RFC4655] can provide suitable paths. 639 6. Re-Optimization of Inter-Domain TE LSPs 641 Re-optimization of a TE LSP is the process of moving the LSP from the 642 current path to a more prefered path. This involves the determination 643 of the preferred path and make-before-break signaling procedures 644 [RFC3209] to minimize traffic disruption. 646 Re-optimization of an inter-domain TE LSP may require a new path in 647 more than one domain. 649 The nature of the inter-domain LSP setup mechanism defines how 650 re-optimization can be applied. If the LSP is contiguous then the 651 signaling of the make-before-break process MUST be initiated by the 652 ingress node as defined in [RFC3209]. But if the re-optimization is 653 limited to a change in path within one domain (that is, if there is 654 no change to the domain border nodes) and nesting or stitching are in 655 use, the H-LSP or stitching segment may be independently re-optimized 656 within the domain without impacting the end-to-end LSP. 658 In all cases, however, the ingress LSP may wish to exert control and 659 coordination over the re-optimization process. For example, a transit 660 domain may be aware of the potential for reoptimization, but not 661 bother because it is not worried by the level of service being 662 provided across the domain. But the cummulative effect on the 663 end-to-end LSP may cause the head-end to worry and trigger an 664 end-to-end reoptimization request (of course, the transit domain may 665 choose to ignore the request). 667 Another benefit to end-to-end reoptimization over per-domain 668 reoptimization for non-contiguous inter-domain LSPs is that 669 per-domain re-optimization is restricted to preserve the domain entry 670 and exit points (since to do otherwise would break the LSP!). But 671 end-to-end reoptimization is more flexible and can select new domain 672 border LSRs. 674 There may be different cost benefit analysis considerations to choose 675 between end-to-end reoptimization and per-domain reoptimization. The 676 greater the number of hops involved in the reoptimization, the higher 677 the risk of traffic disruption. The shorter the segment reoptimized, 678 the lower the chance of making a substantial improvement on the 679 qulaity of the end-to-end LSP. Administrative policies should be 680 applied in this area with care. 682 [RFC4736] describes mechanisms that allow: 684 - The ingress node to request each node with a loose next hop to 685 re-evaluate the current path in order to search for a more optimal 686 path. 688 - A node with a loose next hop to inform the ingress node that a 689 better path exists. 691 These mechanisms SHOULD be used for re-optimization of a contiguous 692 inter-domain TE LSP. 694 Note that end-to-end reoptimization may involve a non-local 695 modification and that might select new entry / exit points. In this 696 case, we can observe that local reoptimization is more easily and 697 flexibly achieved using nesting or stitching. Further, the "locality 698 principle" (i.e., the idea of keeping information only where it is 699 needed) is best achieved using stitching or nesting. That said, a 700 contiguous LSP can easily be modified to take advantage of local 701 reoptimizations (as defined in [RFC4736]) even if this would require 702 the dissemination of information and the invocation of signaling 703 outside the local domain. 705 7. Backward Compatibility 707 The procedures in this document are backward compatible with esiting 708 deployments. 710 - Ingress LSRs are not required to support the extensions in this 711 document to provision intra-domain LSPs. The default behavior by 712 transit LSRs that receive a Path message that does not have the 713 "Contiguous LSP" bit set in the Attributes Flags TLV of the 714 LSP_Attribtes object or does not even have the object present is 715 to allow all modes of inter-domain TE LSP, so back-level ingress 716 LSRs are able to initiate inte-domain LSPs. 718 - Transit, non-border LSRs are not required to perform any special 719 processing and will pass the LSP_Attributes object onwards 720 unmodified according to the rules of [RFC2205]. Thus back-level 721 transit LSRs are fully supported. 723 - Domain border LSRs are likely to be upgraded before inter-domain 724 TE LSPs are allowed. This is because of the need to establish 725 policy, administrative, and security conrols before permitting 726 inter-domain LSPs to be signaled across a domain border. Thus 727 legacy domain border LSRs do not need to be considered. 729 The RRO additions in this document are fully backward compatible. 731 8. Security Considerations 733 A separate document is being prepared to examine the security aspects 734 of RSVP-TE signaling with special reference to multi-domain 735 scenarios [MPLS-GMPLS-SEC]. [RFC4726] provides an overview of the 736 requirements for security in an MPLS-TE or GMPLS multi-domain 737 environment. 739 Before electing to utilise inter-domain signaling for MPLS-TE, the 740 administrators of neighboring domains MUST satisfy themselves of the 741 existence of a suiable trust relationship between the domains. In the 742 absence of such a relationship, the administrators SHOULD decide not 743 to deploy inter-domain signaling, and SHOULD disable RSVP-TE on any 744 inter-domain interfaces. 746 When signaling an inter-domain RSVP-TE LSP, an operator MAY make use 747 of the security features already defined for RSVP-TE [RFC3209]. This 748 may require some coordination between the domains to share the keys 749 (see [RFC2747] and [RFC3097]), and care is required to ensure that 750 the keys are changed sufficiently frequently. Note that this may 751 involve additional synchronization, should the domain border nodes 752 be protected with FRR, since the MP and PLR should also share the 753 key. 755 For an inter-domain TE LSP, especially when it traverses different 756 administrative or trust domains, the following mechanisms SHOULD be 757 provided to an operator (also see [RFC4216]): 759 1) A way to enforce policies and filters at the domain borders 760 to process the incoming inter-domain TE LSP setup requests 761 (Path messages) based on certain agreed trust and service 762 levels/contracts between domains. Various LSP attributes 763 such as bandwidth, priority, etc. could be part of such a 764 contract. 766 2) A way for the operator to rate-limit LSP setup requests 767 or error notifications from a particular domain. 769 3) A mechanism to allow policy-based outbound RSVP message 770 processing at the domain border node, which may involve 771 filtering or modification of certain addresses in RSVP 772 objects and messages. 774 Additionally, an operator may wish to reduce the signaling 775 interactions between domains to improve security. For example, the 776 operator might not trust the neighboring domain to supply correct or 777 trustable restart information [RSVP-RESTART] and might ensure that 778 the availablity of restart function is not configured in the Hello 779 message exchange across the domain border. Thus, suitable 780 configuration MUST be provided in an RSVP-TE implementation to 781 enable the operator to control optional protocol features that may be 782 considered a security risk. 784 Some examples of the policies described above are as follows: 786 A) An operator may choose to implement some kind of ERO filtering 787 policy on the domain border node to disallow or ignore hops 788 within the domain from being identified in the ERO of an 789 incoming Path message. That is, the policy is that a node 790 outside the domain cannot specify the path of the LSP inside the 791 domain. The domain border LSR can make implement this policy in 792 one of two ways: 793 - It can reject the Path message. 794 - It can ignore the hops in the ERO that lie within the domain. 796 B) In order to preserve confidentiality of network topology, an 797 operator may choose to disallow recording of hops within the 798 domain in the RRO or may choose to filter out certain recorded 799 RRO addresses at the domain border node. 801 C) An operator may require the border node to modify the addresses 802 of certain messages like PathErr or Notify originated from hops 803 within the domain. 805 Note that the detailed specification of such policies and their 806 implementation are outside the scope of this document. 808 OAM mechanisms including [BFD-MPLS] and [RFC4379] are commonly used 809 to verify he connectivity of end-to-end LSPs and to trace their 810 paths. Where the LSPs are inter-domain LSPs, such OAM techniques MAY 811 require to be intercepted or modified at domain borders, or to be 812 passed transparently across domains. Further discussion of this topic 813 can be found in [INTERAS-PING] and [MPLS-GMPLS-SEC]. 815 9. IANA Considerations 817 IANA is requested to make the codepoint allocations described in the 818 following sections. 820 9.1. Attribute Flags for LSP_Attributes Object 822 A new bit is to be allocated from the "Attributes Flags" sub-registry 823 of the "RSVP TE Parameters" registry. 825 Path Resv RRO 826 Bit Name message message sub-object 827 ---- ------------------------------- -------- -------- ---------- 828 XX Contiguous LSP Yes No Yes 830 The value XX is to be defined by IANA. A value of 4 is suggested. 832 9.2. New Error Codes 834 New RSVP error codes/values are required to be allocated from the 835 "Error Codes and Globally-Defined Error Value Sub-Codes" sub-registry 836 of the "RSVP Parameters" registry. 838 For the existing error code "Policy control failure" (value 2), two 839 new error values are suggested as follows. The values are suggested 840 and are for IANA determination. 842 103 = Inter-domain policy failure 843 104 = Inter-domain explicit route rejected 845 For the existing error code "Routing Problem" (value 24), two new 846 error values are suggested as follows. The values are suggested and 847 are for IANA determination. 849 21 = Contiguous LSP type not supported 850 22 = ERO conflicts with inter-domain signaling method 852 10. Acknowledgements 854 The authors would like to acknowledge the input and helpful comments 855 from Kireeti Kompella on various aspects discussed in the document. 856 Deborah Brungard and Dimitri Papdimitriou provided thorough reviews. 858 11. References 860 11.1. Normative References 862 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 863 Requirement Levels", BCP 14, RFC 2119, March 1997. 865 [RFC2205] Braden, R., Zhang, L., Berson, S., Herzog, S. and S. Jamin, 866 "Resource ReSerVation Protocol (RSVP) -- Version 1, 867 Functional Specification", RFC 2205, September 1997. 869 [RFC3209] Awduche, D., et al, "Extensions to RSVP for LSP Tunnels", 870 RFC 3209, December 2001. 872 [RFC3473] Berger, L., et al, "Generalized Multi-Protocol Label 873 Switching (GMPLS) Signaling Resource ReserVation Protocol - 874 Traffic Engineering (RSVP-TE) Extensions", RFC 3473, 875 January 2003. 877 [RFC4206] Kompella K., Rekhter Y., "LSP Hierarchy with Generalized 878 MPLS TE", RFC 4206, October 2005. 880 [RFC4420] Farrel, A., et al, "Encoding of Attributes for 881 Multiprotocol Label Switching (MPLS) Label Switched Path 882 (LSP) Establishment Using RSVP-TE", RFC 4420, February 883 2006. 885 [LSP-STITCHING] Ayyangar, A., Kompella, K., and Vasseur, JP., "Label 886 Switched Path Stitching with Generalized MPLS Traffic 887 Engineering", draft-ietf-ccamp-lsp-stitching, (work in 888 progress). 890 11.2. Informative References 892 [RFC2747] Baker, F., Lindell, B., and Talwar, B., "RSVP Cryptographic 893 Authentication", RFC 2747, January 2000. 895 [RFC3097] Braden, R., and Zhang, L., "RSVP Cryptographic 896 Authentication -- Updated Message Type Value", RFC 3097, 897 April 2001. 899 [RFC4090] Ping Pan, et al, "Fast Reroute Extensions to RSVP-TE 900 for LSP Tunnels", RFC 4090, May 2005. 902 [RFC4105] LeRoux, JL., Vasseur, JP., Boyle, J., et al, "Requirements 903 for Inter-Area MPLS Traffic Engineering", RFC 4105, June 904 2005. 906 [RFC4208] G. Swallow et al, "GMPLS UNI: RSVP-TE Support for the 907 Overlay Model", RFC 4208, October 2005. 909 [RFC4216] Zhang, R., et al, "MPLS Inter-Autonomous System (AS) 910 Traffic Engineering (TE) Requirements", RFC 4216, November 911 2005. 913 [RFC4379] Kompella, K. and G. Swallow, "Detecting Multi-Protocol 914 Label Switched (MPLS) Data Plane Failures", RFC 4379, 915 February 2006. 917 [RFC4561] Vasseur, JP., Ali, Z., and Sivabalan, S., "Definition of a 918 Record Route Object (RRO) Node-Id Sub-Object", RFC 4561, 919 June 2006. 921 [RFC4655] Ash, G., Farrel, A., and Vasseur, JP., "Path Computation 922 Element (PCE)-Based Architecture", RFC 4655, August 2006. 924 [RFC4726] Farrel, A., Vasseur, J.P., and Ayyangar, Arthi, "A 925 Framework for Inter-Domain Multiprotocol Label Switching 926 Traffic Engineering", RFC 4726, November 2006. 928 [RFC4736] Vasseur, JP., et al, "Reoptimization of Multiprotocol Label 929 Switching (MPLS) Traffic Engineering (TE) Loosely Routed 930 Label Switched Path (LSP)", RFC 4736, November 2006. 932 [RFC4874] Lee, CY., Farrel, A., and De Cnodder, S., "Exclude Routes - 933 Extension to Resource ReserVation Protocol-Traffic 934 Engineering (RSVP-TE)", RFC 4874, April 2007. 936 [BFD-MPLS] Aggarwal, R., et al, "BFD For MPLS LSPs", 937 draft-ietf-bfd-mpls, (work in progress). 939 [CRANKBACK] Farrel, A., et al, "Crankback Signaling Extensions for 940 MPLS and GMPLS RSVP-TE", draft-ietf-ccamp-crankback, (work 941 in progress). 943 [INTERAS-PING] Nadeau, T., and Swallow, G., "Detecting MPLS Data 944 Plane Failures in Inter-AS and inter-provider Scenarios", 945 draft-nadeau-mpls-interas-lspping, work in progress. 947 [INTER-DOMAIN-PD-PATH-COMP] Vasseur, JP., Ayyangar, A., and 948 Zhang, R., "A Per-domain path computation method for 949 computing Inter-domain Traffic Engineering (TE) Label 950 Switched Paths (LSPs)", draft-ietf-ccamp-inter-domain-pd- 951 path-comp, (work in progress). 953 [MPLS-GMPLS-SEC] Luyuan Fang, et al., "Security Framework for MPLS 954 and GMPLS Networks", draft-fang-mpls-gmpls-security- 955 framework, work in progress. 957 [RSVP-RESTART] Satyanarayana, A., and Rahman, R., "Extensions to 958 GMPLS RSVP Graceful Restart", draft-ietf-ccamp-rsvp- 959 restart-ext, work in progress. 961 [SEGMENT-PROTECTION] Berger, L., et al, "GMPLS Based Segment 962 Recovery", draft-ietf-ccamp-gmpls-segment-recovery, (work 963 in progress). 965 12. Authors' Addresses 967 Adrian Farrel 968 Old Dog Consulting 970 E-mail: adrian@olddog.co.uk 972 Arthi Ayyangar 973 Nuova Systems 974 2600 San Tomas Expressway 975 Santa Clara, CA 95051 976 US 978 Email: arthi@nuovasystems.com 980 Jean Philippe Vasseur 981 Cisco Systems, Inc. 982 300 Beaver Brook Road 983 Boxborough , MA - 01719 984 USA 986 Email: jpv@cisco.com 988 Intellectual Property Statement 990 The IETF takes no position regarding the validity or scope of any 991 Intellectual Property Rights or other rights that might be claimed to 992 pertain to the implementation or use of the technology described in 993 this document or the extent to which any license under such rights 994 might or might not be available; nor does it represent that it has 995 made any independent effort to identify any such rights. Information 996 on the procedures with respect to rights in RFC documents can be 997 found in BCP 78 and BCP 79. 999 Copies of IPR disclosures made to the IETF Secretariat and any 1000 assurances of licenses to be made available, or the result of an 1001 attempt made to obtain a general license or permission for the use of 1002 such proprietary rights by implementers or users of this 1003 specification can be obtained from the IETF on-line IPR repository at 1004 http://www.ietf.org/ipr. 1006 The IETF invites any interested party to bring to its attention any 1007 copyrights, patents or patent applications, or other proprietary 1008 rights that may cover technology that may be required to implement 1009 this standard. Please address the information to the IETF at ietf- 1010 ipr@ietf.org. 1012 Disclaimer of Validity 1014 This document and the information contained herein are provided on an 1015 "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS 1016 OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND 1017 THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS 1018 OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF 1019 THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED 1020 WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. 1022 Copyright Statement 1024 Copyright (C) The IETF Trust (2007). 1026 This document is subject to the rights, licenses and restrictions 1027 contained in BCP 78, and except as set forth therein, the authors 1028 retain all their rights.