idnits 2.17.1 draft-chen-mpls-p2mp-ingress-protection-11.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The document seems to lack an Authors' Addresses Section. ** There are 22 instances of too long lines in the document, the longest one being 1 character in excess of 72. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == The document seems to lack the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords. (The document does seem to have the reference to RFC 2119 which the ID-Checklist requires). -- The document date (February 14, 2014) is 3717 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Missing Reference: 'S' is mentioned on line 131, but not defined == Missing Reference: 'Ra' is mentioned on line 134, but not defined == Missing Reference: 'Rb' is mentioned on line 134, but not defined == Missing Reference: 'L3' is mentioned on line 134, but not defined == Unused Reference: 'RFC1700' is defined on line 1138, but no explicit reference was found in the text == Unused Reference: 'RFC2119' is defined on line 1141, but no explicit reference was found in the text == Unused Reference: 'RFC3692' is defined on line 1144, but no explicit reference was found in the text == Unused Reference: 'RFC2205' is defined on line 1147, but no explicit reference was found in the text == Unused Reference: 'RFC3031' is defined on line 1151, but no explicit reference was found in the text == Unused Reference: 'RFC3209' is defined on line 1154, but no explicit reference was found in the text == Unused Reference: 'RFC3473' is defined on line 1158, but no explicit reference was found in the text == Unused Reference: 'RFC4461' is defined on line 1166, but no explicit reference was found in the text == Unused Reference: 'P2MP-FRR' is defined on line 1175, but no explicit reference was found in the text == Unused Reference: 'RFC2702' is defined on line 1182, but no explicit reference was found in the text == Unused Reference: 'RFC3032' is defined on line 1186, but no explicit reference was found in the text ** Obsolete normative reference: RFC 1700 (Obsoleted by RFC 3232) ** Downref: Normative reference to an Informational RFC: RFC 4461 -- Possible downref: Normative reference to a draft: ref. 'P2MP-FRR' Summary: 4 errors (**), 0 flaws (~~), 17 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Internet Engineering Task Force H. Chen, Ed. 3 Internet-Draft Huawei Technologies 4 Intended status: Standards Track R. Torvi, Ed. 5 Expires: August 18, 2014 Juniper Networks 6 February 14, 2014 8 Extensions to RSVP-TE for LSP Ingress Local Protection 9 draft-chen-mpls-p2mp-ingress-protection-11.txt 11 Abstract 13 This document describes extensions to Resource Reservation Protocol - 14 Traffic Engineering (RSVP-TE) for locally protecting the ingress node 15 of a Traffic Engineered (TE) Label Switched Path (LSP) in a Multi- 16 Protocol Label Switching (MPLS) and Generalized MPLS (GMPLS) network. 18 Status of this Memo 20 This Internet-Draft is submitted to IETF in full conformance with the 21 provisions of BCP 78 and BCP 79. 23 Internet-Drafts are working documents of the Internet Engineering 24 Task Force (IETF). Note that other groups may also distribute 25 working documents as Internet-Drafts. The list of current Internet- 26 Drafts is at http://datatracker.ietf.org/drafts/current/. 28 Internet-Drafts are draft documents valid for a maximum of six months 29 and may be updated, replaced, or obsoleted by other documents at any 30 time. It is inappropriate to use Internet-Drafts as reference 31 material or to cite them other than as "work in progress." 33 This Internet-Draft will expire on August 18, 2014. 35 Copyright Notice 37 Copyright (c) 2014 IETF Trust and the persons identified as the 38 document authors. All rights reserved. 40 This document is subject to BCP 78 and the IETF Trust's Legal 41 Provisions Relating to IETF Documents 42 (http://trustee.ietf.org/license-info) in effect on the date of 43 publication of this document. Please review these documents 44 carefully, as they describe your rights and restrictions with respect 45 to this document. Code Components extracted from this document must 46 include Simplified BSD License text as described in Section 4.e of 47 the Trust Legal Provisions and are provided without warranty as 48 described in the Simplified BSD License. 50 Table of Contents 52 1. Co-authors . . . . . . . . . . . . . . . . . . . . . . . . . . 3 53 2. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 54 2.1. An Example of Ingress Local Protection . . . . . . . . . . 3 55 2.2. Ingress Local Protection with FRR . . . . . . . . . . . . 4 56 3. Ingress Failure Detection . . . . . . . . . . . . . . . . . . 4 57 3.1. Backup and Source Detect Failure . . . . . . . . . . . . . 4 58 3.2. Backup Detects Failure . . . . . . . . . . . . . . . . . . 5 59 3.3. Source Detects Failure . . . . . . . . . . . . . . . . . . 5 60 3.4. Next Hops Detect Failure . . . . . . . . . . . . . . . . . 5 61 3.5. Comparing Different Detection Modes . . . . . . . . . . . 6 62 4. Backup Forwarding State . . . . . . . . . . . . . . . . . . . 6 63 4.1. Forwarding State for Backup LSP . . . . . . . . . . . . . 7 64 4.2. Forwarding State on Next Hops . . . . . . . . . . . . . . 7 65 5. Protocol Extensions . . . . . . . . . . . . . . . . . . . . . 7 66 5.1. INGRESS_PROTECTION Object . . . . . . . . . . . . . . . . 8 67 5.1.1. Subobject: Backup Ingress IPv4/IPv6 Address . . . . . 10 68 5.1.2. Subobject: Ingress IPv4/IPv6 Address . . . . . . . . . 11 69 5.1.3. Subobject: Traffic Descriptor . . . . . . . . . . . . 11 70 5.1.4. Subobject: Label-Routes . . . . . . . . . . . . . . . 12 71 6. Behavior of Ingress Protection . . . . . . . . . . . . . . . . 13 72 6.1. Overview . . . . . . . . . . . . . . . . . . . . . . . . . 13 73 6.1.1. Relay-Message Method . . . . . . . . . . . . . . . . . 13 74 6.1.2. Proxy-Ingress Method . . . . . . . . . . . . . . . . . 13 75 6.1.3. Comparing Two Methods . . . . . . . . . . . . . . . . 14 76 6.2. Ingress Behavior . . . . . . . . . . . . . . . . . . . . . 15 77 6.2.1. Relay-Message Method . . . . . . . . . . . . . . . . . 15 78 6.2.2. Proxy-Ingress Method . . . . . . . . . . . . . . . . . 16 79 6.3. Backup Ingress Behavior . . . . . . . . . . . . . . . . . 17 80 6.3.1. Backup Ingress Behavior in Off-path Case . . . . . . . 17 81 6.3.2. Backup Ingress Behavior in On-path Case . . . . . . . 20 82 6.3.3. Failure Detection . . . . . . . . . . . . . . . . . . 21 83 6.4. Merge Point Behavior . . . . . . . . . . . . . . . . . . . 21 84 6.5. Revertive Behavior . . . . . . . . . . . . . . . . . . . . 22 85 6.5.1. Revert to Primary Ingress . . . . . . . . . . . . . . 22 86 6.5.2. Global Repair by Backup Ingress . . . . . . . . . . . 23 87 7. Security Considerations . . . . . . . . . . . . . . . . . . . 23 88 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 23 89 9. Contributors . . . . . . . . . . . . . . . . . . . . . . . . . 24 90 10. Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . 25 91 11. References . . . . . . . . . . . . . . . . . . . . . . . . . . 25 92 11.1. Normative References . . . . . . . . . . . . . . . . . . . 25 93 11.2. Informative References . . . . . . . . . . . . . . . . . . 26 94 A. Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . 26 96 1. Co-authors 98 Ning So, Autumn Liu, Alia Atlas, Yimin Shen, Fengman Xu, Mehmet Toy, 99 Lei Liu 101 2. Introduction 103 For MPLS LSPs it is important to have a fast-reroute method for 104 protecting its ingress node as well as transit nodes. This is not 105 covered either in the fast-reroute method defined in [RFC4090] or in 106 the P2MP fast-reroute extensions to fast-reroute in [RFC4875]. 108 An alternate approach to local protection (fast-reroute) is to use 109 global protection and set up a second backup LSP (whether P2MP or 110 P2P) from a backup ingress to the egresses. The main disadvantage of 111 this is that the backup LSP may reserve additional network bandwidth. 113 This specification defines a simple extension to RSVP-TE for local 114 protection of the ingress node of a P2MP or P2P LSP. 116 2.1. An Example of Ingress Local Protection 118 Figure 1 shows an example of using a backup P2MP LSP to locally 119 protect the ingress of a primary P2MP LSP, which is from ingress R1 120 to three egresses: L1, L2 and L3. The backup LSP is from backup 121 ingress Ra to the next hops R2 and R4 of ingress R1. 123 [R2]******[R3]*****[L1] 124 * | **** Primary LSP 125 * | ---- Backup LSP 126 * / .... BFD Session 127 * / $ Link 128 [R1]*******[R4]****[R5]*****[L2] $ 129 $ . / / * $ 130 $ . / / * 131 [S] . / / * 132 $ . / / * 133 $ ./ / * 134 [Ra]----[Rb] [L3] 136 Figure 1: Backup P2MP LSP for Locally Protecting Ingress 138 Source S may send the traffic simultaneously to both primary ingress 139 R1 and backup ingress Ra. R1 imports the traffic into the primary 140 LSP. Ra normally does not put the traffic into the backup LSP. 142 Ra should be able to detect the failure of R1 and switch the traffic 143 within 10s of ms. The exact method by which Ra does so is out of 144 scope. Different options are discussed in this draft. 146 When Ra detects the failure of R1, it imports the traffic from S into 147 the backup LSP to R1's next hops R2 and R4, where the traffic is 148 merged into the primary LSP, and then sent to egresses L1, L2 and L3. 150 Note that the backup egress must be one logical hop away from the 151 ingress. A logical hop is a direct link or a tunnel such as a GRE 152 tunnel, over which RSVP-TE messages may be exchanged. 154 2.2. Ingress Local Protection with FRR 156 Through using the ingress local protection and the FRR, we can 157 locally protect the ingress node, all the links and the intermediate 158 nodes of an LSP. The traffic switchover time is within tens of 159 milliseconds whenever the ingress, any of the links and the 160 intermediate nodes of the LSP fails. 162 The ingress node of the LSP can be locally protected through using 163 the ingress local protection. All the links and all the intermediate 164 nodes of the LSP can be locally protected through using the FRR. 166 3. Ingress Failure Detection 168 Exactly how the failure of the ingress (e.g. R1 in Figure 1) is 169 detected is out of scope for this document. However, it is necessary 170 to discuss different modes for detecting the failure because they 171 determine what must be signaled and what is the required behavior for 172 the traffic source, backup ingress, and merge-points. 174 3.1. Backup and Source Detect Failure 176 Backup and Source Detect Failure or Backup-Source-Detect for short 177 means that both the backup ingress and the source are concurrently 178 responsible for detecting the failures of the primary ingress. 180 In normal operations, the source sends the traffic to the primary 181 ingress. It switches the traffic to the backup ingress when it 182 detects the failure of the primary ingress. 184 The backup ingress does not import any traffic from the source into 185 the backup LSP in normal operations. When it detects the failure of 186 the primary ingress, it imports the traffic from the source into the 187 backup LSP to the next hops of the primary ingress, where the traffic 188 is merged into the primary LSP. 190 Note that the source may locally distinguish between the failure of 191 the primary ingress and that of the link between the source and the 192 primary ingress. When the source detects the failure of the link, it 193 may continue to send the traffic to the primary ingress via another 194 link between the source and the primary ingress if there is one. 196 3.2. Backup Detects Failure 198 Backup Detects Failure or Backup-Detect means that the backup ingress 199 is responsible for detecting the failure of the primary ingress of an 200 LSP. The source SHOULD send the traffic simultaneously to both the 201 primary ingress and backup ingress. 203 The backup ingress does not import any traffic from the source into 204 the backup LSP in normal operations. When it detects the failure of 205 the primary ingress, it imports the traffic from the source into the 206 backup LSP to the next hops of the primary ingress, where the traffic 207 is merged into the primary LSP. 209 Note that the backup ingress may locally distinguish between the 210 failure of the primary ingress and that of the link between the 211 backup ingress and the primary ingress through two BFDs between the 212 backup ingress and the primary ingress. One is through the link, and 213 the other is not. If the first BFD is down and the second is up, the 214 link fails and the primary ingress does not. 216 3.3. Source Detects Failure 218 Source Detects Failure or Source-Detect means that the source is 219 responsible for detecting the failure of the primary ingress of an 220 LSP. The backup ingress is ready to import the traffic from the 221 source into the backup LSP after the backup LSP is up. 223 In normal operations, the source sends the traffic to the primary 224 ingress. When the source detects the failure of the primary ingress, 225 it switches the traffic to the backup ingress, which delivers the 226 traffic to the next hops of the primary ingress through the backup 227 LSP, where the traffic is merged into the primary LSP. 229 3.4. Next Hops Detect Failure 231 Next Hops Detect Failure or Next-Hop-Detect means that each of the 232 next hops of the primary ingress of an LSP is responsible for 233 detecting the failure of the primary ingress. 235 In normal operations, the source sends the traffic to both the 236 primary ingress and the backup ingress. Both ingresses deliver the 237 traffic to the next hops of the primary ingress. Each of the next 238 hops selects the traffic from the primary ingress and sends the 239 traffic to the destinations of the LSP. 241 When each of the next hops detects the failure of the primary 242 ingress, it switches to receive the traffic from the backup ingress 243 and then sends the traffic to the destinations. 245 3.5. Comparing Different Detection Modes 247 +----------+--------------+----------------+--------+-------------------+ 248 |\_Behavior|Traffic Always|Backup Ingress |Next-Hop|Incorrect Failure | 249 | \______ |Sent to |Activation of |Select |Detection Cause | 250 |Detection\|Backup Ingress|Forwarding Entry|Stream |Traffic Duplication| 251 |Mode | | | |(Ingress does FRR) | 252 +----------+--------------+----------------+--------+-------------------+ 253 |Backup- | | | | | 254 |Source- | No | Yes | No | No | 255 |Detect | | | | | 256 +----------+--------------+----------------+--------+-------------------+ 257 |Backup- | Yes | Yes | No | Yes | 258 |Detect | | | | | 259 +----------+--------------+----------------+--------+-------------------+ 260 |Source- | No | No | No | No | 261 |Detect | | (Always Active)| | | 262 +----------+--------------+----------------+--------+-------------------+ 263 |Next-Hop- | Yes | No | Yes |(If Ingress-Next- | 264 |Detect | | (Always Active)| |Hop link fails, | 265 | | | | |stream selection | 266 | | | | |at Next-Next-Hops | 267 | | | | |can mitigate) | 268 +----------+--------------+----------------+--------+-------------------+ 270 A primary goal of failure detection and FRR protection is to avoid 271 traffic duplication, particularly along the P2MP. A reasonable 272 assumption when this ingress protection is in use is that the ingress 273 is also trying to provide link and node protection. When the failure 274 cannot be accurately identified as that of the ingress, this can lead 275 to the ingress sending traffic on bypass to the next-next-hop(s) for 276 node-protection while the backup ingress is sending traffic to its 277 next-hop(s) if Next-Hop-Detect mode is used. RSVP Path messages from 278 the bypass may help to eventually resolve this by removing the 279 forwarding entry for receiving the traffic from the next-hop. 281 4. Backup Forwarding State 283 Before the primary ingress fails, the backup ingress is responsible 284 for creating the necessary backup LSPs to the next hops of the 285 ingress. These LSPs might be multiple bypass P2P LSPs that avoid the 286 ingress. Alternately, the backup ingress could choose to use a 287 single backup P2MP LSP as a bypass or detour to protect the primary 288 ingress of a primary P2MP LSP. 290 The backup ingress may be off-path or on-path of an LSP. When a 291 backup ingress is not any node of the LSP, we call the backup ingress 292 is off-path. When a backup ingress is a next-hop of the primary 293 ingress of the LSP, we call it is on-path. If the backup ingress is 294 on-path, the primary forwarding state associated with the primary LSP 295 SHOULD be clearly separated from the backup LSP(s) state. 296 Specifically in Backup-Detect mode, the backup ingress will receive 297 traffic from the primary ingress and from the traffic source; only 298 the former should be forwarded until failure is detected even if the 299 backup ingress is the only next-hop. 301 4.1. Forwarding State for Backup LSP 303 A forwarding entry for a backup LSP is created on the backup ingress 304 after the LSP is set up. Depending on the failure-detection mode 305 (e.g., source-detect), it may be used to forward received traffic or 306 simply be inactive (e.g., backup-detect) until required. In either 307 case, when the primary ingress fails, this forwarding entry is used 308 to import the traffic into the backup LSP to the next hops of the 309 primary ingress, where the traffic is merged into the primary LSP. 311 The forwarding entry for a backup LSP is a local implementation 312 issue. In one device, it may have an inactive flag. This inactive 313 forwarding entry is not used to forward any traffic normally. When 314 the primary ingress fails, it is changed to active, and thus the 315 traffic from the source is imported into the backup LSP. 317 4.2. Forwarding State on Next Hops 319 When Next-Hop-Detect is used, a forwarding entry for a backup LSP is 320 created on each of the next hops of the primary ingress of the LSP. 321 This forwarding entry does not forward any traffic normally. When 322 the primary ingress fails, it is used to import/select the traffic 323 from the backup LSP into the primary LSP. 325 5. Protocol Extensions 327 A new object INGRESS_PROTECTION is defined for signaling ingress 328 local protection. It is backward compatible. 330 5.1. INGRESS_PROTECTION Object 332 The INGRESS_PROTECTION object with the FAST_REROUTE object in a PATH 333 message is used to control the backup for protecting the primary 334 ingress of a primary LSP. The primary ingress MUST insert this 335 object into the PATH message to be sent to the backup ingress for 336 protecting the primary ingress. It has the following format: 338 Class-Num = TBD C-Type = TBD 340 0 1 2 3 341 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 1 342 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 343 | Length (bytes) | Class-Num | C-Type | 344 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 345 | Secondary LSP ID | Flags | Options | DM | 346 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 347 ~ (Subobjects) ~ 348 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 350 Flags 351 0x01 Ingress local protection available 352 0x02 Ingress local protection in use 353 0x04 Bandwidth protection 355 Options 356 0x01 Revert to Ingress 357 0x02 Ingress-Proxy/Relay-Message 358 0x04 P2MP Backup 360 DM (Detection Mode) 361 0x00 Backup-Source-Detect 362 0x01 Backup-Detect 363 0x02 Source-Detect 364 0x03 Next-Hop-Detect 366 For backward compatible, the two high-order bits of the Class-Num in 367 the object are set as follows: 369 o Class-Num = 0bbbbbbb for the object in a message not on LSP path. 370 The entire message should be rejected and an "Unknown Object 371 Class" error returned. 373 o Class-Num = 10bbbbbb for the object in a message on LSP path. The 374 node should ignore the object, neither forwarding it nor sending 375 an error message. 377 The Secondary LSP ID in the object is an LSP ID that the primary 378 ingress has allocated for a protected LSP tunnel. The backup ingress 379 will use this LSP ID to set up a new LSP from the backup ingress to 380 the destinations of the protected LSP tunnel. This allows the new 381 LSP to share resources with the old one. 383 The flags are used to communicate status information from the backup 384 ingress to the primary ingress. 386 o Ingress local protection available: The backup ingress sets this 387 flag after backup LSPs are up and ready for locally protecting the 388 primary ingress. The backup ingress sends this to the primary 389 ingress to indicate that the primary ingress is locally protected. 391 o Ingress local protection in use: The backup ingress sets this flag 392 when it detects a failure in the primary ingress. The backup 393 ingress keeps it and does not send it to the primary ingress since 394 the primary ingress is down. 396 o Bandwidth protection: The backup ingress sets this flag if the 397 backup LSPs guarantee to provide desired bandwidth for the 398 protected LSP against the primary ingress failure. 400 The options are used by the primary ingress to specify the desired 401 behavior to the backup ingress and next-hops. 403 o Revert to Ingress: The primary ingress sets this option indicating 404 that the traffic for the primary LSP successfully re-signaled will 405 be switched back to the primary ingress from the backup ingress 406 when the primary ingress is restored. 408 o Ingress-Proxy/Relay-Message: This option is set to one indicating 409 that Ingress-Proxy method is used. It is set to zero indicating 410 that Relay-Message method is used. 412 o P2MP Backup: This option is set to ask for the backup ingress to 413 use P2MP backup LSP to protect the primary ingress. Note that one 414 spare bit of the flags in the FAST-REROUTE object can be used to 415 indicate whether P2MP or P2P backup LSP is desired for protecting 416 an ingress and intermediate node. 418 The DM (Detection Mode) is used by the primary ingress to specify a 419 desired failure detection mode. 421 o Backup-Source-Detect (0x00): The backup ingress and the source are 422 concurrently responsible for detecting the failure involving the 423 primary ingress and redirecting the traffic. 425 o Backup-Detect (0x01): The backup ingress is responsible for 426 detecting the failure and redirecting the traffic. 428 o Source-Detect (0x02): The source is responsible for detecting the 429 failure and redirecting the traffic. 431 o Next-Hop-Detect (0x03): The next hops of the primary ingress are 432 responsible for detecting the failure and selecting the traffic. 434 The INGRESS_PROTECTION object may contain some of the sub objects 435 described below. 437 5.1.1. Subobject: Backup Ingress IPv4/IPv6 Address 439 When the primary ingress of a protected LSP sends a PATH message with 440 an INGRESS_PROTECTION object to the backup ingress, the object may 441 have a Backup Ingress IPv4/IPv6 Address sub object containing an 442 IPv4/IPv6 address belonging to the backup ingress. The formats of 443 the sub object for Backup Ingress IPv4/IPv6 Address is given below: 445 0 1 2 3 446 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 1 447 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 448 | Type | Length | Reserved (zeros) | 449 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 450 | IPv4 address | 451 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 453 Type: TBD-1 Backup Ingress IPv4 Address 454 Length: Total length of the subobject in bytes, including 455 the Type and Length fields. The Length is always 8. 456 Reserved: Reserved two bytes are set to zeros. 457 IPv4 address: A 32-bit unicast, host address. 459 0 1 2 3 460 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 1 461 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 462 | Type | Length | Reserved (zeros) | 463 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 464 | | 465 ~ IPv6 address (16 bytes) ~ 466 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 468 Type: TBD-2 Backup Ingress IPv6 Address 469 Length: Total length of the subobject in bytes, including 470 the Type and Length fields. The Length is always 20. 471 Reserved: Reserved two bytes are set to zeros. 472 IPv6 address: A 128-bit unicast, host address. 474 5.1.2. Subobject: Ingress IPv4/IPv6 Address 476 The INGRESS_PROTECTION object in a PATH message from the primary 477 ingress to the backup ingress may have an Ingress IPv4/IPv6 Address 478 sub object containing an IPv4/IPv6 address belonging to the primary 479 ingress. The sub object has the following format: 481 0 1 2 3 482 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 1 483 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 484 | Type | Length | Reserved (zeros) | 485 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 486 | IPv4 address | 487 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 489 Type: TBD-3 Ingress IPv4 Address 490 Length: Total length of the subobject in bytes, including 491 the Type and Length fields. The Length is always 8. 492 Reserved: Reserved two bytes are set to zeros. 493 IPv4 address: A 32-bit unicast, host address. 495 0 1 2 3 496 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 1 497 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 498 | Type | Length | Reserved (zeros) | 499 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 500 | | 501 ~ IPv6 address (16 bytes) ~ 502 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 504 Type: TBD-4 Backup Ingress IPv6 Address 505 Length: Total length of the subobject in bytes, including 506 the Type and Length fields. The Length is always 20. 507 Reserved: Reserved two bytes are set to zeros. 508 IPv6 address: A 128-bit unicast, host address. 510 5.1.3. Subobject: Traffic Descriptor 512 The INGRESS_PROTECTION object in a PATH message from the primary 513 ingress to the backup ingress may have a Traffic Descriptor sub 514 object describing the traffic to be mapped to the backup LSP on the 515 backup ingress for locally protecting the primary ingress. The sub 516 object has the following format: 518 0 1 2 3 519 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 1 520 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 521 | Type | Length | Reserved (zeros) | 522 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 523 | Traffic Element 1 | 524 ~ ~ 525 | Traffic Element n | 526 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 528 Type: TBD-5/TBD-6/TBD-7 Interface/IPv4/6 Prefix 529 Length: Total length of the subobject in bytes, including 530 the Type and Length fields. 531 Reserved: Reserved two bytes are set to zeros. 533 The Traffic Descriptor sub object may contain multiple Traffic 534 Elements of same type as follows. 536 o Interface Traffic (Type TBD-5): Each of the Traffic Elements is a 537 32 bit index of an interface, from which the traffic is imported 538 into the backup LSP. 540 o IPv4/6 Prefix Traffic (Type TBD-6/TBD-7): Each of the Traffic 541 Elements is an IPv4/6 prefix, containing an 8-bit prefix length 542 followed by an IPv4/6 address prefix, whose length, in bits, was 543 specified by the prefix length, padded to a byte boundary. 545 5.1.4. Subobject: Label-Routes 547 The INGRESS_PROTECTION object in a PATH message from the primary 548 ingress to the backup ingress will have a Label-Routes sub object 549 containing the labels and routes that the next hops of the ingress 550 use. The sub object has the following format: 552 0 1 2 3 553 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 1 554 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 555 | Type | Length | Reserved (zeros) | 556 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 557 ~ (Subobjects) ~ 558 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 560 Type: TBD-8 Label-Routes 561 Length: Total length of the subobject in bytes, including 562 the Type and Length fields. 563 Reserved: Reserved two bytes are set to zeros. 565 The Subobjects in the Label-Routes are copied from the Subobjects in 566 the RECORD_ROUTE objects contained in the RESV messages that the 567 primary ingress receives from its next hops for the protected LSP. 568 They MUST contain the first hops of the LSP, each of which is paired 569 with its label. 571 6. Behavior of Ingress Protection 573 6.1. Overview 575 There are four parts of ingress protection: 1) setting up the 576 necessary backup LSP forwarding state; 2) identifying the failure and 577 providing the fast repair (as discussed in Sections 2 and 3); 3) 578 maintaining the RSVP-TE control plane state until a global repair can 579 be done; and 4) performing the global repair(see Section 5.5). 581 There are two different proposed signaling approaches to obtain 582 ingress protection. They both use the same new INGRESS-PROTECTION 583 object. The object is sent in both PATH and RESV messages. 585 6.1.1. Relay-Message Method 587 The primary ingress relays the information for ingress protection of 588 an LSP to the backup ingress via PATH messages. Once the LSP is 589 created, the ingress of the LSP sends the backup ingress a PATH 590 message with an INGRESS-PROTECTION object with Label-Routes 591 subobject, which is populated with the next-hops and labels. This 592 provides sufficient information for the backup ingress to create the 593 appropriate forwarding state and backup LSP(s). 595 The ingress also sends the backup ingress all the other PATH messages 596 for the LSP with an empty INGRESS-PROTECTION object. Thus, the 597 backup ingress has access to all the PATH messages needed for 598 modification to be sent to refresh control-plane state after a 599 failure. 601 The advantages of this method include: 1) the primary LSP is 602 independent of the backup ingress; 2) simple; 3) less configuration; 603 and 4) less control traffic. 605 6.1.2. Proxy-Ingress Method 607 Conceptually, a proxy ingress is created that starts the RSVP 608 signaling. The explicit path of the LSP goes from the proxy ingress 609 to the backup ingress and then to the real ingress. The behavior and 610 signaling for the proxy ingress is done by the real ingress; the use 611 of a proxy ingress address avoids problems with loop detection. 613 [ traffic source ] *** Primary LSP 614 $ $ --- Backup LSP 615 $ $ $$ Link 616 $ $ 617 [ proxy ingress ] [ backup ] 618 [ & ingress ] | 619 * | 620 *****[ MP ]----| 622 Figure 2: Example Protected LSP with Proxy Ingress Node 624 The backup ingress must know the merge points or next-hops and their 625 associated labels. This is accomplished by having the RSVP PATH and 626 RESV messages go through the backup ingress, although the forwarding 627 path need not go through the backup ingress. If the backup ingress 628 fails, the ingress simply removes the INGRESS-PROTECTION object and 629 forwards the PATH messages to the LSP's next-hop(s). If the ingress 630 has its LSP configured for ingress protection, then the ingress can 631 add the backup ingress and itself to the ERO and start forwarding the 632 PATH messages to the backup ingress. 634 Slightly different behavior can apply for the on-path and off-path 635 cases. In the on-path case, the backup ingress is a next hop node 636 after the ingress for the LSP. In the off-path, the backup ingress 637 is not any next-hop node after the ingress for all associated sub- 638 LSPs. 640 The key advantage of this approach is that it minimizes the special 641 handling code requires. Because the backup ingress is on the 642 signaling path, it can receive various notifications. It easily has 643 access to all the PATH messages needed for modification to be sent to 644 refresh control-plane state after a failure. 646 6.1.3. Comparing Two Methods 648 +-------+-----------+------+--------+-----------------+---------+ 649 | |Primary LSP|Simple|Config |PATH Msg from |Reuse | 650 |Method |Depends on | |Proxy- |Backup to primary|Some of | 651 | |Backup | |Ingress-|RESV Msg from |Existing | 652 | |Ingress | |ID |Primary to backup|Functions| 653 +-------+-----------+------+--------+-----------------+---------+ 654 |Relay- | No |Yes | No | No | Yes- | 655 |Message| | | | | | 656 +-------+-----------+------+--------+-----------------+---------+ 657 |Proxy- | Yes |Yes- | Yes | Yes | Yes | 658 |Ingress| | | | | | 659 +-------+-----------+------+--------+-----------------+---------+ 661 6.2. Ingress Behavior 663 The primary ingress must be configured with four pieces of 664 information for ingress protection. 666 o Backup Ingress Address: The primary ingress must know an IP 667 address for it to be included in the INGRESS-PROTECTION object. 669 o Failure Detection Mode: The primary ingress must know what failure 670 detection mode is to be used: Backup-Source-Detect, Backup-Detect, 671 Source-Detect, or Next-Hop-Detect. 673 o Proxy-Ingress-Id (only needed for Proxy-Ingress Method): The 674 Proxy-Ingress-Id is only used in the Record Route Object for 675 recording the proxy-ingress. If no proxy-ingress-id is specified, 676 then a local interface address that will not otherwise be included 677 in the Record Route Object can be used. A similar technique is 678 used in [RFC4090 Sec 6.1.1]. 680 o Application Traffic Identifier: The primary ingress and backup 681 ingress must both know what application traffic should be directed 682 into the LSP. If a list of prefixes in the Traffic Descriptor 683 sub-object will not suffice, then a commonly understood 684 Application Traffic Identifier can be sent between the primary 685 ingress and backup ingress. The exact meaning of the identifier 686 should be configured similarly at both the primary ingress and 687 backup ingress. The Application Traffic Identifier is understood 688 within the unique context of the primary ingress and backup 689 ingress. 691 With this additional information, the primary ingress can create and 692 signal the necessary RSVP extensions to support ingress protection. 694 6.2.1. Relay-Message Method 696 To protect the ingress of an LSP, the ingress does the following 697 after the LSP is up. 699 1. Select a PATH message. 701 2. If the backup ingress is off-path, then send the backup ingress a 702 PATH message with the content from the selected PATH message and 703 an INGRESS-PROTECTION object; else (the backup ingress is a next 704 hop, i.e., on-path case) add an INGRESS-PROTECTION object into 705 the existing PATH message to the backup ingress (i.e., the next 706 hop). The INGRESS-PROTECTION object contains the Traffic- 707 Descriptor sub-object, the Backup Ingress Address sub-object and 708 the Label-Routes sub-object. The DM (Detection Mode) in the 709 object is set to indicate the failure detection mode desired. 710 The flags is set to indicate whether a Backup P2MP LSP is 711 desired. If not yet allocated, allocate a second LSP-ID to be 712 used in the INGRESS-PROTECTION object. The Label-Routes sub- 713 object contains the next-hops of the ingress and their labels. 715 3. For each of the other PATH messages, if the node to which the 716 message is sent is not the backup ingress, then send the backup 717 ingress a PATH message with the content copied from the message 718 to the node and an empty INGRESS-PROTECTION object; else send the 719 node the message with an empty INGRESS-PROTECTION object. 721 6.2.2. Proxy-Ingress Method 723 The primary ingress is responsible for starting the RSVP signaling 724 for the proxy-ingress node. To do this, the following is done for 725 the RSVP PATH message. 727 1. Compute the EROs for the LSP as normal for the ingress. 729 2. If the selected backup ingress node is not the first node on the 730 path (for all sub-LSPs), then insert at the beginning of the ERO 731 first the backup ingress node and then the ingress node. 733 3. In the PATH RRO, instead of recording the ingress node's address, 734 replace it with the Proxy-Ingress-Id. 736 4. Leave the HOP object populated as usual with information for the 737 ingress-node. 739 5. Add the INGRESS-PROTECTION object to the PATH message. Allocate 740 a second LSP-ID to be used in the INGRESS-PROTECTION object. 741 Include the Backup Ingress Address (IPv4 or IPv6) sub-object and 742 the Traffic-Descriptor sub-object. Set the control-options to 743 indicate the failure detection mode desired. Set or clear the 744 flag indicating that a Backup P2MP LSP is desired. 746 6. Optionally, add the FAST-REROUTE object [RFC4090] to the Path 747 message. Indicate whether one-to-one backup is desired. 748 Indicate whether facility backup is desired. 750 7. The RSVP PATH message is sent to the backup node as normal. 752 If the ingress detects that it can't communicate with the backup 753 ingress, then the ingress should instead send the PATH message to the 754 next-hop indicated in the ERO computed in step 1. Once the ingress 755 detects that it can communicate with the backup ingress, the ingress 756 SHOULD follow the steps 1-7 to obtain ingress failure protection. 758 When the ingress node receives an RSVP PATH message with an INGRESS- 759 PROTECTION object and the object specifies that node as the ingress 760 node and the PHOP as the backup ingress node, the ingress node SHOULD 761 check the Failure Scenario specified in the INGRESS-PROTECTION object 762 and, if it is not the Next-Hop-Detect, then the ingress node SHOULD 763 remove the INGRESS-PROTECTION object from the PATH message before 764 sending it out. Additionally, the ingress node must store that it 765 will install ingress forwarding state for the LSP rather than 766 midpoint forwarding. 768 When an RSVP RESV message is received by the ingress, it uses the 769 NHOP to determine whether the message is received from the backup 770 ingress or from a different node. The stored associated PATH message 771 contains an INGRESS-PROTECTION object that identifies the backup 772 ingress node. If the RESV message is not from the backup node, then 773 ingress forwarding state should be set up, and the INGRESS-PROTECTION 774 object MUST be added to the RESV before it is sent to the NHOP, which 775 should be the backup node. If the RESV message is from the backup 776 node, then the LSP should be considered available for use. 778 If the backup ingress node is on the forwarding path, then a RESV is 779 received with an INGRESS-PROTECTION object and an NHOP that matches 780 the backup ingress. In this case, the ingress node's address will 781 not appear after the backup ingress in the RRO. The ingress node 782 should set up ingress forwarding state, just as is done if the LSP 783 weren't ingress-node protected. 785 6.3. Backup Ingress Behavior 787 An LER determines that the ingress local protection is requested for 788 an LSP if the INGRESS_PROTECTION object is included in the PATH 789 message it receives for the LSP. The LER can further determine that 790 it is the backup ingress if one of its addresses is in the Backup 791 Ingress Address sub-object of the INGRESS-PROTECTION object. The LER 792 as the backup ingress will assume full responsibility of the ingress 793 after the primary ingress fails. In addition, the LER determines 794 that it is off-path if it is not a next hop of the primary ingress. 796 6.3.1. Backup Ingress Behavior in Off-path Case 798 The backup ingress considers itself as a PLR and the primary ingress 799 as its next hop and provides a local protection for the primary 800 ingress. It behaves very similarly to a PLR providing fast-reroute 801 where the primary ingress is considered as the failure-point to 802 protect. Where not otherwise specified, the behavior given in 803 [RFC4090] for a PLR should apply. 805 The backup ingress SHOULD follow the control-options specified in the 806 INGRESS-PROTECTION object and the flags and specifications in the 807 FAST-REROUTE object. This applies to providing a P2MP backup if the 808 "P2MP backup" is set, a one-to-one backup if "one-to-one desired" is 809 set, facility backup if the "facility backup desired" is set, and 810 backup paths that support the desired bandwidth, and administrative- 811 colors that are requested. 813 If multiple INGRESS-PROTECTION objects have been received via 814 multiple PATH messages for the same LSP, then the most recent one 815 that specified a Traffic-Descriptor sub-object MUST be the one used. 817 The backup ingress creates the appropriate forwarding state based on 818 failure detection mode specified. For the Source-Detect and Next- 819 Hop-Detect, this means that the backup ingress forwards any received 820 identified traffic into the backup LSP tunnel(s) to the merge 821 point(s). For the Backup-Detect and Backup-Source-Detect, this means 822 that the backup ingress creates state to quickly determine the 823 primary ingress has failed and switch to sending any received 824 identified traffic into the backup LSP tunnel(s) to the merge 825 point(s). 827 When the backup ingress sends a RESV message to the primary ingress, 828 it should add an INGRESS-PROTECTION object into the message. It 829 SHOULD set or clear the flags in the object to report "Ingress local 830 protection available", "Ingress local protection in use", and 831 "bandwidth protection". 833 If the backup ingress doesn't have a backup LSP tunnel to all the 834 merge points, it SHOULD clear "Ingress local protection available". 835 [Editor Note: It is possible to indicate the number or which are 836 unprotected via a sub-object if desired.] 838 When the primary ingress fails, the backup ingress redirects the 839 traffic from a source into the backup P2P LSPs or the backup P2MP LSP 840 transmitting the traffic to the next hops of the primary ingress, 841 where the traffic is merged into the protected LSP. 843 In this case, the backup ingress keeps the PATH message with the 844 INGRESS_PROTECTION object received from the primary ingress and the 845 RESV message with the INGRESS_PROTECTION object to be sent to the 846 primary ingress. The backup ingress sets the "local protection in 847 use" flag in the RESV message, indicating that the backup ingress is 848 actively redirecting the traffic into the backup P2P LSPs or the 849 backup P2MP LSP for locally protecting the primary ingress failure. 851 Note that the RESV message with this piece of information will not be 852 sent to the primary ingress because the primary ingress has failed. 854 If the backup ingress has not received any PATH message from the 855 primary ingress for an extended period of time (e.g., a cleanup 856 timeout interval) and a confirmed primary ingress failure did not 857 occur, then the standard RSVP soft-state removal SHOULD occur. The 858 backup ingress SHALL remove the state for the PATH message from the 859 primary ingress, and tear down the one-to-one backup LSPs for 860 protecting the primary ingress if one-to-one backup is used or unbind 861 the facility backup LSPs if facility backup is used. 863 When the backup ingress receives a PATH message from the primary 864 ingress for locally protecting the primary ingress of a protected 865 LSP, it checks to see if any critical information has been changed. 866 If the next hops of the primary ingress are changed, the backup 867 ingress SHALL update its backup LSP(s). 869 6.3.1.1. Relay-Message Method 871 When the backup ingress receives a PATH message with the INGRESS- 872 PROTECTION object, it examines the object to learn what traffic 873 associated with the LSP and what ingress failure detection mode is 874 being used. It determines the next-hops to be merged to by examining 875 the Label-Routes sub-object in the object. If the Traffic-Descriptor 876 sub-object isn't included, this object is considered "empty". 878 The backup ingress stores the PATH message received from the primary 879 ingress, but does NOT forward it. 881 The backup ingress MUST respond with a RESV to the PATH message 882 received from the primary ingress. If the INGRESS-PROTECTION object 883 is not "empty", the backup ingress SHALL send the RESV message with 884 the state indicating protection is available after the backup LSP(s) 885 are successfully established. 887 6.3.1.2. Proxy-Ingress Method 889 The backup ingress determines the next-hops to be merged to by 890 collecting the set of the pair of (IPv4/IPv6 sub-object, Label sub- 891 object) from the Record Route Object of each RESV that are closest to 892 the top and not the Ingress router; this should be the second to the 893 top pair. If a Label-Routes sub-object is included in the INGRESS- 894 PROTECTION object, the included IPv4/IPv6 sub-objects are used to 895 filter the set down to the specific next-hops where protection is 896 desired. A RESV message must have been received before the Backup 897 Ingress can create or select the appropriate backup LSP. 899 When the backup ingress receives a PATH message with the INGRESS- 900 PROTECTION object, the backup ingress examines the object to learn 901 what traffic associated with the LSP and what ingress failure 902 detection mode is being used. The backup ingress forwards the PATH 903 message to the ingress node with the normal RSVP changes. 905 When the backup ingress receives a RESV message with the INGRESS- 906 PROTECTION object, the backup ingress records an IMPLICIT-NULL label 907 in the RRO. Then the backup ingress forwards the RESV message to the 908 ingress node, which is acting for the proxy ingress. 910 6.3.2. Backup Ingress Behavior in On-path Case 912 An LER as the backup ingress determines that it is on-path if one of 913 its addresses is a next hop of the primary ingress and the primary 914 ingress is not its next hop via checking the PATH message with the 915 INGRESS_PROTECTION object received from the primary ingress. The LER 916 on-path sends the corresponding PATH messages without any 917 INGRESS_PROTECTION object to its next hops. It creates a number of 918 backup P2P LSPs or a backup P2MP LSP from itself to the other next 919 hops (i.e., the next hops other than the backup ingress) of the 920 primary ingress. The other next hops are from the Label-Routes sub 921 object. 923 It also creates a forwarding entry, which sends/multicasts the 924 traffic from the source to the next hops of the backup ingress along 925 the protected LSP when the primary ingress fails. The traffic is 926 described by the Traffic-Descriptor. 928 After the forwarding entry is created, all the backup P2P LSPs or the 929 backup P2MP LSP is up and associated with the protected LSP, the 930 backup ingress sends the primary ingress the RESV message with the 931 INGRESS_PROTECTION object containing the state of the local 932 protection such as "local protection available" flag set to one, 933 which indicates that the primary ingress is locally protected. 935 When the primary ingress fails, the backup ingress sends/multicasts 936 the traffic from the source to its next hops along the protected LSP 937 and imports the traffic into each of the backup P2P LSPs or the 938 backup P2MP LSP transmitting the traffic to the other next hops of 939 the primary ingress, where the traffic is merged into protected LSP. 941 During the local repair, the backup ingress continues to send the 942 PATH messages to its next hops as before, keeps the PATH message with 943 the INGRESS_PROTECTION object received from the primary ingress and 944 the RESV message with the INGRESS_PROTECTION object to be sent to the 945 primary ingress. It sets the "local protection in use" flag in the 946 RESV message. 948 6.3.3. Failure Detection 950 Failure detection happens much faster than RSVP, whether via a link- 951 level notification or BFD. As discussed, there are different modes 952 for detecting it. The backup ingress MUST have properly set up its 953 forwarding state to either always forward the specified traffic into 954 the backup LSP(s) for the Source-Detect and Next-Hop-Detect modes or 955 to swap from discarding to forwarding when a failure is detected for 956 the Backup-Source-Detect and Backup-Detect modes. 958 For facility backup LSPs, the correct inner MPLS label to use must be 959 determined. For the ingress-proxy method, that MPLS label comes 960 directly from the RRO of the RESV. For the relay-message method, 961 that MPLS label comes from the Label-Routes sub-object in the non- 962 empty INGRESS-PROTECTION object. 964 As described in [RFC4090], it is necessary to refresh the PATH 965 messages via the backup LSP(s). The Backup Ingress MUST wait to 966 refresh the backup PATH messages until it can accurately detect that 967 the ingress node has failed. An example of such an accurate 968 detection would be that the IGP has no bi-directional links to the 969 ingress node and the last change was long enough in the past that 970 changes should have been received (i.e., an IGP network convergence 971 time or approximately 2-3 seconds) or a BFD session to the primary 972 ingress' loopback address has failed and stayed failed after the 973 network has reconverged. 975 As described in [RFC4090 Section 6.4.3], the backup ingress, acting 976 as PLR, SHOULD modify - including removing any INGRESS-PROTECTION and 977 FAST-REROUTE objects - and send any saved PATH messages associated 978 with the primary LSP. 980 6.4. Merge Point Behavior 982 An LSR that is serving as a Merge Point may need to support the 983 INGRESS-PROTECTION object and functionality defined in this 984 specification if the LSP is ingress-protected where the failure 985 scenario is Next-Hop-Detect. An LSR can determine that it must be a 986 merge point if it is not the ingress, it is not the backup ingress 987 (determined by examining the Backup Ingress Address (IPv4 or IPv6) 988 sub-object in the INGRESS-PROTECTION object), and the PHOP is the 989 ingress node. 991 In that case, when the LSR receives a PATH message with an INGRESS- 992 PROTECTION object, the LSR MUST remove the INGRESS-PROTECTION object 993 before forwarding on the PATH message. If the failure scenario 994 specified is Next-Hop-Detect, the MP must connect up the fast-failure 995 detection (as configured) to accepting backup traffic received from 996 the backup node. There are a number of different ways that the MP 997 can enforce not forwarding traffic normally received from the backup 998 node. For instance, first, any LSPs set up from the backup node 999 should not be signaled with an IMPLICIT NULL label and second, the 1000 associated label for the ingress- protected LSP could be set to 1001 normally discard inside that context. 1003 When the MP receives a RESV message whose matching PATH state had an 1004 INGRESS-PROTECTION object, the MP SHOULD add the INGRESS-PROTECTION 1005 object to the RESV message before forwarding it. The Backup PATH 1006 handling is as described in [RFC4090] and [RFC4875]. 1008 6.5. Revertive Behavior 1010 Upon a failure event in the (primary) ingress of a protected LSP, the 1011 protected LSP is locally repaired by the backup ingress. There are a 1012 couple of basic strategies for restoring the LSP to a full working 1013 path. 1015 - Revert to Primary Ingress: When the primary ingress is restored, 1016 it re-signals each of the LSPs that start from the primary 1017 ingress. The traffic for every LSP successfully re-signaled is 1018 switched back to the primary ingress from the backup ingress. 1020 - Global Repair by Backup Ingress: After determining that the 1021 primary ingress of an LSP has failed, the backup ingress computes 1022 a new optimal path, signals a new LSP along the new path, and 1023 switches the traffic to the new LSP. 1025 6.5.1. Revert to Primary Ingress 1027 If "Revert to Primary Ingress" is desired for a protected LSP, the 1028 (primary) ingress of the LSP re-signals the LSP that starts from the 1029 primary ingress after the primary ingress restores. When the LSP is 1030 re-signaled successfully, the traffic is switched back to the primary 1031 ingress from the backup ingress and redirected into the LSP starting 1032 from the primary ingress. 1034 It is possible that the Ingress failure was inaccurately detected, 1035 that the Ingress recovers before the Backup Ingress does Global 1036 Repair, or that the Ingress has the ability to take over an LSP based 1037 on receiving the associated RESVs. 1039 If the ingress can resignal the PATH messages for the LSP, then the 1040 ingress can specify the "Revert to Ingress" control-option in the 1041 INGRESS-PROTECTION object. Doing so may cause a duplication of 1042 traffic while the Ingress starts sending traffic again before the 1043 Backup Ingress stops; the alternative is to drop traffic for a short 1044 period of time. 1046 Additionally, the Backup Ingress can set the "Revert To Ingress" 1047 control-option as a request for the Ingress to take over. 1049 6.5.2. Global Repair by Backup Ingress 1051 When the backup ingress has determined that the primary ingress of 1052 the protected LSP has failed (e.g., via the IGP), it can compute a 1053 new path and signal a new LSP along the new path so that it no longer 1054 relies upon local repair. To do this, the backup ingress uses the 1055 same tunnel sender address in the Sender Template Object and uses the 1056 previously allocated second LSP-ID in the INGRESS-PROTECTION object 1057 of the PATH message as the LSP-ID of the new LSP. This allows the 1058 new LSP to share resources with the old LSP. 1060 When the backup ingress has determined that the primary ingress of 1061 the protected LSP has failed (e.g., via the IGP), it can compute a 1062 new path and signal a new LSP along the new path so that it no longer 1063 relies upon local repair. To do this, the backup ingress uses the 1064 same tunnel sender address in the Sender Template Object and uses the 1065 previously allocated second LSP-ID in the INGRESS-PROTECTION object 1066 of the PATH message as the LSP-ID of the new LSP. This allows the 1067 new LSP to share resources with the old LSP. In addition, if the 1068 Ingress recovers, the Backup Ingress SHOULD send it RESVs with the 1069 INGRESS-PROTECTION object where either the "Force to Backup" or 1070 "Revert to Ingress" is specified. The Secondary LSP ID should be the 1071 unused LSP ID - while the LSP ID signaled in the RESV will be that 1072 currently active. The Ingress can learn from the RESVs what to 1073 signal. Even if the Ingress does not take over, the RESVs notify it 1074 that the particular LSP IDs are in use. The Backup Ingress can 1075 reoptimize the new LSP as necessary until the Ingress recovers. 1076 Alternately, the Backup Ingress can create a new LSP with no 1077 bandwidth reservation that duplicates the path(s) of the protected 1078 LSP, move traffic to the new LSP, delete the protected LSP, and then 1079 resignal the new LSP with bandwidth. 1081 7. Security Considerations 1083 In principle this document does not introduce new security issues. 1084 The security considerations pertaining to RFC 4090, RFC 4875 and 1085 other RSVP protocols remain relevant. 1087 8. IANA Considerations 1089 TBD 1091 9. Contributors 1093 Renwei Li 1094 Huawei Technologies 1095 2330 Central Expressway 1096 Santa Clara, CA 95050 1097 USA 1098 Email: renwei.li@huawei.com 1100 Quintin Zhao 1101 Huawei Technologies 1102 Boston, MA 1103 USA 1104 Email: quintin.zhao@huawei.com 1106 Zhenbin Li 1107 Huawei Technologies 1108 2330 Central Expressway 1109 Santa Clara, CA 95050 1110 USA 1111 Email: zhenbin.li@huawei.com 1113 Boris Zhang 1114 Telus Communications 1115 200 Consilium Pl Floor 15 1116 Toronto, ON M1H 3J3 1117 Canada 1118 Email: Boris.Zhang@telus.com 1120 Markus Jork 1121 Juniper Networks 1122 10 Technology Park Drive 1123 Westford, MA 01886 1124 USA 1125 Email: mjork@juniper.net 1127 10. Acknowledgement 1129 The authors would like to thank Rahul Aggarwal, Eric Osborne, Ross 1130 Callon, Loa Andersson, Michael Yue, Olufemi Komolafe, Rob Rennison, 1131 Neil Harrison, Kannan Sampath, and Ronhazli Adam for their valuable 1132 comments and suggestions on this draft. 1134 11. References 1136 11.1. Normative References 1138 [RFC1700] Reynolds, J. and J. Postel, "Assigned Numbers", RFC 1700, 1139 October 1994. 1141 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1142 Requirement Levels", BCP 14, RFC 2119, March 1997. 1144 [RFC3692] Narten, T., "Assigning Experimental and Testing Numbers 1145 Considered Useful", BCP 82, RFC 3692, January 2004. 1147 [RFC2205] Braden, B., Zhang, L., Berson, S., Herzog, S., and S. 1148 Jamin, "Resource ReSerVation Protocol (RSVP) -- Version 1 1149 Functional Specification", RFC 2205, September 1997. 1151 [RFC3031] Rosen, E., Viswanathan, A., and R. Callon, "Multiprotocol 1152 Label Switching Architecture", RFC 3031, January 2001. 1154 [RFC3209] Awduche, D., Berger, L., Gan, D., Li, T., Srinivasan, V., 1155 and G. Swallow, "RSVP-TE: Extensions to RSVP for LSP 1156 Tunnels", RFC 3209, December 2001. 1158 [RFC3473] Berger, L., "Generalized Multi-Protocol Label Switching 1159 (GMPLS) Signaling Resource ReserVation Protocol-Traffic 1160 Engineering (RSVP-TE) Extensions", RFC 3473, January 2003. 1162 [RFC4090] Pan, P., Swallow, G., and A. Atlas, "Fast Reroute 1163 Extensions to RSVP-TE for LSP Tunnels", RFC 4090, 1164 May 2005. 1166 [RFC4461] Yasukawa, S., "Signaling Requirements for Point-to- 1167 Multipoint Traffic-Engineered MPLS Label Switched Paths 1168 (LSPs)", RFC 4461, April 2006. 1170 [RFC4875] Aggarwal, R., Papadimitriou, D., and S. Yasukawa, 1171 "Extensions to Resource Reservation Protocol - Traffic 1172 Engineering (RSVP-TE) for Point-to-Multipoint TE Label 1173 Switched Paths (LSPs)", RFC 4875, May 2007. 1175 [P2MP-FRR] 1176 Le Roux, J., Aggarwal, R., Vasseur, J., and M. Vigoureux, 1177 "P2MP MPLS-TE Fast Reroute with P2MP Bypass Tunnels", 1178 draft-leroux-mpls-p2mp-te-bypass , March 1997. 1180 11.2. Informative References 1182 [RFC2702] Awduche, D., Malcolm, J., Agogbua, J., O'Dell, M., and J. 1183 McManus, "Requirements for Traffic Engineering Over MPLS", 1184 RFC 2702, September 1999. 1186 [RFC3032] Rosen, E., Tappan, D., Fedorkow, G., Rekhter, Y., 1187 Farinacci, D., Li, T., and A. Conta, "MPLS Label Stack 1188 Encoding", RFC 3032, January 2001. 1190 Appendix A. Authors' Addresses 1192 Huaimo Chen 1193 Huawei Technologies 1194 Boston, MA 1195 USA 1196 Email: huaimo.chen@huawei.com 1198 Ning So 1199 Tata Communications 1200 2613 Fairbourne Cir. 1201 Plano, TX 75082 1202 USA 1203 Email: ning.so@tatacommunications.com 1205 Autumn Liu 1206 Ericsson 1207 300 Holger Way 1208 San Jose, CA 95134 1209 USA 1210 Email: autumn.liu@ericsson.com 1211 Raveendra Torvi 1212 Juniper Networks 1213 10 Technology Park Drive 1214 Westford, MA 01886 1215 USA 1216 Email: rtorvi@juniper.net 1218 Alia Atlas 1219 Juniper Networks 1220 10 Technology Park Drive 1221 Westford, MA 01886 1222 USA 1223 Email: akatlas@juniper.net 1225 Yimin Shen 1226 Juniper Networks 1227 10 Technology Park Drive 1228 Westford, MA 01886 1229 USA 1230 Email: yshen@juniper.net 1232 Fengman Xu 1233 Verizon 1234 2400 N. Glenville Dr 1235 Richardson, TX 75082 1236 USA 1237 Email: fengman.xu@verizon.com 1239 Mehmet Toy 1240 Comcast 1241 1800 Bishops Gate Blvd. 1242 Mount Laurel, NJ 08054 1243 USA 1244 Email: mehmet_toy@cable.comcast.com 1246 Lei Liu 1247 UC Davis 1248 USA 1249 Email: liulei.kddi@gmail.com