idnits 2.17.1 draft-ietf-6man-flow-update-03.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 a both a reference to RFC 2119 and the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords. RFC 2119 keyword, line 80: '...ue set by the source MUST be delivered...' RFC 2119 keyword, line 82: '... b. "IPv6 nodes MUST NOT assume any m...' RFC 2119 keyword, line 84: '...uter performance SHOULD NOT be depende...' RFC 2119 keyword, line 188: '...o say that hosts SHOULD set a pseudo-r...' Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (February 26, 2011) is 4780 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Outdated reference: A later version (-03) exists of draft-gont-6man-flowlabel-security-01 == Outdated reference: A later version (-03) exists of draft-hu-flow-label-cases-02 == Outdated reference: A later version (-07) exists of draft-ietf-6man-flow-3697bis-00 -- Obsolete informational reference (is this intentional?): RFC 2460 (Obsoleted by RFC 8200) -- Obsolete informational reference (is this intentional?): RFC 2629 (Obsoleted by RFC 7749) -- Obsolete informational reference (is this intentional?): RFC 3697 (Obsoleted by RFC 6437) Summary: 1 error (**), 0 flaws (~~), 4 warnings (==), 4 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 6MAN S. Amante 3 Internet-Draft Level 3 4 Intended status: Informational B. Carpenter 5 Expires: August 28, 2011 Univ. of Auckland 6 S. Jiang 7 Huawei Technologies Co., Ltd 8 February 26, 2011 10 Rationale for update to the IPv6 flow label specification 11 draft-ietf-6man-flow-update-03 13 Abstract 15 Various published proposals for use of the IPv6 flow label are 16 incompatible with its original specification in RFC 3697. 17 Furthermore, very little practical use is made of the flow label, 18 partly due to some uncertainties about the correct interpretation of 19 the specification. This document discusses and motivates changes to 20 the specification in order to clarify it, and to introduce some 21 additional flexibility. 23 Status of this Memo 25 This Internet-Draft is submitted in full conformance with the 26 provisions of BCP 78 and BCP 79. 28 Internet-Drafts are working documents of the Internet Engineering 29 Task Force (IETF). Note that other groups may also distribute 30 working documents as Internet-Drafts. The list of current Internet- 31 Drafts is at http://datatracker.ietf.org/drafts/current/. 33 Internet-Drafts are draft documents valid for a maximum of six months 34 and may be updated, replaced, or obsoleted by other documents at any 35 time. It is inappropriate to use Internet-Drafts as reference 36 material or to cite them other than as "work in progress." 38 This Internet-Draft will expire on August 28, 2011. 40 Copyright Notice 42 Copyright (c) 2011 IETF Trust and the persons identified as the 43 document authors. All rights reserved. 45 This document is subject to BCP 78 and the IETF Trust's Legal 46 Provisions Relating to IETF Documents 47 (http://trustee.ietf.org/license-info) in effect on the date of 48 publication of this document. Please review these documents 49 carefully, as they describe your rights and restrictions with respect 50 to this document. Code Components extracted from this document must 51 include Simplified BSD License text as described in Section 4.e of 52 the Trust Legal Provisions and are provided without warranty as 53 described in the Simplified BSD License. 55 Table of Contents 57 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 58 2. Impact of current specification . . . . . . . . . . . . . . . 3 59 3. Changes to specification . . . . . . . . . . . . . . . . . . . 6 60 4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 7 61 5. Security Considerations . . . . . . . . . . . . . . . . . . . 8 62 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 8 63 7. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 8 64 8. Change log . . . . . . . . . . . . . . . . . . . . . . . . . . 9 65 9. Informative References . . . . . . . . . . . . . . . . . . . . 9 66 Appendix A. Alternative Approaches . . . . . . . . . . . . . . . 10 67 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 11 69 1. Introduction 71 The flow label field in the IPv6 header was reserved but left 72 experimental by [RFC2460], which mandates only that "Hosts or routers 73 that do not support the functions of the Flow Label field are 74 required to set the field to zero when originating a packet, pass the 75 field on unchanged when forwarding a packet, and ignore the field 76 when receiving a packet." 78 The flow label field was normatively specified by [RFC3697]. In 79 particular, we quote three rules from that RFC: 80 a. "The Flow Label value set by the source MUST be delivered 81 unchanged to the destination node(s)." 82 b. "IPv6 nodes MUST NOT assume any mathematical or other properties 83 of the Flow Label values assigned by source nodes." 84 c. "Router performance SHOULD NOT be dependent on the distribution 85 of the Flow Label values. Especially, the Flow Label bits alone 86 make poor material for a hash key." 88 Additionally, RFC 3697 leaves it undefined what method a host should 89 adopt by default to choose the value of the flow label, if no 90 specific method is in use. It was expected that various signaling 91 methods might be defined for agreeing on values of the flow label, 92 but no such methods have been standardised, except a pre-existing 93 option in RSVP [RFC2205]. 95 The flow label is hardly used in practice in widespread IPv6 96 implementations, although some operating systems do set it 97 [McGann05]. To some extent this is due to the main focus being on 98 basic deployment of IPv6, but the absence of a default method of 99 choosing the flow label value means that most host implementations 100 simply set it to zero. There is also anecdotal evidence that the 101 rules quoted above have led to uncertainty about exactly what is 102 possible. Furthermore, various use cases have been proposed that 103 infringe one or another of the rules. None of these proposals has 104 been accepted as a standard and in practice there is no significant 105 deployment of any mechanism to set the flow label. 107 The intention of this document is to explain this situation in more 108 detail and to motivate changes to RFC 3697 intended to remove the 109 uncertainties and encourage active usage of the flow label. It does 110 not formally update RFC 3697, but it serves as background material 111 for [I-D.ietf-6man-flow-3697bis]. 113 2. Impact of current specification 115 Rule (a) makes it impossible for the routing system to use the flow 116 label as any form of dynamic routing tag. This was a conscious 117 choice in the early design of IPv6 and there appears to be no 118 practical possibility of revisiting this choice at this stage in the 119 deployment of IPv6, which uses conventional routing mechanisms like 120 those used for IPv4. However, this rule also makes it impossible to 121 make any use at all of the flow label unless hosts choose to set it. 122 It also forbids clearing the flow label for security reasons. 124 This last point highlights the security properties, or rather the 125 lack of them, of the flow label. The flow label field is always 126 unprotected as it travels through the network, because there is no 127 IPv6 header checksum, and the flow label is not included in transport 128 pseudo-header checksums, nor in IPsec checksums. As a result, 129 intentional and malicious changes to its value cannot be detected. 130 Also, it could be used as a covert data channel, since apparently 131 pseudo-random flow label values could in fact consist of covert data. 132 If the flow label were to carry quality of service semantics, then 133 like the diffserv code point [RFC2474], it would not be intrinsically 134 trustworthy across domain boundaries. As a result, some security 135 specialists believe that flow labels should be cleared for safety. 136 These points must be considered when discussing the immutability of 137 the flow label across domain boundaries. 139 Rule (b) appears to forbid any usage in which the bits of the flow 140 label are encoded with a specific semantic meaning. However, the 141 words "MUST NOT assume" are to be interpreted precisely - if a router 142 knows by configuration or by signaling that the flow label has been 143 assigned in a certain way, it can make use of that knowledge. It is 144 not made clear by the rule that there is an implied distinction 145 between stateless models (in which case no assumption may be made) 146 and stateful models (in which the router has explicit knowledge). 148 If the word "alone" is overlooked, rule (c) has sometimes been 149 interpreted to forbid the use of the flow label as part of a hash 150 used by load distribution mechanisms. In this case too, the word 151 "alone" is to be interpreted precisely - a router is allowed to 152 combine the flow label value with other data in order to produce a 153 uniformly distributed hash. 155 Both before and after these rules were laid down, a considerable 156 number of proposals for use of the flow label were published that 157 seem incompatible with them. Numerous examples and an analysis are 158 presented in [I-D.hu-flow-label-cases]. Those examples propose use 159 cases in which some or all of the following apply: 160 o The flow label may be changed by intermediate systems. 161 o It doesn't matter if the flow label is changed, because the 162 receiver doesn't use it. 164 o Some or all bits of the flow label are encoded: they have specific 165 meanings understood by routers and switches along the path. 166 o The encoding is related to the required quality of service, as 167 well as identifying a flow. 168 o The flow label is used to control forwarding or switching in some 169 way. 171 These proposals all require either some form of encoding of semantics 172 in the bits of the flow label, or the ability for routers to modify 173 the flow label, or both. Thus they appear to infringe the rules from 174 RFC 3697 quoted above. 176 We can conclude that a considerable number of researchers and 177 designers have been stymied by RFC 3697. On the other hand, some 178 other proposals discussed in [I-D.hu-flow-label-cases] appear to be 179 compatible with RFC 3697. Several are based on the originator of a 180 packet choosing a pseudo-random flow label for each flow, which is 181 one option suggested in RFC 3697. Thus, we can also conclude that 182 there is a useful role for this approach. 184 If our goal is for the flow label to be used in practice, the 185 conflict between the various approaches creates a dilemma. There 186 appear to be two major options: 187 1. Discourage locally defined and/or stateful use of the flow label. 188 Strengthen RFC 3697 to say that hosts SHOULD set a pseudo-random 189 label value, without creating state, which would clarify and 190 limit its possible uses. In particular, its use for load 191 distribution and balancing would be encouraged. 192 2. Relax the rules to encourage locally defined and/or stateful use 193 of the flow label. This approach would make the flow label 194 completely mutable and would exclude use cases depending on 195 strict end-to-end immutability. It would encourage applications 196 of a pseudo-random flow label, such as load distribution, on a 197 local basis, but it would exclude end-to-end applications. 199 During 2010 there was considerable debate about these options 200 and variants of them, with a variety of proposals in previous 201 versions of this document and in mailing list discussions. After 202 these discussions, there appears to be a view that simplicity should 203 prevail, and that complicated proposals such as defining quality of 204 service semantics in the flow label, or sub-dividing the flow label 205 field into smaller sub-fields, will not prove efficient or 206 deployable, especially in high speed routers. There is also a 207 clearly expressed view that using the flow label for various forms of 208 stateless load distribution is the best simple application for it. 209 At the same time, it is necessary to recognize that the strict 210 immutability rule has drawbacks as noted above. 212 Even under the rules of RFC 3697, the flow label is intrinsically 213 untrustworthy, because modifications en route cannot be detected. 214 For this reason, even with the current strict immutability rule, 215 downstream nodes cannot rely on the value being unchanged. In this 216 sense, any use of the flow label must be viewed as an optimisation on 217 a best effort basis; a packet with a changed (or zero) flow label 218 value should never cause a hard failure. 220 The remainder of this document discusses specific modifications to 221 the standard, which are defined normatively in a companion document 222 [I-D.ietf-6man-flow-3697bis]. 224 3. Changes to specification 226 Although RFC 3697 requires the flow label to be delivered unchanged, 227 as noted above, it is not included in any transport layer pseudo- 228 header checksums nor in IPsec authentication [RFC4302]. Both RFC 229 2460 and RFC 3697 define the default flow label to be zero. At the 230 time of writing, this is the observed value in an overwhelming 231 proportion of IPv6 packets; the most widespread operating systems and 232 applications do not set it, and routers do not rely on it. Thus 233 there is no reason to expect operational difficulties if a careful 234 change is made to the rules of RFC 3697. 236 In particular, the facts that the label is not checksummed and rarely 237 used mean that the current strict immutability of the label can be 238 moderated without serious operational consequences. 240 The purposes of the proposed changes are to remove the uncertainties 241 left by RFC 3697, in order to encourage setting of the flow label by 242 default, and to enable its generic use. The proposed generic use is 243 to encourage pseudo-random flow labels that can be used to assist 244 load distribution balancing. There should be no impact on existing 245 IETF specifications other than RFC 3697 and no impact on currently 246 operational software and hardware. 248 A secondary purpose is to modify the immutability of the flow label 249 in a limited way, to allow hosts that do not set the flow label to 250 benefit from it nevertheless. The fact that the flow label may in 251 practice be changed en route is also reflected in the reformulation 252 of the rules. 254 A general description of the changes follows. The normative text is 255 to be found in [I-D.ietf-6man-flow-3697bis]. 257 The definition of a flow is subtly changed from RFC 3697 to allow any 258 node, not just the source node, to set the flow label value. 260 However, it is recommended that sources should set a pseudo-random 261 flow label value in all flows, replacing the less precise 262 recommendation made in Section 3 of RFC 3697. Both stateful and 263 stateless methods of assigning a pseudo-random value could be used. 265 Section 3 of RFC 3697 also allows nodes to participate in an 266 unspecified method of flow state establishment. The changes do not 267 remove that option, but it is made clear that stateless models are 268 also possible and are the recommended default. 270 The main novelty is that a forwarding node (typically a first-hop or 271 ingress router) may set the flow label value if the source has not 272 done so, according to the same recommendations that apply to the 273 source. This might place a considerable processing load on ingress 274 routers, even if they adopted a stateless method of flow 275 identification and label assignment. 277 The immutability of the flow label, once it has been set, is not 278 changed. However, some qualifications are placed on this property, 279 to allow for the fact that the flow label is an unprotected field and 280 might be changed undetectably. No Internet-wide mechanism can depend 281 mathematically on immutable flow labels. The new rules require that 282 flow labels exported to the Internet should always be either zero or 283 pseudo-random, but even this cannot be relied on mathematically. Use 284 cases need to be robust against non-conforming flow label values. 285 This will also enhance compatibility with any legacy hosts that set 286 the flow label according to RFC 2460 or RFC 3697. 288 4. Discussion 290 The following are some practical consequences of the above changes: 291 o Sending hosts that are not updated will in practice continue to 292 send all-zero labels. If there is no label-setting router along 293 the path taken by a packet, the label will be delivered as zero. 294 o Sending hosts conforming to the new specification will by default 295 choose pseudo-random labels between 1 and 0xFFFFF. 296 o Sending hosts may continue to send all-zero labels, in which case 297 an ingress router may set pseudo-random labels between 1 and 298 0xFFFFF. 299 o The flow label is no longer unrealistically asserted to be 300 strictly immutable; it is recognised that it may, incorrectly, be 301 changed en route. In some circumstances this will break end-to- 302 end usage, e.g. potential detection of third-party spoofing 303 attacks [I-D.gont-6man-flowlabel-security]. 304 o The expected default usage of the flow label is some form of 305 stateless load distribution, such as the ECMP/LAG usage defined in 306 [I-D.carpenter-flow-ecmp]. 308 o If the new rules are followed, all IPv6 traffic flows on the 309 Internet should have zero or pseudo-random flow label values. 311 From an operational viewpoint, existing IPv6 hosts that set a default 312 (zero) flow label value and ignore the flow label on receipt will be 313 unaffected by implementations of the new specification. In general, 314 it is assumed that hosts will ignore the value of the flow label on 315 receipt; it cannot be relied on as an end-to-end signal. However, 316 this doesn't apply if a cryptographically generated label is being 317 used to detect attackers [I-D.gont-6man-flowlabel-security]. 319 Similarly, routers that ignore the flow label will be unaffected by 320 implementations of the specification. 322 Hosts that set a default (zero) flow label but are in a domain where 323 routers set a pseudo-random label as recommended in Section 3 will 324 benefit from whatever flow label handling is used on the path. 326 Hosts and routers that adopt the recommended pseudo-random mechanism 327 will enhance the performance of any load balancing devices that 328 include the flow label in the hash used to select a particular path 329 or server, even when packets leave the local domain. 331 5. Security Considerations 333 See [I-D.ietf-6man-flow-3697bis] and 334 [I-D.gont-6man-flowlabel-security] for full discussion. 336 6. IANA Considerations 338 This document requests no action by IANA. 340 7. Acknowledgements 342 The authors are grateful to Qinwen Hu for general discussion about 343 the flow label and for his work in searching the literature. 344 Valuable comments and contributions were made by Fred Baker, Steve 345 Blake, Remi Despres, Alan Ford, Fernando Gont, Brian Haberman, Tony 346 Hain, Joel Halpern, Chris Morrow, Thomas Narten, Pekka Savola, Mark 347 Smith, Pascal Thubert, Iljitsch van Beijnum, and other participants 348 in the 6man working group. 350 This document was produced using the xml2rfc tool [RFC2629]. 352 8. Change log 354 draft-ietf-6man-flow-update-03: updated to be in styep with RFC 355 3697bis, 2011-02-26 357 draft-ietf-6man-flow-update-02: repurposed as rationale for update of 358 RFC 3697, 2011-01-31 360 draft-ietf-6man-flow-update-01: clarified that this is not a formal 361 update of RFC 3697, clarified text about domains exporting 362 inappropriate labels, 2011-01-10 364 draft-ietf-6man-flow-update-00: adopted as WG document at IETF 79, 365 mutability rules adjusted according to WG discussion, 2010-12-03 367 draft-carpenter-6man-flow-update-04: even more simplified according 368 to WG discussion, 2010-09-16 370 draft-carpenter-6man-flow-update-03: futher simplified according to 371 WG discussion, 2010-05-07 373 draft-carpenter-6man-flow-update-02: revised and simplified according 374 to WG discussion, 2010-04-13 376 draft-carpenter-6man-flow-update-01: revised according to mail list 377 discussion, 2010-03-05 379 draft-carpenter-6man-flow-update-00: original version, 2010-02-18 381 9. Informative References 383 [I-D.carpenter-flow-ecmp] 384 Carpenter, B. and S. Amante, "Using the IPv6 flow label 385 for equal cost multipath routing and link aggregation in 386 tunnels", draft-carpenter-flow-ecmp-03 (work in progress), 387 October 2010. 389 [I-D.gont-6man-flowlabel-security] 390 Gont, F., "Security Assessment of the IPv6 Flow Label", 391 draft-gont-6man-flowlabel-security-01 (work in progress), 392 November 2010. 394 [I-D.hu-flow-label-cases] 395 Hu, Q. and B. Carpenter, "Survey of proposed use cases for 396 the IPv6 flow label", draft-hu-flow-label-cases-02 (work 397 in progress), September 2010. 399 [I-D.ietf-6man-flow-3697bis] 400 Amante, S., Carpenter, B., Jiang, S., and J. Rajahalme, 401 "IPv6 Flow Label Specification", 402 draft-ietf-6man-flow-3697bis-00 (work in progress), 403 January 2011. 405 [I-D.martinbeckman-ietf-ipv6-fls-ipv6flowswitching] 406 Beckman, M., "IPv6 Dynamic Flow Label Switching (FLS)", 407 draft-martinbeckman-ietf-ipv6-fls-ipv6flowswitching-03 408 (work in progress), March 2007. 410 [McGann05] 411 McGann, O. and D. Malone, "Flow Label Filtering 412 Feasibility", European Conference on Computer Network 413 Defence , 2005. 415 [RFC2205] Braden, B., Zhang, L., Berson, S., Herzog, S., and S. 416 Jamin, "Resource ReSerVation Protocol (RSVP) -- Version 1 417 Functional Specification", RFC 2205, September 1997. 419 [RFC2460] Deering, S. and R. Hinden, "Internet Protocol, Version 6 420 (IPv6) Specification", RFC 2460, December 1998. 422 [RFC2474] Nichols, K., Blake, S., Baker, F., and D. Black, 423 "Definition of the Differentiated Services Field (DS 424 Field) in the IPv4 and IPv6 Headers", RFC 2474, 425 December 1998. 427 [RFC2629] Rose, M., "Writing I-Ds and RFCs using XML", RFC 2629, 428 June 1999. 430 [RFC3697] Rajahalme, J., Conta, A., Carpenter, B., and S. Deering, 431 "IPv6 Flow Label Specification", RFC 3697, March 2004. 433 [RFC4302] Kent, S., "IP Authentication Header", RFC 4302, 434 December 2005. 436 Appendix A. Alternative Approaches 438 A model was discussed in an earlier version of this document which 439 defined a notion of 'flow label domain' analogous to a differentiated 440 services domain [RFC2474]. This model would have encouraged local 441 usage of the flow label as an alternative to any form of generic use, 442 but it required complex rules for the behaviour of domain boundary 443 routers, and proved controversial in discussion. 445 Two even more complex alternative approaches were also considered and 446 rejected. 448 The first was to distinguish locally significant flow labels from 449 those conforming to RFC 3697 by setting or clearing the most 450 significant bit (MSB) of the flow label. This led to quite 451 complicated rules, seems impossible to make fully self-consistent, 452 and was not considered practical. 454 The second was to use a specific differentiated services code point 455 (DSCP)[RFC2474] in the Traffic Class octet instead of the MSB of the 456 flow label itself, to flag a locally defined behaviour. A more 457 elaborate version of this was proposed in 458 [I-D.martinbeckman-ietf-ipv6-fls-ipv6flowswitching]. There are two 459 issues with this approach. One is that DSCP values are themselves 460 only locally significant, inconsistent with the end-to-end nature of 461 the original flow label definition. Secondly, it seems unwise to 462 meld the semantics of differentiated services, which are currently 463 deployed, with the unknown future semantics of flow label usage. 464 However, this approach, while not recommended, does not appear to 465 violate any basic principles if applied strictly within a single 466 differentiated services domain. 468 Authors' Addresses 470 Shane Amante 471 Level 3 Communications, LLC 472 1025 Eldorado Blvd 473 Broomfield, CO 80021 474 USA 476 Email: shane@level3.net 478 Brian Carpenter 479 Department of Computer Science 480 University of Auckland 481 PB 92019 482 Auckland, 1142 483 New Zealand 485 Email: brian.e.carpenter@gmail.com 486 Sheng Jiang 487 Huawei Technologies Co., Ltd 488 Huawei Building, No.3 Xinxi Rd., 489 Shang-Di Information Industry Base, Hai-Dian District, Beijing 490 P.R. China 492 Email: shengjiang@huawei.com