idnits 2.17.1 draft-ietf-v6ops-design-choices-00.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** 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 216: '... A router MUST be able to determi...' Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (February 14, 2013) is 4079 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Outdated reference: A later version (-16) exists of draft-ietf-idr-dynamic-cap-14 == Outdated reference: A later version (-11) exists of draft-ietf-opsec-lla-only-02 == Outdated reference: A later version (-06) exists of draft-ietf-v6ops-enterprise-incremental-ipv6-01 Summary: 1 error (**), 0 flaws (~~), 4 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 V6OPS Working Group P. Matthews 3 Internet-Draft Alcatel-Lucent 4 Intended status: Informational February 14, 2013 5 Expires: August 18, 2013 7 Design Choices for IPv6 Networks 8 draft-ietf-v6ops-design-choices-00 10 Abstract 12 This document presents advice on the design choices that arise when 13 designing IPv6 networks (both dual-stack and IPv6-only). The 14 intended audience is someone designing an IPv6 network who is 15 knowledgeable about best current practices around IPv4 network 16 design, and wishes to learn the corresponding practices for IPv6. 18 Status of this Memo 20 This Internet-Draft is submitted 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, 2013. 35 Copyright Notice 37 Copyright (c) 2013 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. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 53 2. Design Choices . . . . . . . . . . . . . . . . . . . . . . . . 3 54 2.1. Mix IPv4 and IPv6 on the Same Link? . . . . . . . . . . . 3 55 2.2. Links with Only Link-Local Addresses? . . . . . . . . . . 4 56 2.3. Link-Local Next-Hop in a Static Route? . . . . . . . . . . 5 57 2.4. Separate or Combined eBGP Sessions? . . . . . . . . . . . 6 58 2.5. eBGP Endpoints: Global or Link-Local Addresses? . . . . . 7 59 3. General Observations . . . . . . . . . . . . . . . . . . . . . 8 60 3.1. Use of Link-Local Addresses . . . . . . . . . . . . . . . 9 61 3.2. Separation of IPv4 and IPv6 . . . . . . . . . . . . . . . 9 62 4. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 10 63 5. Security Considerations . . . . . . . . . . . . . . . . . . . 10 64 6. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 10 65 7. Informative References . . . . . . . . . . . . . . . . . . . . 10 66 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . . 12 68 1. Introduction 70 This document presents advice on the design choices that arise when 71 designing IPv6 networks (both dual-stack and IPv6-only). The 72 intended audience is someone designing an IPv6 network who is 73 knowledgeable about best current practices around IPv4 network 74 design, and wishes to learn the corresponding practices for IPv6. 76 The focus of the document is on design choices where there are 77 differences between IPv4 and IPv6, either in the range of possible 78 alternatives (e.g. the extra possibilities introduced by link-local 79 addresses in IPv6) or the recommended alternative. The document 80 presents the alternatives and discusses the pros and cons in detail. 81 Where consensus currently exists around the best practice, this is 82 documented; otherwise the document simply summarizes the current 83 state of the discussion. Thus this document serves to both to 84 document the reasoning behind best current practices for IPv6, and to 85 allow a designer to make an intelligent choice where no such 86 consensus exists. 88 This document does not present advice on strategies for adding IPv6 89 to a network, nor does it discuss transition mechanisms. For advice 90 in these areas, see [RFC6180] for general advice, [RFC6782] for 91 wireline service providers, [RFC6342] for mobile network providers, 92 [RFC5963] for exchange point operators, [I-D.ietf-v6ops-icp-guidance] 93 for content providers, and both [RFC4852] and 94 [I-D.ietf-v6ops-enterprise-incremental-ipv6] for enterprises. Nor 95 does the document cover the ins and outs of creating an IPv6 96 addressing plan; for advice in this area, see [RFC5375]. 98 This document focuses on unicast network design only. It does not 99 cover multicast, nor supporting infrastructure such as DNS. 101 The current version is still work in progress, and it is expected 102 that the presentation and discussion of additional design choices 103 will be added as the document matures. 105 2. Design Choices 107 This section consists of a list of specific design choices a network 108 designer faces when designing an IPv6-only or dual-stack network, 109 along with guidance and advice to the designer when making a choice. 111 2.1. Mix IPv4 and IPv6 on the Same Link? 113 Should IPv4 and IPv6 traffic be logically separated on a link? That 114 is: 116 a. Mix IPv4 and IPv6 traffic on the same layer 2 connection, OR 118 b. Separate IPv4 and IPv6 by using separate physical or logical 119 links (e.g., two physical links or two VLANs on the same link)? 121 Option (a) implies a single layer 3 interface at each end with both 122 IPv4 and IPv6 addresses; while option (b) implies two layer 3 123 interfaces, one for IPv4 addresses and one with IPv6 addresses. 125 The advantages of option (a) include: 127 o Requires only half as many layer 3 interfaces as option (b), thus 128 providing better scaling; 130 o May require fewer physical ports, thus saving money; 132 o Can make the QoS implementation much easier (for example, rate- 133 limiting the combined IPv4 and IPv6 traffic to or from a 134 customer); 136 o Provides better support for the expected future of increasing IPv6 137 traffic and decreasing IPv4 traffic; 139 o And is generally conceptually simpler. 141 For these reasons, there is a pretty strong consensus in the operator 142 community that option (a) is the preferred way to go. 144 However, there can be times when option (b) is the pragmatic choice. 145 Most commonly, option (b) is used to work around limitations in 146 network equipment. One big example is the generally poor level of 147 support today for individual statistics on IPv4 traffic vs IPv6 148 traffic when option (a) is used. Other, device-specific, limitations 149 exist as well. It is expected that these limitations will go away as 150 support for IPv6 matures, making option (b) less and less attractive 151 until the day that IPv4 is finally turned off. 153 Most networks today use option (a) wherever possible. 155 2.2. Links with Only Link-Local Addresses? 157 Should the link: 159 a. Use only link-local addresses ("unnumbered"), OR 161 b. Have global or unique-local addresses assigned in addition to 162 link-locals? 164 There are two advantages of unnumbered links. The first advantage is 165 ease of configuration. In a network with a large number of 166 unnumbered links, the operator can just enable an IGP on each router, 167 without going through the tedious process of assigning and tracking 168 the addresses for each link. The second advantage is security. 169 Since link-local addresses are unroutable, the associated interfaces 170 cannot be attacked from an off-link device. This implies less effort 171 around maintaining security ACLs. 173 Countering this advantage are various disadvantages to unnumbered 174 links in IPv6: 176 o It is not possible to ping an interface that has only a link-local 177 address from a device that is not directly attached to the link. 178 Thus, to troubleshoot, one must typically log into a device that 179 is directly attached to the device in question, and execute the 180 ping from there. 182 o A traceroute passing over the unnumbered link will return the 183 loopback or system address of the router, rather than the address 184 of the interface itself. 186 o On some devices, by default the link-layer address of the 187 interface is derived from the MAC address assigned to interface. 188 When this is done, swapping out the interface hardware (e.g. 189 interface card) will cause the link-layer address to change. In 190 some cases (peering config, ACLs, etc) this may require additional 191 changes. However, many devices allow the link-layer address of an 192 interface to be explicitly configured, which avoids this issue. 194 o The practice of naming router interfaces using DNS names is 195 difficult-to-impossible when using LLAs only. 197 o It is not possible to identify the interface or link (in a 198 database, email, etc) by just giving its address. 200 For more discussion on the pros and cons, see 201 [I-D.ietf-opsec-lla-only]. 203 Today, most operators use numbered links (option b). 205 2.3. Link-Local Next-Hop in a Static Route? 207 What form of next-hop address should one use in a static route? 209 a. Use the far-end's link-local address as the next-hop address, OR 210 b. Use the far-end's GUA/ULA address as the next-hop address? 212 Recall that the IPv6 specs for OSPF [RFC5340] and ISIS [RFC5308] 213 dictate that they always use link-locals for next-hop addresses. For 214 static routes, [RFC4861] section 8 says: 216 A router MUST be able to determine the link-local address for each 217 of its neighboring routers in order to ensure that the target 218 address in a Redirect message identifies the neighbor router by 219 its link-local address. For static routing, this requirement 220 implies that the next-hop router's address should be specified 221 using the link-local address of the router. 223 This implies that using a GUA or ULA as the next hop will prevent a 224 router from sending Redirect messages for packets that "hit" this 225 static route. All this argues for using a link-local as the next-hop 226 address in a static route. 228 However, there are two cases where using a link-local address as the 229 next-hop clearly does not work. One is when the static route is an 230 indirect (or multi-hop) static route. The second is when the static 231 route is redistributed into another routing protocol. In these 232 cases, the above text from RFC 4861 notwithstanding, either a GUA or 233 ULA must be used. 235 Furthermore, many network operators are concerned about the 236 dependency of the default link-local address on an underlying MAC 237 address, as described in the previous section. 239 Today most operators use GUAs as next-hop addresses. 241 2.4. Separate or Combined eBGP Sessions? 243 For a dual-stack peering connection where eBGP is used as the routing 244 protocol, then one can either: 246 a. Use one BGP session to carry both IPv4 and IPv6 routes, OR 248 b. Use two BGP sessions, a session over IPv4 carrying IPv4 routes 249 and a session over IPv6 carrying IPv6 routes. 251 The main advantage of (a) is a reduction in the number of BGP 252 sessions compared with (b). 254 However, there are a number of concerns with option (a): 256 o On most existing implementations, adding or removing an address 257 family to an established BGP session will cause the router to tear 258 down and re-establish the session. Thus adding the IPv6 family to 259 an existing session carrying just IPv4 routes will disrupt the 260 session, and the eventual removal of IPv4 from the dual IPv4/IPv6 261 session will also disrupt the session. This disruption problem 262 will persist until something similar to [I-D.ietf-idr-dynamic-cap] 263 or [I-D.ietf-idr-bgp-multisession] is widely deployed. 265 o Whatever selection you make for the underlying transport protocol 266 (v4 or v6) will likely look funny at some date. Using two 267 sessions is appropriate both now and in the future. 269 o Carrying (for example) IPv6 routes over IPv4 means that route 270 information is transported over a different transport plane than 271 the data packets themselves. If v6 connectivity goes down locally 272 without v4 also going down, then v6 routes will still be 273 exchanged, thus leading to a blackhole. 275 o In some implementations, carrying v4 routes in a BGP session over 276 v6 transport (or vica-versa) results in the BGP next-hops in the 277 wrong address family, which must be fixed using routing policy 278 before the routes can be used. 280 Given these disadvantages, option (b) is the better choice in most 281 situations, and this is the choice selected in most networks today. 283 2.5. eBGP Endpoints: Global or Link-Local Addresses? 285 When running eBGP over IPv6, there are two options for the addresses 286 to use at each end of the eBGP session (or more properly, the 287 underlying TCP session): 289 a. Use link-local addresses for the eBGP session, OR 291 b. Use global addresses for the eBGP session. 293 Note that the choice here is the addresses to use for the eBGP 294 sessions, and not whether the link itself has global (or unique- 295 local) addresses. In particular, it is quite possible for the eBGP 296 session to use link-local addresses even when the link has global 297 addresses. 299 The big attraction for option (a) is security: an eBGP session using 300 link-local addresses is impossible to attack from a device that is 301 off-link. This provides very strong protection against TCP RST and 302 similar attacks. Though there are other ways to get an equivalent 303 level of security (e.g. GTSM [RFC5082], MD5 [RFC5925], or ACLs), 304 these other ways require additional configuration which can be 305 forgotten or potentially mis-configured. 307 However, there are a number of small disadvantages to using link- 308 local addresses: 310 o Using link-local addresses only works for single-hop eBGP 311 sessions; it does not work for multi-hop sessions. 313 o One must use "next-hop self" at both endpoints, otherwise 314 redistributing routes learned via eBGP into iBGP will not work. 315 (Some products enable "next-hop self" in this situation 316 automatically). 318 o Operators and their tools are used to referring to eBGP sessions 319 by address only, something that is not possible with link-local 320 addresses. 322 o If one is configuring parallel eBGP sessions for IPv4 and IPv6 323 routes, then using link-local addresses for the IPv6 session 324 introduces an extra difference between the two sessions which 325 could otherwise be avoided. 327 o On some products, an eBGP session using a link-local address is 328 more complex to configure than a session that use a global 329 address. 331 o If hardware or other issues cause one to move the cable to a 332 different local interface, then reconfiguration is required at 333 both ends: at the local end because the interface has changed (and 334 with link-local addresses, the interface must always be specified 335 along with the address), and at the remote end because the link- 336 local address has likely changed. (Contrast this with using 337 global addresses, where less re-configuration is required at the 338 local end, and no reconfiguration is required at the remote end). 340 o Finally, a strict interpretation of RFC 2545 can be seen as 341 forbidding running eBGP between link-local addresses, as RFC 2545 342 requires the BGP next-hop field to contain at least a global 343 address. 345 For these reasons, most operators today choose to have their eBGP 346 sessions use global addresses. 348 3. General Observations 350 There are two themes that run though many of the design choices in 351 this document. This section presents some general discussion on 352 these two themes. 354 3.1. Use of Link-Local Addresses 356 The proper use of link-local addresses is a common theme in the IPv6 357 network design choices. Link-layer addresses are, of course, always 358 present in an IPv6 network, but current network design practice 359 mostly ignores them, despite efforts such as 360 [I-D.ietf-opsec-lla-only]. 362 There are three main reasons for this current practice: 364 o Network operators are concerned about the volatility of link-local 365 addresses based on MAC addresses, despite the fact that this 366 concern can be overcome by manually-configuring link-local 367 addresses; 369 o It is impossible to ping a link-local address from a device that 370 is not on the same subnet. This is a troubleshooting 371 disadvantage, though it can also be viewed as a security 372 advantage. 374 o Most operators are currently running networks that carry both IPv4 375 and IPv6 traffic, and wish to harmonize their IPv4 and IPv6 design 376 and operational practices where possible. 378 3.2. Separation of IPv4 and IPv6 380 Currently, most operators are running or planning to run networks 381 that carry both IPv4 and IPv6 traffic. Hence the question: To what 382 degree should IPv4 and IPv6 be kept separate? As can be seen above, 383 this breaks into two sub-questions: To what degree should IPv4 and 384 IPv6 traffic be kept separate, and to what degree should IPv4 and 385 IPv6 routing information be kept separate? 387 The general consensus around the first question is that IPv4 and IPv6 388 traffic should generally be mixed together. This recommendation is 389 driven by the operational simplicity of mixing the traffic, plus the 390 general observation that the service being offered to the end user is 391 Internet connectivity and most users do not know or care about the 392 differences between IPv4 and IPv6. Thus it is very desirable to mix 393 IPv4 and IPv6 on the same link to the end user. On other links, 394 separation is possible but more operationally complex, though it does 395 occasionally allow the operator to work around limitations on network 396 devices. The situation here is roughly comparable to IP and MPLS 397 traffic: many networks mix the two traffic types on the same links 398 without issues. 400 By contrast, there is more of an argument for carrying IPv6 routing 401 information over IPv6 transport, while leaving IPv4 routing 402 information on IPv4 transport. By doing this, one gets fate-sharing 403 between the control and data plane for each IP protocol version: if 404 the data plane fails for some reason, then often the control plane 405 will too. 407 4. IANA Considerations 409 This document makes no requests of IANA. 411 5. Security Considerations 413 (TBD) 415 6. Acknowledgements 417 Many, many people in the V6OPS working group provided comments and 418 suggestions that made their way into this document. A partial list 419 includes: Rajiv Asati, Fred Baker, Michael Behringer, Marc Blanchet, 420 Ron Bonica, Randy Bush, Cameron Byrne, Brian Carpenter, KK 421 Chittimaneni, Tim Chown, Lorenzo Colitti, Gert Doering, Bill Fenner, 422 Kedar K Gaonkar, Chris Grundemann, Steinar Haug, Ray Hunter, Joel 423 Jaeggli, Victor Kuarsingh, Ivan Pepelnjak, Alexandru Petrescu, Rob 424 Shakir, Mark Smith, Jean-Francois Tremblay, Tina Tsou, Dan York, and 425 Xuxiaohu. 427 I would also like to thank Pradeep Jain and Alastair Johnson for 428 helpful comments on a very preliminary version of this document. 430 7. Informative References 432 [I-D.ietf-idr-bgp-multisession] 433 Scudder, J., Appanna, C., and I. Varlashkin, "Multisession 434 BGP", draft-ietf-idr-bgp-multisession-07 (work in 435 progress), September 2012. 437 [I-D.ietf-idr-dynamic-cap] 438 Ramachandra, S. and E. Chen, "Dynamic Capability for 439 BGP-4", draft-ietf-idr-dynamic-cap-14 (work in progress), 440 December 2011. 442 [I-D.ietf-opsec-lla-only] 443 Behringer, M. and E. Vyncke, "Using Only Link-Local 444 Addressing Inside an IPv6 Network", 445 draft-ietf-opsec-lla-only-02 (work in progress), 446 October 2012. 448 [I-D.ietf-v6ops-enterprise-incremental-ipv6] 449 Chittimaneni, K., Chown, T., Howard, L., Kuarsingh, V., 450 Pouffary, Y., and E. Vyncke, "Enterprise IPv6 Deployment 451 Guidelines", 452 draft-ietf-v6ops-enterprise-incremental-ipv6-01 (work in 453 progress), September 2012. 455 [I-D.ietf-v6ops-icp-guidance] 456 Carpenter, B. and S. Jiang, "IPv6 Guidance for Internet 457 Content and Application Service Providers", 458 draft-ietf-v6ops-icp-guidance-05 (work in progress), 459 January 2013. 461 [RFC4852] Bound, J., Pouffary, Y., Klynsma, S., Chown, T., and D. 462 Green, "IPv6 Enterprise Network Analysis - IP Layer 3 463 Focus", RFC 4852, April 2007. 465 [RFC4861] Narten, T., Nordmark, E., Simpson, W., and H. Soliman, 466 "Neighbor Discovery for IP version 6 (IPv6)", RFC 4861, 467 September 2007. 469 [RFC5082] Gill, V., Heasley, J., Meyer, D., Savola, P., and C. 470 Pignataro, "The Generalized TTL Security Mechanism 471 (GTSM)", RFC 5082, October 2007. 473 [RFC5308] Hopps, C., "Routing IPv6 with IS-IS", RFC 5308, 474 October 2008. 476 [RFC5340] Coltun, R., Ferguson, D., Moy, J., and A. Lindem, "OSPF 477 for IPv6", RFC 5340, July 2008. 479 [RFC5375] Van de Velde, G., Popoviciu, C., Chown, T., Bonness, O., 480 and C. Hahn, "IPv6 Unicast Address Assignment 481 Considerations", RFC 5375, December 2008. 483 [RFC5925] Touch, J., Mankin, A., and R. Bonica, "The TCP 484 Authentication Option", RFC 5925, June 2010. 486 [RFC5963] Gagliano, R., "IPv6 Deployment in Internet Exchange Points 487 (IXPs)", RFC 5963, August 2010. 489 [RFC6180] Arkko, J. and F. Baker, "Guidelines for Using IPv6 490 Transition Mechanisms during IPv6 Deployment", RFC 6180, 491 May 2011. 493 [RFC6342] Koodli, R., "Mobile Networks Considerations for IPv6 494 Deployment", RFC 6342, August 2011. 496 [RFC6782] Kuarsingh, V. and L. Howard, "Wireline Incremental IPv6", 497 RFC 6782, November 2012. 499 Author's Address 501 Philip Matthews 502 Alcatel-Lucent 503 600 March Road 504 Ottawa, Ontario K2K 2E6 505 Canada 507 Phone: +1 613-784-3139 508 Email: philip_matthews@magma.ca