idnits 2.17.1 draft-ietf-tsvwg-ecn-mpls-00.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 17. -- Found old boilerplate from RFC 3978, Section 5.5, updated by RFC 4748 on line 966. -- Found old boilerplate from RFC 3979, Section 5, paragraph 1 on line 977. -- Found old boilerplate from RFC 3979, Section 5, paragraph 2 on line 984. -- Found old boilerplate from RFC 3979, Section 5, paragraph 3 on line 990. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust Copyright Line does not match the current year -- The document seems to lack a disclaimer for pre-RFC5378 work, but may have content which was first submitted before 10 November 2008. If you have contacted all the original authors and they are all willing to grant the BCP78 rights to the IETF Trust, then this is fine, and you can ignore this comment. If not, you may need to add the pre-RFC5378 disclaimer. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (February 20, 2007) is 6275 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) == Unused Reference: 'RFC2475' is defined on line 840, but no explicit reference was found in the text == Unused Reference: 'I-D.briscoe-tsvwg-re-ecn-border-cheat' is defined on line 881, but no explicit reference was found in the text ** Downref: Normative reference to an Informational RFC: RFC 2475 ** Downref: Normative reference to an Informational RFC: RFC 3260 == Outdated reference: A later version (-09) exists of draft-briscoe-tsvwg-re-ecn-tcp-03 == Outdated reference: A later version (-20) exists of draft-ietf-nsis-rmd-08 Summary: 3 errors (**), 0 flaws (~~), 5 warnings (==), 7 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group B. Davie 3 Internet-Draft Cisco Systems, Inc. 4 Intended status: Standards Track B. Briscoe 5 Expires: August 24, 2007 J. Tay 6 BT Research 7 February 20, 2007 9 Explicit Congestion Marking in MPLS 10 draft-ietf-tsvwg-ecn-mpls-00.txt 12 Status of this Memo 14 By submitting this Internet-Draft, each author represents that any 15 applicable patent or other IPR claims of which he or she is aware 16 have been or will be disclosed, and any of which he or she becomes 17 aware will be disclosed, in accordance with Section 6 of BCP 79. 19 Internet-Drafts are working documents of the Internet Engineering 20 Task Force (IETF), its areas, and its working groups. Note that 21 other groups may also distribute working documents as Internet- 22 Drafts. 24 Internet-Drafts are draft documents valid for a maximum of six months 25 and may be updated, replaced, or obsoleted by other documents at any 26 time. It is inappropriate to use Internet-Drafts as reference 27 material or to cite them other than as "work in progress." 29 The list of current Internet-Drafts can be accessed at 30 http://www.ietf.org/ietf/1id-abstracts.txt. 32 The list of Internet-Draft Shadow Directories can be accessed at 33 http://www.ietf.org/shadow.html. 35 This Internet-Draft will expire on August 24, 2007. 37 Copyright Notice 39 Copyright (C) The IETF Trust (2007). 41 Abstract 43 RFC 3270 defines how to support the Diffserv architecture in MPLS 44 networks, including how to encode Diffserv Code Points (DSCPs) in an 45 MPLS header. DSCPs may be encoded in the EXP field, while other uses 46 of that field are not precluded. RFC3270 makes no statement about 47 how Explicit Congestion Notification (ECN) marking might be encoded 48 in the MPLS header. This draft defines how an operator might define 49 some of the EXP codepoints for explicit congestion notification, 50 without precluding other uses. 52 Requirements Language 54 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 55 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 56 document are to be interpreted as described in RFC 2119 [RFC2119]. 58 Table of Contents 60 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4 61 1.1. Change History . . . . . . . . . . . . . . . . . . . . . . 4 62 1.2. Background . . . . . . . . . . . . . . . . . . . . . . . . 4 63 1.3. Intent . . . . . . . . . . . . . . . . . . . . . . . . . . 5 64 1.4. Terminology . . . . . . . . . . . . . . . . . . . . . . . 5 65 2. Use of MPLS EXP Field for ECN . . . . . . . . . . . . . . . . 6 66 3. Per-domain ECT checking . . . . . . . . . . . . . . . . . . . 8 67 4. ECN-enabled MPLS domain . . . . . . . . . . . . . . . . . . . 9 68 4.1. Pushing (adding) one or more labels to an IP packet . . . 9 69 4.2. Pushing one or more labels onto an MPLS labelled packet . 9 70 4.3. Congestion experienced in an interior MPLS node . . . . . 9 71 4.4. Crossing a Diffserv Domain Boundary . . . . . . . . . . . 10 72 4.5. Popping an MPLS label (not the end of the stack) . . . . . 10 73 4.6. Popping the last MPLS label in the stack . . . . . . . . . 10 74 4.7. Diffserv Tunneling Models . . . . . . . . . . . . . . . . 11 75 4.8. Extension to Pre-Congestion Notification . . . . . . . . . 11 76 4.8.1. Label Push onto IP packet . . . . . . . . . . . . . . 12 77 4.8.2. Pushing Additional MPLS Labels . . . . . . . . . . . . 12 78 4.8.3. Admission Control or Pre-emption Marking inside 79 MPLS domain . . . . . . . . . . . . . . . . . . . . . 12 80 4.8.4. Popping an MPLS Label (not end of stack) . . . . . . . 12 81 4.8.5. Popping the last MPLS Label to expose IP header . . . 12 82 5. ECN-disabled MPLS domain . . . . . . . . . . . . . . . . . . . 13 83 6. The use of more codepoints with E-LSPs and L-LSPs . . . . . . 13 84 7. Relationship to tunnel behavior in RFC 3168 . . . . . . . . . 14 85 7.1. Alternative approach to support ECN in an MPLS domain . . 14 86 8. Example Uses . . . . . . . . . . . . . . . . . . . . . . . . . 15 87 8.1. RFC3168-style ECN . . . . . . . . . . . . . . . . . . . . 15 88 8.2. ECN Co-existence with Diffserv E-LSPs . . . . . . . . . . 15 89 8.3. Congestion-feedback-based Traffic Engineering . . . . . . 16 90 8.4. PCN flow admission control and flow pre-emption . . . . . 16 91 9. Deployment Considerations . . . . . . . . . . . . . . . . . . 17 92 9.1. Marking non-ECN Capable Packets . . . . . . . . . . . . . 17 93 9.2. Non-ECN capable routers in an MPLS Domain . . . . . . . . 18 94 10. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 18 95 11. Security Considerations . . . . . . . . . . . . . . . . . . . 18 96 12. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 19 97 13. References . . . . . . . . . . . . . . . . . . . . . . . . . . 19 98 13.1. Normative References . . . . . . . . . . . . . . . . . . . 19 99 13.2. Informative References . . . . . . . . . . . . . . . . . . 20 100 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 21 101 Intellectual Property and Copyright Statements . . . . . . . . . . 22 103 1. Introduction 105 1.1. Change History 107 [Note to RFC Editor: This section to be removed before publication] 109 This version (draft-ietf-tsvwg-ecn-mpls-00.txt) differs from the last 110 (draft-davie-mpls-ecn-01.txt) only in title, date, and updated 111 references. 113 Changes from draft-davie-ecn-mpls-00 to draft-davie-ecn-mpls-01: 115 o Corrected the description of ECN-MPLS marking proposed in 116 [Shayman], which closely corresponds to that proposed in this 117 document. 119 o Pre-congestion notification (PCN) marking is now described in a 120 way that does not require normative references to PCN 121 specifications. PCN discussion now serves only to illustrate how 122 the ECN marking concepts can be extended to cover more complex 123 scenarios, with PCN being an example. 125 o Added specification of behavior when MPLS encapsulated packets 126 cross from an ECN-enabled domain to a domain that is not ECN- 127 enabled. 129 o Clarified that copying MPLS ECN or PCN marking into exposed IP 130 header on egress is not mandatory 132 o Fixed typos and nits 134 1.2. Background 136 [RFC3270] defines how to support the Diffserv architecture in MPLS 137 networks, including how to encode Diffserv Code Points (DSCPs) in an 138 MPLS header. DSCPs may be encoded in the EXP field, while other uses 139 of that field are not precluded. RFC3270 makes no statement about 140 how Explicit Congestion Notification (ECN) marking might be encoded 141 in the MPLS header. This draft defines how an operator might define 142 some of the EXP codepoints for explicit congestion notification, 143 without precluding other uses. In parallel to the activity defining 144 the addition of ECN to IP [RFC3168], two proposals were made to add 145 ECN to MPLS [Floyd][Shayman]. These proposals, however, fell by the 146 wayside. With ECN for IP now being a proposed standard, and 147 developing interest in using pre-congestion notification (PCN) for 148 admission control and flow pre-emption 149 [I-D.briscoe-tsvwg-cl-architecture], there is consequent interest in 150 being able to support ECN across IP networks consisting of MPLS- 151 enabled domains. Therefore it is necessary to specify the protocol 152 for including ECN in the MPLS shim header, and the protocol behavior 153 of edge MPLS nodes. 155 We note that in [RFC3168] there are four codepoints used for ECN 156 marking, which are encoded using two bits of the IP header. The MPLS 157 EXP field is the logical place to encode ECN codepoints, but with 158 only 3 bits (8 codepoints) available, and with the same field being 159 used to convey DSCP information as well, there is a clear incentive 160 to conserve the number of codepoints consumed for ECN purposes. 161 Efficient use of the EXP field has been a focus of prior drafts 162 [Floyd] [Shayman] and we draw on those efforts in this draft as well. 164 1.3. Intent 166 Our intent is to specify how the MPLS shim header[RFC3032] should 167 denote ECN marking and how MPLS nodes should understand whether the 168 transport for a packet will be ECN capable. We offer this as a 169 building block, from which to build different congestion notification 170 systems. We do not intend to specify how the resulting congestion 171 notification is fed back to an upstream node that can mitigate 172 congestion. For instance, unlike [Shayman], we do not specify edge- 173 to-edge MPLS domain feedback, but we also do not preclude it. 174 Nonetheless, we do specify how the egress node of an MPLS domain 175 should copy congestion notification from the MPLS shim into the 176 underlying IP header if the ECN is to be carried onward towards the 177 IP receiver. But we do NOT mandate that MPLS congestion notification 178 must be copied into the IP header for onward transmission. This 179 draft aims to be generic for any use of congestion notification in 180 MPLS. PCN or traffic engineering are merely two of many motivating 181 applications (see Section 8.) 183 1.4. Terminology 185 This document draws freely on the terminology of ECN [RFC3168] and 186 MPLS [RFC3031]. For ease of reference, we have included some 187 definitions here, but refer the reader to the references above for 188 complete specifications of the relevant technologies: 190 o CE: Congestion Experienced. One of the states with which a packet 191 may be marked in a network supporting ECN. A packet is marked in 192 this state by an ECN-capable router, to indicate that this router 193 was experiencing congestion at the time the packet arrived. 195 o ECT: ECN-capable Transport. One of the ECN states which a packet 196 may be in when it is sent by an end system. An end system marks a 197 packet with an ECT codepoint to indicate that the end-points of 198 the transport protocol are ECN-capable. A router may not mark a 199 packet as CE unless the packet was marked ECT when it arrived. 201 o Not-ECT: Not ECN capable transport. An end system marks a packet 202 with this codepoint to indicate that the end-points of the 203 transport protocol are not ECN-capable. A congested router cannot 204 mark such packets as CE, and thus can only drop them to indicate 205 congestion. 207 o EXP field. A 3 bit field in the MPLS label header [RFC3032] which 208 may be used to convey Diffserv information (and used in this draft 209 to carry ECN information). 211 o PHP. Penultimate Hop Popping. An MPLS operation in which the 212 penultimate Label Switching Router (LSR) on a Label Switched Path 213 (LSP) removes the top label from the packet before forwarding the 214 packet to the final LSR on the LSP. 216 2. Use of MPLS EXP Field for ECN 218 We propose that LSRs configured for explicit congestion notification 219 should use the EXP field in the MPLS shim header. However, RFC 3270 220 already defines use of codepoints in the EXP field for differentiated 221 services. Although it does not preclude other compatible uses of the 222 EXP field, this clearly seems to limit the space available for ECN, 223 given the field is only 3 bits (8 codepoints). 225 RFC 3270 defines two possible approaches for requesting 226 differentiated service treatment from an LSR. 228 o In the E-LSP approach, different codepoints of the EXP field in 229 the MPLS shim header are used to indicate the packet's per hop 230 behavior (PHB). 232 o In the L-LSP approach, an MPLS label is assigned for each PHB 233 scheduling class (PSC, as defined in [RFC3260], so that an LSR 234 determines both its forwarding and its scheduling behavior from 235 the label. 237 If an MPLS domain uses the L-LSP approach, there is likely to be 238 space in the EXP field for ECN codepoint(s). Where the E-LSP 239 approach is used, then codepoint space in the EXP field is likely to 240 be scarce. This draft focuses on interworking ECN marking with the 241 E-LSP approach as it is the tougher problem. Consequently the same 242 approach can also be applied with L-LSPs. 244 We recommend that explicit congestion notification in MPLS should use 245 codepoints instead of bits in the EXP field. Since not every PHB 246 will need an associated ECN codepoint and in some applications a 247 given PHB might need two ECN codepoints (see, for 248 example,[I-D.briscoe-tsvwg-cl-architecture]) it would be wasteful to 249 assign a dedicated bit for ECN. 251 For each PHB that uses ECN marking, we assume one EXP codepoint will 252 be defined meaning not congestion marked (Not-CM), and at least one 253 other codepoint will be defined meaning congestion marked (CM). 254 Therefore, each PHB that uses ECN marking will consume at least two 255 EXP codepoints. But PHBs that do not use ECN marking will only 256 consume one. 258 Further, we wish to use minimal space in the MPLS shim header to tell 259 interior LSRs whether each packet will be received by an ECN-capable 260 transport (ECT). Nonetheless, we must ensure that an end-point that 261 would not understand an ECN mark will not receive one, otherwise it 262 will not be able to respond to congestion as it should. In the past, 263 three solutions to this problem have been proposed: 265 o One possible approach is for congested LSRs to mark the ECN field 266 in the underlying IP header at the bottom of the label stack. 267 Although many commercial LSRs routinely access the IP header for 268 other reasons (ECMP), there are numerous drawbacks to attempting 269 to find an IP header beneath an MPLS label stack. Notably, there 270 is the challenge of detecting the absence of an IP header when 271 non-IP packets are carried on an LSP. Therefore we will not 272 consider this approach further. 274 o In the scheme suggested by [Floyd] ECT and CE are overloaded into 275 one bit, so that a 0 means ECT while a 1 might either mean Not-ECT 276 or it might mean CE. A packet that has been marked as having 277 experienced congestion upstream, and then is picked out for 278 marking at a second congested LSR, will be dropped by the second 279 LSR since it cannot determine whether the packet has previously 280 experienced congestion or if ECN is not supported by the 281 transport. 283 While such an approach seemed potentially palatable, we do not 284 recommend it here for the following reasons. In some cases we 285 wish to be able to use ECN marking long before actual congestion 286 (e.g. pre-congestion notification). In these circumstances, 287 marking rates at each LSR might be non-negligible most of the 288 time, so the chances of a previously marked packet encountering an 289 LSR that wants to mark it again will also be non-negligible. In 290 the case where CE and not-ECT are indistinguishable to core 291 routers, such a scenario could lead to unacceptable drop rates. 292 If the typical marking rate at every router or LSR is p, and the 293 typical diameter of the network of LSRs is d, then the probability 294 that a marked packet will be chosen for marking more than once is 295 1-[Pr(never marked) + Pr(marked at exactly one hop)] = 1- [(1-p)^d 296 + dp(1-p)^(d-1)]. For instance, with 6 LSRs in a row, each 297 marking ECN with 1% probability, the chances of a packet that is 298 already marked being chosen for marking a second time is 0.15%. 299 The bit overloading scheme would therefore introduce a drop rate 300 of 0.15% unnecessarily. Given that most modern core networks are 301 sized to introduce near-zero packet drop, it may be unacceptable 302 to drop over one in a thousand packets unnecessarily. 304 o A third possible approach was suggested by [Shayman]. In this 305 scheme, interior LSRs assume that the endpoints are ECN-capable, 306 but this assumption is checked when the final label is popped. If 307 an interior LSR has marked ECN in the EXP field of the shim 308 header, but the IP header says the endpoints are not ECN capable, 309 the edge router (or penultimate router, if using penultimate hop 310 popping) drops the packet. We recommend this scheme, which we 311 call `per-domain ECT checking', and define it more precisely in 312 the following section. Its chief drawback is that it can cause 313 packets to be forwarded after encountering congestion only to be 314 dropped at the egress of the MPLS domain. The rationale for this 315 decision is given in Section 9.1. 317 3. Per-domain ECT checking 319 For the purposes of this discussion, we define the egress nodes of an 320 MPLS domain as the nodes that pop the last MPLS label from the label 321 stack, exposing the IP (or, potentially non-IP) header. Note that 322 such a node may be the ultimate or penultimate hop of an LSP, 323 depending on whether penultimate hop popping (PHP) is employed. 325 In the per-domain ECT checking approach, the egress nodes take 326 responsibility for checking whether the transport is ECN capable. 327 This draft does not specify how these nodes should pass on congestion 328 notification, because different approaches are likely in different 329 scenarios. However, if congestion notification in the MPLS header is 330 copied into the IP header, the procedure MUST conform to the 331 specification given here. 333 If congestion notification is passed to the transport without first 334 passing it onward in the IP header, the approach used must take 335 similar care to check that the transport is ECN capable before 336 passing it ECN markings. Specifically, if the transport for a 337 particular congestion marked MPLS packet is found not to be ECN- 338 capable, the packet MUST be dropped at this egress node. 340 In the per-domain ECT checking approach, only the egress nodes check 341 whether an IP packet is destined for an ECN-capable transport. 342 Therefore, any single LSR within an MPLS domain MUST NOT be 343 configured to enable ECN marking unless all the egress LSRs 344 surrounding it are already configured to handle ECN marking. 346 We call a domain surrounded by ECN-capable egress LSRs an ECN-enabled 347 MPLS domain. This term only implies that all the egress LSRs are 348 ECN-enabled; some interior LSRs may not be ECN-enabled. For 349 instance, it would be possible to use legacy LSRs incapable of 350 supporting ECN in the interior of an MPLS domain as long as all the 351 egress LSRs were ECN-capable. Note that if PHP is used, the 352 "penultimate hop" routers which perform the pop operation do need to 353 be ECN-enabled, since they are acting in this context as egress LSRs. 355 4. ECN-enabled MPLS domain 357 In the following subsections we describe various operations affecting 358 the ECN marking of a packet that may be performed at MPLS edge and 359 core LSRs. 361 4.1. Pushing (adding) one or more labels to an IP packet 363 On encapsulating an IP packet with an MPLS label stack, the ECN field 364 must be translated from the IP packet into the MPLS EXP field. The 365 Not-CM (not congestion marked) state is set in the MPLS EXP field if 366 the ECN status of the IP packet is "Not ECT" or ECT(1) or ECT(0). 367 The CM state is set if the ECN status of the IP packet is "CE". If 368 more than one label is pushed at one time, the same value should be 369 placed in the EXP value of all label stack entries. 371 4.2. Pushing one or more labels onto an MPLS labelled packet 373 The EXP field is copied directly from the topmost label before the 374 push to the newly added outer label. If more than one label is being 375 pushed, the same EXP value is copied to all label stack entries. 377 4.3. Congestion experienced in an interior MPLS node 379 If the EXP codepoint of the packet maps to a PHB that uses ECN 380 marking and the marking algorithm requires the packet to be marked, 381 the CM state is set (irrespective of whether it is already in the CM 382 state). 384 If the buffer is full, a packet is dropped. 386 4.4. Crossing a Diffserv Domain Boundary 388 If an MPLS-encapsulated packet crosses a Diffserv domain boundary, it 389 may be the case that the two domains use different encodings of the 390 same PHB in the EXP field. In such cases, the EXP field must be 391 rewritten at the domain boundary. If the PHB is one that supports 392 ECN, then the appropriate ECN marking should also be preserved when 393 the EXP field is mapped at the boundary. 395 If an MPLS-encapsulated packet that is in the CM state crosses from a 396 domain that is ECN-enabled (as defined in Section 3) to a domain that 397 is not ECN-enabled, then it is necessary to perform the egress 398 checking procedures at the egress LSR of the ECN-enabled domain. 399 This means that if the encapsulated packet is not ECN capable, the 400 packet MUST be dropped. Note that this implies the egress LSR must 401 be able to look beneath the MPLS header without popping the label 402 stack. 404 The related issue of Diffserv tunnel models is discussed in 405 Section 4.7. 407 4.5. Popping an MPLS label (not the end of the stack) 409 When a packet has more than one MPLS label in the stack and the top 410 label is popped, another MPLS label is exposed. In this case the ECN 411 information should be transferred from the outer EXP field to the 412 inner MPLS label in the following manner. If the inner EXP field is 413 Not-CM, the inner EXP field is set to the same CM or Not-CM state as 414 the outer EXP field. If the inner EXP field is CM, it remains 415 unchanged whatever the outer EXP field. Note that an inner value of 416 CM and an outer value of not-CM should be considered anomalous, and 417 SHOULD be logged in some way by the LSR. 419 4.6. Popping the last MPLS label in the stack 421 When the last MPLS label is popped from the packet, its payload is 422 exposed. If that packet is not IP, and does not have any capability 423 equivalent to ECT, it is assumed Not-ECT and treated as such. That 424 means that if the EXP value of the MPLS header was CM, the packet 425 MUST be dropped. 427 Assuming an IP packet was exposed, we have to examine whether that 428 packet is ECT or not. A Not-ECT packet MUST be dropped if the EXP 429 field is CM. 431 For the remainder of this section, we describe the behavior that is 432 required if the ECN information is to be transferred from the MPLS 433 header into the exposed IP header for onward transmission. As noted 434 in Section 1.3, such behavior is not mandated by this document, but 435 may be selected by an operator. 437 If the inner IP packet is Not-ECT, its ECN field remains unchanged if 438 the EXP field is Not-CM. If the ECN field of the inner packet is set 439 to ECT(0), ECT(1) or CE, the ECN field remains unchanged if the EXP 440 field is set to Not-CM. The ECN field is set to CE if the EXP field 441 is CM. Note that an inner value of CE and an outer value of not-CM 442 should be considered anomalous, and SHOULD be logged in some way by 443 the LSR. 445 4.7. Diffserv Tunneling Models 447 [RFC3270] describes three tunneling models for Diffserv support 448 across MPLS Domains, referred to as the uniform, short pipe, and pipe 449 models. The differences between these models lie in whether the 450 Diffserv treatment that applies to a packet while it travels along a 451 particular LSP is carried to the last hop of the LSP and beyond the 452 last hop. Depending on which mode is preferred by an operator, the 453 EXP value or DSCP value of an exposed header following a label pop 454 may or may not be dependent on the EXP value of the label that is 455 removed by the pop operation. We believe that in the case of ECN 456 marking, the use of these models should only apply to the encoding of 457 the Diffserv PHB in the EXP value, and that the choice of codepoint 458 for ECN should always be made based on the procedures described 459 above, independent of the tunneling model. 461 4.8. Extension to Pre-Congestion Notification 463 This section describes how the preceding mechanisms can be extended 464 to support PCN [I-D.briscoe-tsvwg-cl-architecture]. Our intent here 465 is to show that the mechanisms are readily extended to more complex 466 scenarios than ECN, but this section may be safely ignored if one is 467 interested only in supporting ECN. 469 The relevant aspects of PCN for the purposes of this discussion are: 471 o PCN uses 3 states rather than 2 for ECN - these are referred to as 472 admission marked (AM), pre-emption marked (PM) and not marked (NM) 473 states. (See Section 8.4 for further discussion of PCN and the 474 possibility of using fewer codepoints.) 476 o A packet can go from NM to AM, from NM to PM, or from AM to PM, 477 but no other transition is possible. 479 o Whereas ECN-capable packets are identified by the ECT value in the 480 IP header, PCN-capability is determined by the PHB of the packet. 482 Thus, to support PCN fully in an MPLS domain for a particular PHB, a 483 total of 3 codepoints need to be allocated for that PHB. These 3 484 codepoints represent the admission marked (AM), pre-emption marked 485 (PM) and not marked (NM) states. The procedures described above need 486 to be slightly modified to support this scenario. The following 487 procedures are invoked when the topmost DSCP or EXP value indicates a 488 PHB that supports PCN. 490 4.8.1. Label Push onto IP packet 492 If the IP packet header indicates AM, set the EXP value of all 493 entries in the label stack to AM. If the IP packet header indicates 494 PM, set the EXP value of all entries in the label stack to PM. For 495 any other marking of the IP header, set the EXP value of all entries 496 in the label stack to NM. 498 4.8.2. Pushing Additional MPLS Labels 500 The procedures of Section 4.2 apply. 502 4.8.3. Admission Control or Pre-emption Marking inside MPLS domain 504 The EXP value can be set to AM or PM according to the same procedures 505 as described in [I-D.briscoe-tsvwg-cl-phb]. For the purposes of this 506 document, it does not matter exactly what algorithms are used to 507 decide when to set AM or PM; all that matters is that if a router 508 would have marked AM (or PM) in the IP header, it should set the EXP 509 value in the MPLS header to the AM (or PM) codepoint. 511 4.8.4. Popping an MPLS Label (not end of stack) 513 When popping an MPLS Label exposes another MPLS label, the AM or PM 514 marking should be transferred to the exposed EXP field in the 515 following manner: if the inner EXP value is NM, then it should be set 516 to the same marking state as the EXP value of the popped label stack 517 entry. If the inner EXP value is AM, it should be unchanged if the 518 popped EXP value was AM, and it should be set to PM if the popped EXP 519 value was PM. If the popped EXP value was NM, this should be logged 520 in some way and the inner EXP value should be unchanged. If the 521 inner EXP value is PM, it should be unchanged whatever the popped EXP 522 value was, but any EXP value other than PM should be logged. 524 4.8.5. Popping the last MPLS Label to expose IP header 526 When popping the last MPLS Label exposes the IP header, there are two 527 cases to consider: 529 o the popping LSR is NOT the egress router of the PCN region, in 530 which case AM or PM marking should be transferred to the exposed 531 IP header field; or 533 o the popping LSR IS the egress router of the PCN region. 535 In the latter case, the behavior of the egress LSR is defined in 536 [I-D.briscoe-tsvwg-cl-architecture] and is beyond the scope of this 537 document. In the former case, the marking should be transferred from 538 the popped MPLS header to the exposed IP header as follows: if the 539 inner IP header value is neither AM nor PM, and the EXP value was NM, 540 then the IP header should be unchanged. For any other EXP value, the 541 IP header should be set to the same marking state as the EXP value of 542 the popped label stack entry. If the inner IP header value is AM, it 543 should be unchanged if the popped EXP value was AM, and it should be 544 set to PM if the popped EXP value was PM. If the popped EXP value 545 was NM, this should be logged in some way and the inner IP header 546 value should be unchanged. If the IP header value is PM, it should 547 be unchanged whatever the popped EXP value was, but any EXP value 548 other than PM should be logged. 550 5. ECN-disabled MPLS domain 552 If ECN is not enabled on all the egress LSRs of a domain, ECN MUST 553 NOT be enabled on any LSRs throughout the domain. If congestion is 554 experienced on any LSR in an ECN-disabled MPLS domain, packets MUST 555 be dropped, NOT marked. The exact algorithm for deciding when to 556 drop packets during congestion (e.g. tail-drop, RED, etc.) is a local 557 matter for the operator of the domain. 559 6. The use of more codepoints with E-LSPs and L-LSPs 561 RFC 3270 gives different options with E-LSPs and L-LSPs and some of 562 those could potentially provide ample EXP codepoints for ECN/PCN. 563 However, deploying L-LSPs vs E-LSPs has many implications such as 564 platform support and operational complexity. The above ECN/PCN MPLS 565 solution should provide some flexibility. If the operator has 566 deployed one L-LSP per PHB scheduling class, then EXP space will be a 567 non-issue and it could be used to achieve more sophisticated ECN/PCN 568 behavior if required. If the operator wants to stick to E-LSPs and 569 uses a handful of EXP codepoints for Diffserv, it may be desirable to 570 operate with a minimum number of extra ECN/PCN codepoints, even if 571 this comes with some compromise on ECN/PCN optimality. See Section 8 572 for discussion of some possible deployment scenarios. 574 7. Relationship to tunnel behavior in RFC 3168 576 [RFC3168] defines two modes of encapsulating ECN-marked IP packets 577 inside additional IP headers when tunnels are used. The two modes 578 are the "full functionality" and "limited functionality" modes. In 579 the full functionality mode, the ECT information from the inner 580 header is copied to the outer header at the tunnel ingress, but the 581 CE information is not. In the limited functionality mode, neither 582 ECT nor CE information is copied to the outer header, and thus ECN 583 cannot be applied to the encapsulated packet. 585 The behavior that is specified in Section 4 of this document 586 resembles the "full functionality" mode in the sense that it conveys 587 some information from inner to outer header, and in the sense that it 588 enables full ECN support along the MPLS LSP (which is analogous to an 589 IP tunnel in this context). However it differs in one respect, which 590 is that the CE information is conveyed from the inner header to the 591 outer header. Our reason for this different design choice is to give 592 interior routers and LSRs more information about upstream marking in 593 multi-bottleneck cases. For instance, the flow pre-emption marking 594 mechanism proposed for PCN works by only considering packets for 595 marking that have not already been marked upstream. Unless existing 596 pre-emption marking is copied from the inner to the outer header at 597 tunnel ingress, the mechanism doesn't pre-empt enough traffic in 598 cases where anomalous events hit multiple MPLS domains at once. 599 [RFC3168] does not give any reasons against conveying CE information 600 from the inner header to the outer in the "full functionality" mode. 601 So, rather than define different encapsulation methods for ECN and 602 PCN, Section 4 defines a common approach for both. 604 7.1. Alternative approach to support ECN in an MPLS domain 606 It is possible to define an approach for MPLS support of ECN that 607 more closely resembles that of the full functionality mode of 608 [RFC3168]. This approach would differ from that described in 609 Section 4 in the following ways: 611 o when pushing one or more MPLS labels onto an IP packet, the not-CM 612 state is set in the EXP field of all label stack entries 614 o when pushing one or more MPLS labels onto an MPLS packet, the 615 not-CM state is set in the EXP field of all newly added label 616 stack entries 618 o when popping an MPLS label and the exposed header is MPLS (i.e. 619 this is not the end of stack), the EXP field of the MPLS packet 620 should be set to CM if the popped label's EXP value was CM and 621 left unchanged otherwise 623 o when popping an MPLS label and the exposed header is IP, the IP 624 ECN field should be set to CE if the EXP value was CM and if the 625 IP header indicated that the packet was ECN capable. If the IP 626 header indicated not-ECT and the EXP value was CM, the packet MUST 627 be dropped. If the EXP value was not-CM, the ECN field in the IP 628 header is unchanged. 630 The advantages of this scheme over that described in Section 4 are 631 greater similarity to [RFC3168], and the ability to determine, at the 632 end of an LSP, that congestion either did or did not occur along that 633 LSP (since the initial state is always not-CM at the start of an 634 LSP). 636 A disadvantage of this approach is that exceptions to this rule are 637 necessary in cases where the marking process on LSRs needs to depend 638 on whether a packet has already suffered upstream marking. The 639 currently proposed pre-emption marking in PCN is an example where 640 such an exception would be necessary (see the discussion at the start 641 of Section 7). 643 8. Example Uses 645 8.1. RFC3168-style ECN 647 [RFC3168] proposes the use of ECN in TCP and introduces the use of 648 ECN-Echo and CWR flags in the TCP header for initialization. The TCP 649 sender responds accordingly (such as not increasing the congestion 650 window) when it receives an ECN-Echo (ECE) ACK packet (that is, an 651 ACK packet with ECN-Echo flag set in the TCP header), then the sender 652 knows that congestion was encountered in the network on the path from 653 the sender to the receiver. 655 It would be possible to enable ECN in an MPLS domain for Diffserv 656 PHBs like AF and best efforts that are expected to be used by TCP and 657 similar transports (e.g. DCCP [RFC4340]). Then end-to-end 658 congestion control in transports capable of understanding ECN would 659 be able to respond to approaching congestion on LSRs without having 660 to rely on packet discard to signal congestion. 662 8.2. ECN Co-existence with Diffserv E-LSPs 664 Many operators today have deployed Diffserv using the E-LSP approach 665 of [RFC3270]. In many cases the number of PHBs used is less than 8, 666 and hence there remain available codepoints in the EXP space. If an 667 operator wished to support ECN for single PHB, this can be 668 accomplished by simply allocated a second codepoint to the PHB for 669 the "CM" state of that PHB and retaining the old codepoint for the 670 "not-CM" state. An operator with only four deployed PHBs could of 671 course enable ECN marking on all those PHBs. It is easy to imagine 672 cases where some PHBs might benefit more from ECN than others - for 673 example, an operator might use ECN on a premium data service but not 674 on a PHB used for best effort internet traffic. 676 As an illustrative example of how the EXP field might be used in this 677 case, consider the example of an operator who is using the aggregated 678 service classes described in [I-D.chan-tsvwg-diffserv-class-aggr]. 679 He may choose to support ECN only for the Assured Elastic Treatment 680 Aggregate, using the EXP codepoint 010 for the not-CM state and 011 681 for the CM state. All other codepoints could be the same as in 682 [I-D.chan-tsvwg-diffserv-class-aggr]. Of course any other 683 combination of EXP values can be used according to the specific set 684 of PHBs and marking conventions used within that operator's network. 686 8.3. Congestion-feedback-based Traffic Engineering 688 Shayman's traffic engineering [Shayman] proposed the use of ECN by an 689 egress LSR feeding back congestion to an ingress LSR to mitigate 690 congestion by employing dynamic traffic engineering techniques such 691 as shifting flows to an alternate path. It proposed a new RSVP 692 TUNNEL CONGESTION message which was sent to the ingress LSR and 693 ignored by transit LSRs. 695 8.4. PCN flow admission control and flow pre-emption 697 [I-D.briscoe-tsvwg-cl-architecture] proposes using pre-congestion 698 notification (PCN) on routers within an edge-to-edge Diffserv region 699 to control admission of new flows to the region and, if necessary, to 700 pre-empt existing flows in response to disasters and other anomalous 701 routing events. In this approach, the current level of PCN marking 702 is picked up by the signalling used to initiate each flow in order to 703 inform the admission control decision for the whole region at once. 704 As an example, a minor extension to RSVP signalling has been proposed 705 [I-D.lefaucheur-rsvp-ecn] to carry this message, but a similar 706 approach has also been proposed that uses NSIS signalling 707 [I-D.ietf-nsis-rmd]. 709 If it is possible for LSRs to signify congestion in MPLS, PCN marking 710 could be used for admission control and flow pre-emption across a 711 Diffserv region, irrespective of whether it contained pure IP 712 routers, MPLS LSRs, or both. Indeed, the solution could be somewhat 713 more efficient to implement if aggregates could identify themselves 714 by their MPLS label. Section 4.8 describes the mechanisms by which 715 the necessary markings for PCN could be carried in the MPLS header. 717 As an illustrative example of how the EXP field might be used in this 718 case, consider the example of an operator who is using the aggregated 719 service classes described in [I-D.chan-tsvwg-diffserv-class-aggr]. 720 He may choose to support PCN only for the Real Time Treatment 721 Aggregate, using the EXP codepoint 100 for the not-marked (NM) state, 722 101 for the Admission Marked (AM) state, and 111 for the Pre-emption 723 Marked (PM) state. All other codepoints could be the same as in 724 [I-D.chan-tsvwg-diffserv-class-aggr]. Of course any other 725 combination of EXP values can be used according to the specific set 726 of PHBs and marking conventions used within that operator's network. 728 It might also be possible to deploy a similar solution using PCN 729 marking over MPLS for just admission control alone, or just flow pre- 730 emption alone, particularly if codepoint space was at a premium in 731 the MPLS EXP field. However, the feasibility of deploying one 732 without the other would require further study. 734 9. Deployment Considerations 736 9.1. Marking non-ECN Capable Packets 738 What is the consequences of marking a packet that is not ECN-capable? 739 Even if it will be dropped before leaving the domain, doesn't this 740 consume resources unnecessarily? 742 The problem only arises if there is congestion downstream of an 743 earlier congested node. It might be that marked packets are carried 744 through this second congested router when, within the underlying IP 745 header they are not ECN capable, so they will be dropped later. Such 746 packets might cause other packets to be marked (or dropped) that 747 would not otherwise have been. 749 We decided to use the per-domain ECT checking approach because it 750 would become optimal as ECN deployment became prevalent. The 751 situation where traffic is carried beyond a congested LSR only to be 752 dropped later should become less prevalent as more transports use 753 ECN. This is why we chose not to use the [Floyd] alternative which 754 introduced a low but persistent level of unnecessary packet drop for 755 all time. Although that scheme did not carry droppable traffic to 756 the edge of the MPLS domain, we felt this was a small price to pay, 757 and it was anyway only of concern until ECN had become more widely 758 deployed. 760 A partial solution would be to preferentially drop packets arriving 761 at a congested router that were already marked. There is no solution 762 to the problem of marking a packet when congestion is caused by 763 another packet that should have been dropped. However, the chance of 764 such an occurrence is very low and the consequences are not 765 significant. It merely causes an application to very occasionally 766 slow down its rate when it did not have to. 768 9.2. Non-ECN capable routers in an MPLS Domain 770 What if an MPLS domain wants to use ECN, but not all legacy routers 771 are able to support it? 773 If the legacy router(s) are used in the interior, this is not a 774 problem. They will simply have to drop the packets if they are 775 congested, rather than mark them, which is the standard behavior for 776 IP routers that are not ECN-enabled. 778 If the legacy router were used as an egress router, it would not be 779 able to check the ECN capability of the transport correctly. An 780 operator in this position would not be able to use this solution and 781 therefore MUST NOT enable ECN unless all egress routers are ECN- 782 capable. 784 10. IANA Considerations 786 This document makes no request of IANA. 788 Note to RFC Editor: this section may be removed on publication as an 789 RFC. 791 11. Security Considerations 793 We believe no new vulnerabilities are introduced by this draft. 795 We have considered whether malicious sources might be able to exploit 796 the fact that interior LSRs will mark packets that are Not-ECT, 797 relying on their egress LSR to drop them. Although this might allow 798 sources to engineer a situation where more traffic is carried across 799 an MPLS domain than should be, we figured that even if we hadn't 800 introduced this feature, these sources would have been able to 801 prevent these LSRs dropping this traffic anyway, simply by setting 802 ECT in the first place. 804 An ECN sender can use the ECN nonce [RFC3540] to detect a misbehaving 805 receiver. The ECN nonce works correctly across an MPLS domain 806 without requiring any specific support from the proposal in this 807 draft. The nonce does not need to be present in the MPLS shim 808 header. As long as the nonce is present in the IP header when the 809 ECN information is copied from the last MPLS shim header, it will be 810 overwritten if congestion has been experienced by an LSR. This is 811 all that is necessary for the sender to detect a misbehaving 812 receiver. 814 An alternative proposal currently in progress in the IETF 815 [I-D.briscoe-tsvwg-re-ecn-tcp] allows the network to prevent 816 misbehavior by senders or receivers or other routers. Like the ECN 817 nonce, it works correctly without requiring any specific support from 818 the proposal in this draft. It uses a bit in the IP header (the RE 819 bit) which is set by the sender and never changed along the path-it 820 is only read by certain policing elements in the network. There is 821 no need for a copy of this bit in the MPLS shim, as policing nodes 822 can examine the IP header if they need to, particularly given they 823 are intended to only be necessary at domain borders where MPLS 824 headers are often removed. 826 12. Acknowledgments 828 Thanks to K.K. Ramakrishnan and Sally Floyd for getting us thinking 829 about this in the first place and for providing advice on tunneling 830 of ECN packets, and to Joe Babiarz, Ben Niven-Jenkins, Phil Eardley, 831 and Ruediger Geib for their comments on the draft. 833 13. References 835 13.1. Normative References 837 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 838 Requirement Levels", BCP 14, RFC 2119, March 1997. 840 [RFC2475] Blake, S., Black, D., Carlson, M., Davies, E., Wang, Z., 841 and W. Weiss, "An Architecture for Differentiated 842 Services", RFC 2475, December 1998. 844 [RFC3031] Rosen, E., Viswanathan, A., and R. Callon, "Multiprotocol 845 Label Switching Architecture", RFC 3031, January 2001. 847 [RFC3032] Rosen, E., Tappan, D., Fedorkow, G., Rekhter, Y., 848 Farinacci, D., Li, T., and A. Conta, "MPLS Label Stack 849 Encoding", RFC 3032, January 2001. 851 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 852 of Explicit Congestion Notification (ECN) to IP", 853 RFC 3168, September 2001. 855 [RFC3260] Grossman, D., "New Terminology and Clarifications for 856 Diffserv", RFC 3260, April 2002. 858 [RFC3270] Le Faucheur, F., Wu, L., Davie, B., Davari, S., Vaananen, 859 P., Krishnan, R., Cheval, P., and J. Heinanen, "Multi- 860 Protocol Label Switching (MPLS) Support of Differentiated 861 Services", RFC 3270, May 2002. 863 13.2. Informative References 865 [Floyd] "A Proposal to Incorporate ECN in MPLS", 1999. 867 Work in progress. http://www.icir.org/floyd/papers/ 868 draft-ietf-mpls-ecn-00.txt 870 [I-D.briscoe-tsvwg-cl-architecture] 871 Briscoe, B., "An edge-to-edge Deployment Model for Pre- 872 Congestion Notification: Admission Control over a 873 DiffServ Region", draft-briscoe-tsvwg-cl-architecture-04 874 (work in progress), October 2006. 876 [I-D.briscoe-tsvwg-cl-phb] 877 Briscoe, B., "Pre-Congestion Notification marking", 878 draft-briscoe-tsvwg-cl-phb-03 (work in progress), 879 October 2006. 881 [I-D.briscoe-tsvwg-re-ecn-border-cheat] 882 Briscoe, B., "Emulating Border Flow Policing using Re-ECN 883 on Bulk Data", draft-briscoe-tsvwg-re-ecn-border-cheat-01 884 (work in progress), June 2006. 886 [I-D.briscoe-tsvwg-re-ecn-tcp] 887 Briscoe, B., "Re-ECN: Adding Accountability for Causing 888 Congestion to TCP/IP", draft-briscoe-tsvwg-re-ecn-tcp-03 889 (work in progress), October 2006. 891 [I-D.chan-tsvwg-diffserv-class-aggr] 892 Chan, K., "Aggregation of DiffServ Service Classes", 893 draft-chan-tsvwg-diffserv-class-aggr-03 (work in 894 progress), January 2006. 896 [I-D.ietf-nsis-rmd] 897 Bader, A., "RMD-QOSM - The Resource Management in Diffserv 898 QOS Model", draft-ietf-nsis-rmd-08 (work in progress), 899 October 2006. 901 [I-D.lefaucheur-rsvp-ecn] 902 Faucheur, F., "RSVP Extensions for Admission Control over 903 Diffserv using Pre-congestion Notification (PCN)", 904 draft-lefaucheur-rsvp-ecn-01 (work in progress), 905 June 2006. 907 [RFC3540] Spring, N., Wetherall, D., and D. Ely, "Robust Explicit 908 Congestion Notification (ECN) Signaling with Nonces", 909 RFC 3540, June 2003. 911 [RFC4340] Kohler, E., Handley, M., and S. Floyd, "Datagram 912 Congestion Control Protocol (DCCP)", RFC 4340, March 2006. 914 [Shayman] "Using ECN to Signal Congestion Within an MPLS Domain", 915 2000. 917 Work in progress. http://www.ee.umd.edu/~shayman/papers.d/ 918 draft-shayman-mpls-ecn-00.txt 920 Authors' Addresses 922 Bruce Davie 923 Cisco Systems, Inc. 924 1414 Mass. Ave. 925 Boxborough, MA 01719 926 USA 928 Email: bsd@cisco.com 930 Bob Briscoe 931 BT Research 932 B54/77, Sirius House 933 Adastral Park 934 Martlesham Heath 935 Ipswich 936 Suffolk IP5 3RE 937 United Kingdom 939 Email: bob.briscoe@bt.com 941 June Tay 942 BT Research 943 B54/77, Sirius House 944 Adastral Park 945 Martlesham Heath 946 Ipswich 947 Suffolk IP5 3RE 948 United Kingdom 950 Email: june.tay@bt.com 952 Full Copyright Statement 954 Copyright (C) The IETF Trust (2007). 956 This document is subject to the rights, licenses and restrictions 957 contained in BCP 78, and except as set forth therein, the authors 958 retain all their rights. 960 This document and the information contained herein are provided on an 961 "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS 962 OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND 963 THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS 964 OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF 965 THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED 966 WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. 968 Intellectual Property 970 The IETF takes no position regarding the validity or scope of any 971 Intellectual Property Rights or other rights that might be claimed to 972 pertain to the implementation or use of the technology described in 973 this document or the extent to which any license under such rights 974 might or might not be available; nor does it represent that it has 975 made any independent effort to identify any such rights. Information 976 on the procedures with respect to rights in RFC documents can be 977 found in BCP 78 and BCP 79. 979 Copies of IPR disclosures made to the IETF Secretariat and any 980 assurances of licenses to be made available, or the result of an 981 attempt made to obtain a general license or permission for the use of 982 such proprietary rights by implementers or users of this 983 specification can be obtained from the IETF on-line IPR repository at 984 http://www.ietf.org/ipr. 986 The IETF invites any interested party to bring to its attention any 987 copyrights, patents or patent applications, or other proprietary 988 rights that may cover technology that may be required to implement 989 this standard. Please address the information to the IETF at 990 ietf-ipr@ietf.org. 992 Acknowledgment 994 Funding for the RFC Editor function is provided by the IETF 995 Administrative Support Activity (IASA).