idnits 2.17.1 draft-ietf-v6ops-addcon-06.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- ** It looks like you're using RFC 3978 boilerplate. You should update this to the boilerplate described in the IETF Trust License Policy document (see https://trustee.ietf.org/license-info), which is required now. -- Found old boilerplate from RFC 3978, Section 5.1 on line 20. -- Found old boilerplate from RFC 3978, Section 5.5, updated by RFC 4748 on line 1427. -- Found old boilerplate from RFC 3979, Section 5, paragraph 1 on line 1438. -- Found old boilerplate from RFC 3979, Section 5, paragraph 2 on line 1445. -- Found old boilerplate from RFC 3979, Section 5, paragraph 3 on line 1451. 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 (October 1, 2007) is 6024 days in the past. Is this intentional? -- Found something which looks like a code comment -- if you have code sections in the document, please surround them with '' and '' lines. Checking references for intended status: Informational ---------------------------------------------------------------------------- -- Looks like a reference, but probably isn't: 'RIID' on line 496 -- Obsolete informational reference (is this intentional?): RFC 2462 (ref. '2') (Obsoleted by RFC 4862) -- Obsolete informational reference (is this intentional?): RFC 2471 (ref. '3') (Obsoleted by RFC 3701) -- Obsolete informational reference (is this intentional?): RFC 3041 (ref. '6') (Obsoleted by RFC 4941) -- Obsolete informational reference (is this intentional?): RFC 3177 (ref. '9') (Obsoleted by RFC 6177) -- Obsolete informational reference (is this intentional?): RFC 3315 (ref. '11') (Obsoleted by RFC 8415) -- Obsolete informational reference (is this intentional?): RFC 3484 (ref. '12') (Obsoleted by RFC 6724) -- Obsolete informational reference (is this intentional?): RFC 3627 (ref. '15') (Obsoleted by RFC 6547) -- Obsolete informational reference (is this intentional?): RFC 3633 (ref. '16') (Obsoleted by RFC 8415) -- Obsolete informational reference (is this intentional?): RFC 3736 (ref. '18') (Obsoleted by RFC 8415) -- Obsolete informational reference (is this intentional?): RFC 4214 (ref. '23') (Obsoleted by RFC 5214) == Outdated reference: A later version (-04) exists of draft-ietf-v6ops-scanning-implications-03 Summary: 1 error (**), 0 flaws (~~), 2 warnings (==), 19 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 IPv6 Operations G. Van de Velde 3 Internet-Draft C. Popoviciu 4 Intended status: Informational Cisco Systems 5 Expires: April 3, 2008 T. Chown 6 University of Southampton 7 O. Bonness 8 C. Hahn 9 T-Systems Enterprise Services GmbH 10 October 1, 2007 12 IPv6 Unicast Address Assignment Considerations 13 15 Status of this Memo 17 By submitting this Internet-Draft, each author represents that any 18 applicable patent or other IPR claims of which he or she is aware 19 have been or will be disclosed, and any of which he or she becomes 20 aware will be disclosed, in accordance with Section 6 of BCP 79. 22 Internet-Drafts are working documents of the Internet Engineering 23 Task Force (IETF), its areas, and its working groups. Note that 24 other groups may also distribute working documents as Internet- 25 Drafts. 27 Internet-Drafts are draft documents valid for a maximum of six months 28 and may be updated, replaced, or obsoleted by other documents at any 29 time. It is inappropriate to use Internet-Drafts as reference 30 material or to cite them other than as "work in progress." 32 The list of current Internet-Drafts can be accessed at 33 http://www.ietf.org/ietf/1id-abstracts.txt. 35 The list of Internet-Draft Shadow Directories can be accessed at 36 http://www.ietf.org/shadow.html. 38 This Internet-Draft will expire on April 3, 2008. 40 Copyright Notice 42 Copyright (C) The IETF Trust (2007). 44 Abstract 46 One fundamental aspect of any IP communications infrastructure is its 47 addressing plan. With its new address architecture and allocation 48 policies, the introduction of IPv6 into a network means that network 49 designers and operators need to reconsider their existing approaches 50 to network addressing. Lack of guidelines on handling this aspect of 51 network design could slow down the deployment and integration of 52 IPv6. This document aims to provide the information and 53 recommendations relevant to planning the addressing aspects of IPv6 54 deployments. The document also provides IPv6 addressing case studies 55 for both an enterprise and an ISP network. 57 Table of Contents 59 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4 60 2. Network Level Addressing Design Considerations . . . . . . . . 5 61 2.1. Global Unique Addresses . . . . . . . . . . . . . . . . . 5 62 2.2. Unique Local IPv6 Addresses . . . . . . . . . . . . . . . 6 63 2.3. 6Bone Address Space . . . . . . . . . . . . . . . . . . . 7 64 2.4. Network Level Design Considerations . . . . . . . . . . . 7 65 2.4.1. Sizing the Network Allocation . . . . . . . . . . . . 8 66 2.4.2. Address Space Conservation . . . . . . . . . . . . . . 9 67 3. Subnet Prefix Considerations . . . . . . . . . . . . . . . . . 9 68 3.1. Considerations for subnet prefixes shorter then /64 . . . 9 69 3.2. Considerations for /64 prefixes . . . . . . . . . . . . . 10 70 3.3. Considerations for subnet prefixes longer then /64 . . . . 10 71 3.3.1. Anycast addresses . . . . . . . . . . . . . . . . . . 10 72 3.3.2. Addresses used by Embedded-RP (RFC3956) . . . . . . . 12 73 3.3.3. ISATAP addresses . . . . . . . . . . . . . . . . . . . 12 74 3.3.4. /126 addresses . . . . . . . . . . . . . . . . . . . . 13 75 3.3.5. /127 addresses . . . . . . . . . . . . . . . . . . . . 13 76 3.3.6. /128 addresses . . . . . . . . . . . . . . . . . . . . 13 77 4. Allocation of the IID of an IPv6 Address . . . . . . . . . . . 13 78 4.1. Automatic EUI-64 Format Option . . . . . . . . . . . . . . 14 79 4.2. Using Privacy Extensions . . . . . . . . . . . . . . . . . 14 80 4.3. Manual/Dynamic Assignment Option . . . . . . . . . . . . . 14 81 5. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 14 82 6. Security Considerations . . . . . . . . . . . . . . . . . . . 15 83 7. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 15 84 8. References . . . . . . . . . . . . . . . . . . . . . . . . . . 15 85 8.1. Normative References . . . . . . . . . . . . . . . . . . . 15 86 8.2. Informative References . . . . . . . . . . . . . . . . . . 15 87 Appendix A. Case Studies . . . . . . . . . . . . . . . . . . . . 17 88 A.1. Enterprise Considerations . . . . . . . . . . . . . . . . 17 89 A.1.1. Obtaining general IPv6 network prefixes . . . . . . . 18 90 A.1.2. Forming an address (subnet) allocation plan . . . . . 19 91 A.1.3. Other considerations . . . . . . . . . . . . . . . . . 19 92 A.1.4. Node configuration considerations . . . . . . . . . . 20 93 A.2. Service Provider Considerations . . . . . . . . . . . . . 21 94 A.2.1. Investigation of objective Requirements for an 95 IPv6 addressing schema of a Service Provider . . . . 21 96 A.2.2. Exemplary IPv6 address allocation plan for a 97 Service Provider . . . . . . . . . . . . . . . . . . . 24 98 A.2.3. Additional Remarks . . . . . . . . . . . . . . . . . . 28 99 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 31 100 Intellectual Property and Copyright Statements . . . . . . . . . . 33 102 1. Introduction 104 The Internet Protocol Version 6 (IPv6) Addressing Architecture [26] 105 defines three main types of addresses: unicast, anycast and 106 multicast. This document focuses on unicast addresses, for which 107 there are currently two principal allocated types: Global Unique 108 Addresses [14] ('globals') and Unique Local IPv6 Addresses [22] 109 (ULAs). In addition until recently there has been 'experimental' 110 6bone address space [3], though its use has been deprecated since 111 June 2006 [17]. 113 The document covers aspects that should be considered during IPv6 114 deployment for the design and planning of an addressing scheme for an 115 IPv6 network. The network's IPv6 addressing plan may be for an IPv6- 116 only network, or for a dual-stack infrastructure where some or all 117 devices have addresses in both protocols. These considerations will 118 help an IPv6 network designer to efficiently and prudently assign the 119 IPv6 address space that has been allocated to their organization. 121 The address assignment considerations are analyzed separately for the 122 two major components of the IPv6 unicast addresses, namely 'Network 123 Level Addressing' (the allocation of subnets) and the 'interface-id'. 124 Thus the document includes a discussion of aspects of address 125 assignment to nodes and interfaces in an IPv6 network. Finally the 126 document provides two examples of deployed address plans in a service 127 provider (ISP) and an enterprise network. 129 Parts of this document highlight the differences that an experienced 130 IPv4 network designer should consider when planning an IPv6 131 deployment, for example: 133 o IPv6 devices will more likely be multi-addressed in comparison 134 with their IPv4 counterparts 135 o The practically unlimited size of an IPv6 subnet (2^64 bits) 136 reduces the requirement to size subnets to device counts for the 137 purposes of (IPv4) address conservation 138 o Even though there is no broadcast for the IPv6 protocol, there is 139 still need to consider the number of devices in a given subnet due 140 to traffic storm and level of traffic generated by hosts 141 o The implications of the vastly increased subnet size on the threat 142 of address-based host scanning and other scanning techniques, as 143 discussed in [30]. 145 We do not discuss here how a site or ISP should proceed with 146 acquiring its globally routable IPv6 address prefix. In each case 147 the prefix received is provider assigned (PA) or provider independent 148 (PI). 150 We do not discuss PI policy here. The observations and 151 recommendations of this text are largely independent of the PA or PI 152 nature of the address block being used. At this time we assume that 153 most commonly an IPv6 network which changes provider will need to 154 undergo a renumbering process, as described in [21]. A separate 155 document [32] makes recommendations to ease the IPv6 renumbering 156 process. 158 This document does not discuss implementation aspects related to the 159 transition between the ULA addresses and the now obsoleted site-local 160 addresses. Most implementations know about Site-local addresses even 161 though they are deprecated, and do not know about ULAs - even though 162 they represent current specification. As result transitioning 163 between these types of addresses may cause difficulties. 165 2. Network Level Addressing Design Considerations 167 This section discusses the kind of IPv6 addresses used at the network 168 level for the IPv6 infrastructure. The kind of addresses that can be 169 considered are Global Unique Addresses and ULAs. We also comment 170 here on the deprecated 6bone address space. 172 2.1. Global Unique Addresses 174 The most commonly used unicast addresses will be Global Unique 175 Addresses ('globals'). No significant considerations are necessary 176 if the organization has an address space assignment and a single 177 prefix is deployed through a single upstream provider. 179 However, a multihomed site may deploy addresses from two or more 180 Service Provider assigned IPv6 address ranges. Here, the network 181 Administrator must have awareness on where and how these ranges are 182 used on the multihomed infrastructure environment. The nature of the 183 usage of multiple prefixes may depend on the reason for multihoming 184 (e.g. resilience failover, load balancing, policy-based routing, or 185 multihoming during an IPv6 renumbering event). IPv6 introduces 186 improved support for multi-addressed hosts through the IPv6 default 187 address selection methods described in RFC3484 [12]. A multihomed 188 host may thus have two addresses, one per prefix (provider), and 189 select source and destination addresses to use as described in that 190 RFC. However multihoming also has some operative and administrative 191 burdens besides chosing multiple addresses per interface [33] 192 [25][24]. 194 2.2. Unique Local IPv6 Addresses 196 ULAs have replaced the originally conceived Site Local addresses in 197 the IPv6 addressing architecture, for reasons described in [19]. 198 ULAs improve on site locals by offering a high probability of the 199 global uniqueness of the prefix used, which can be beneficial in the 200 case of (deliberate or accidental) leakage, or where networks are 201 merged. ULAs are akin to the private address space [1] assigned for 202 IPv4 networks, except that in IPv6 networks we may expect to see ULAs 203 used alongside global addresses, with ULAs used internally and 204 globals used externally. Thus use of ULAs does not imply use of NAT 205 for IPv6. 207 The ULA address range allows network administrators to deploy IPv6 208 addresses on their network without asking for a globally unique 209 registered IPv6 address range. A ULA prefix is 48 bits, i.e. a /48, 210 the same as the currently recommended allocation for a site from the 211 globally routable IPv6 address space [9]. 213 A site willing to use ULA address space can have either (a) multiple 214 /48 prefixes (e.g. a /44) and wishes to use ULAs, or (b) has one /48 215 and wishes to use ULAs or (c) a site has a less-than-/48 prefix (e.g. 216 a /56 or /64) and wishes to use ULAs. In all above cases the ULA 217 addresses can be randomly chosen according the principles specified 218 in [19]. Using a random chosen ULA address will be conform in case 219 (a) provide suboptimal aggregation capability, while in case (c) 220 there will be overconsumption of address space. 222 ULAs provide the means to deploy a fixed addressing scheme that is 223 not affected by a change in service provider and the corresponding PA 224 global addresses. Internal operation of the network is thus 225 unaffected during renumbering events. Nevertheless, this type of 226 address must be used with caution. 228 A site using ULAs may or may not also deploy global addresses. In an 229 isolated network ULAs may be deployed on their own. In a connected 230 network, that also deploys global addresses, both may be deployed, 231 such that hosts become multiaddressed (one global and one ULA 232 address) and the IPv6 default address selection algorithm will pick 233 the appropriate source and destination addresses to use, e.g. ULAs 234 will be selected where both the source and destination hosts have ULA 235 addresses. Because a ULA and a global site prefix are both /48 236 length, an administrator can choose to use the same subnetting (and 237 host addressing) plan for both prefixes. 239 As an example of the problems ULAs may cause, when using IPv6 240 multicast within the network, the IPv6 default address selection 241 algorithm prefers the ULA address as the source address for the IPv6 242 multicast streams. This is NOT a valid option when sending an IPv6 243 multicast stream to the IPv6 Internet for two reasons. For one, 244 these addresses are not globally routable so RPF checks for such 245 traffic will fail outside the internal network. The other reason is 246 that the traffic will likely not cross the network boundary due to 247 multicast domain control and perimeter security policies. 249 In principle ULAs allow easier network mergers than RFC1918 addresses 250 do for IPv4 because ULA prefixes have a high probability of 251 uniqueness, if the prefix is chosen as described in the RFC. 253 The usage of ULAs should be carefully considered even when not 254 attached to the IPv6 Internet as some IPv6 specifications were 255 created before the existence of ULA addresses. 257 2.3. 6Bone Address Space 259 The 6Bone address space was used before the RIRs started to 260 distribute 'production' IPv6 prefixes. The 6Bone prefixes have a 261 common first 16 bits in the IPv6 Prefix of 3FFE::/16. This address 262 range is deprecated as of 6th June 2006 [17] and must not be used on 263 any new IPv6 network deployments. Sites using 6bone address space 264 should renumber to production address space using procedures as 265 defined in [21]. 267 2.4. Network Level Design Considerations 269 IPv6 provides network administrators with a significantly larger 270 address space, enabling them to be very creative in how they can 271 define logical and practical address plans. The subnetting of 272 assigned prefixes can be done based on various logical schemes that 273 involve factors such as: 274 o Using existing systems 275 * translate the existing subnet number into IPv6 subnet id 276 * translate the VLAN id into IPv6 subnet id 277 o Rethink 278 * allocate according to your need 279 o Aggregation 280 * Geographical Boundaries - by assigning a common prefix to all 281 subnets within a geographical area 282 * Organizational Boundaries - by assigning a common prefix to an 283 entire organization or group within a corporate infrastructure 284 * Service Type - by reserving certain prefixes for predefined 285 services such as: VoIP, Content Distribution, wireless 286 services, Internet Access, Security areas etc. This type of 287 addressing may create dependencies on IP addresses that can 288 make renumbering harder if the nodes or interfaces supporting 289 those services on the network are sparse within the topology. 291 Such logical addressing plans have the potential to simplify network 292 operations and service offerings, and to simplify network management 293 and troubleshooting. A very large network would also have no need to 294 consider using private address space for its infrastructure devices, 295 simplifying network management. 297 The network designer must however keep in mind several factors when 298 developing these new addressing schemes for networks with and without 299 global connectivity: 300 o Prefix Aggregation - The larger IPv6 addresses can lead to larger 301 routing tables unless network designers are actively pursuing 302 aggregation. While prefix aggregation will be enforced by the 303 service provider, it is beneficial for the individual 304 organizations to observe the same principles in their network 305 design process 306 o Network growth - The allocation mechanism for flexible growth of a 307 network prefix, documented in RFC3531 [13] can be used to allow 308 the network infrastructure to grow and be numbered in a way that 309 is likely to preserve aggregation (the plan leaves 'holes' for 310 growth) 311 o ULA usage in large networks - Networks which have a large number 312 of 'sites' that each deploy a ULA prefix which will by default be 313 a 'random' /48 under fc00::/7 will have no aggregation of those 314 prefixes. Thus the end result may be cumbersome because the 315 network will have large amounts of non-aggregated ULA prefixes. 316 However, there is no rule to disallow large networks to use a 317 single ULA for all 'sites', as a ULA still provides 16 bits for 318 subnetting to be used internally 319 o It is possible that as registry policies evolve, a small site may 320 experience an increase in prefix length when renumbering, e.g. 321 from /48 to /56. For this reason, the best practice is number 322 subnets compactly rather than sparsely, and to use low-order bits 323 as much as possible when numbering subnets. In other words, even 324 if a /48 is allocated, act as though only a /56 is available. 325 Clearly, this advice does not apply to large sites and enterprises 326 that have an intrinsic need for a /48 prefix. 328 2.4.1. Sizing the Network Allocation 330 We do not discuss here how a network designer sizes their application 331 for address space. By default a site will receive a /48 prefix [9] , 332 however different RIR service regions policies may suggest 333 alternative default assignments or let the ISPs to decide on what 334 they believe is more appropriate for their specific case [29]. The 335 default provider allocation via the RIRs is currently a /32 [31]. 336 These allocations are indicators for a first allocation for a 337 network. Different sizes may be obtained based on the anticipated 338 address usage [31]. There are examples of allocations as large as 339 /19 having been made from RIRs to providers at the time of writing. 341 2.4.2. Address Space Conservation 343 Despite the large IPv6 address space which enables easier subnetting, 344 it still is important to ensure an efficient use of this resource. 345 Some addressing schemes, while facilitating aggregation and 346 management, could lead to significant numbers of addresses being 347 unused. Address conservation requirements are less stringent in IPv6 348 but they should still be observed. 350 The proposed HD [10] value for IPv6 is 0.94 compared to the current 351 value of 0.96 for IPv4. Note that for IPv6 HD is calculated for 352 sites (e.g. on a basis of /48), instead of based on addresses like 353 with IPv4. 355 3. Subnet Prefix Considerations 357 This section analyzes the considerations applied to define the subnet 358 prefix of the IPv6 addresses. The boundaries of the subnet prefix 359 allocation are specified in RFC4291 [26]. In this document we 360 analyze their practical implications. Based on RFC4291 [26] it is 361 legal for any IPv6 unicast address starting with binary address '000' 362 to have a subnet prefix larger than, smaller than or of equal to 64 363 bits. Each of these three options is discussed in this document. 365 3.1. Considerations for subnet prefixes shorter then /64 367 An allocation of a prefix shorter then 64 bits to a node or interface 368 is considered bad practice. One exception to this statement is when 369 using 6to4 technology where a /16 prefix is utilised for the pseudo- 370 interface [8]. The shortest subnet prefix that could theoretically 371 be assigned to an interface or node is limited by the size of the 372 network prefix allocated to the organization. 374 A possible reason for choosing the subnet prefix for an interface 375 shorter then /64 is that it would allow more nodes to be attached to 376 that interface compared to a prescribed length of 64 bits. This 377 however is unnecessary for most networks considering that 2^64 378 provides plenty of node addresses. 380 The subnet prefix assignments can be made either by manual 381 configuration, by a stateful Host Configuration Protocol [11], by a 382 stateful prefix delegation mechanism [16] or implied by stateless 383 autoconfiguration from prefix RAs. 385 3.2. Considerations for /64 prefixes 387 Based on RFC3177 [9], 64 bits is the prescribed subnet prefix length 388 to allocate to interfaces and nodes. 390 When using a /64 subnet length, the address assignment for these 391 addresses can be made either by manual configuration, by a stateful 392 Host Configuration Protocol [11] [18] or by stateless 393 autoconfiguration [2]. 395 Note that RFC3177 strongly prescribes 64 bit subnets for general 396 usage, and that stateless autoconfiguration option is only defined 397 for 64 bit subnets. However, implementations could use proprietary 398 mechanism for stateless autoconfiguration for different then 64 bit 399 prefix length. 401 3.3. Considerations for subnet prefixes longer then /64 403 Address space conservation is the main motivation for using a subnet 404 prefix length longer than 64 bits, however this kind of address 405 conservation is of futile benefit compared with the additional 406 considerations one must make when creating and maintain an IPv6 407 address plan. 409 The address assignment can be made either by manual configuration or 410 by a stateful Host Configuration Protocol [11]. 412 When assigning a subnet prefix of more then 80 bits, according to 413 RFC4291 [26] "u" and "g" bits (respectively the 81st and 82nd bit) 414 need to be taken into consideration and should be set correctly. In 415 currently implemented IPv6 protocol stacks, the relevance of the "u" 416 (universal/local) bit and "g" (the individual/group) bit are marginal 417 and typically will not show an issue when configured wrongly, however 418 future implementations may turn out differently. 420 When using subnet lengths longer then 64 bits, it is important to 421 avoid selecting addresses that may have a predefined use and could 422 confuse IPv6 protocol stacks. The alternate usage may not be a 423 simple unicast address in all cases. The following points should be 424 considered when selecting a subnet length longer then 64 bits. 426 3.3.1. Anycast addresses 428 3.3.1.1. Subnet Router Anycast Address 430 RFC4291 [26] provides a definition for the required Subnet Router 431 Anycast Address as follows: 433 | n bits | 128-n bits | 434 +--------------------------------------------+----------------+ 435 | subnet prefix | 00000000000000 | 436 +--------------------------------------------+----------------+ 438 It is recommended to avoid allocating this IPv6 address to an device 439 which expects to have a normal unicast address. No additional 440 dependencies for the subnet prefix while the EUI-64 and IID 441 dependencies will be discussed later in this document. 443 3.3.1.2. Reserved IPv6 Subnet Anycast Addresses 445 RFC2526 [4] stated that within each subnet, the highest 128 interface 446 identifier values are reserved for assignment as subnet anycast 447 addresses. 449 The construction of a reserved subnet anycast address depends on the 450 type of IPv6 addresses used within the subnet, as indicated by the 451 format prefix in the addresses. 453 The first type of Subnet Anycast addresses have been defined as 454 follows for EUI-64 format: 456 | 64 bits | 57 bits | 7 bits | 457 +------------------------------+------------------+------------+ 458 | subnet prefix | 1111110111...111 | anycast ID | 459 +------------------------------+------------------+------------+ 461 The anycast address structure implies that it is important to avoid 462 creating a subnet prefix where the bits 65 to 121 are defined as 463 "1111110111...111" (57 bits in total) so that confusion can be 464 avoided. 466 For other IPv6 address types (that is, with format prefixes other 467 than those listed above), the interface identifier is not in EUI-64 468 format and may be other than 64 bits in length; these reserved subnet 469 anycast addresses for such address types are constructed as follows: 471 | n bits | 121-n bits | 7 bits | 472 +------------------------------+------------------+------------+ 473 | subnet prefix | 1111111...111111 | anycast ID | 474 +------------------------------+------------------+------------+ 475 | interface identifier field | 477 In the case discussed above there is no additional dependency for the 478 subnet prefix with the exception of the EUI-64 and an IID dependency. 479 These will be discussed later in this document. 481 3.3.2. Addresses used by Embedded-RP (RFC3956) 483 Embedded-RP [20] reflects the concept of integrating the Rendezvous 484 Point (RP) IPv6 address into the IPv6 multicast group address. Due 485 to this embedding and the fact that the length of the IPv6 address 486 AND the IPv6 multicast address are 128 bits, it is not possible to 487 have the complete IPv6 address of the multicast RP embedded as such. 489 This resulted in a restriction of 15 possible RP-addresses per prefix 490 that can be used with embedded-RP. The space assigned for the 491 embedded-RP is based on the 4 low order bits, while the remainder of 492 the Interface ID [RIID] is set to all '0'. 494 [IPv6-prefix (64 bits)][60 bits all '0'][RIID] 496 Where: [RIID] = 4 bit. 498 This format implies that when selecting subnet prefixes longer then 499 64, and the bits beyond the 64th one are non-zero, the subnet can not 500 use embedded-RP. 502 In addition it is discouraged to assign a matching embedded-RP IPv6 503 address to a device that is not a real Multicast Rendezvous Point, 504 eventhough it would not generate major problems. 506 3.3.3. ISATAP addresses 508 ISATAP [23] is an experimental automatic tunneling protocol used to 509 provide IPv6 connectivity over an IPv4 campus or enterprise 510 environment. In order to leverage the underlying IPv4 511 infrastructure, the IPv6 addresses are constructed in a special 512 format. 514 An IPv6 ISATAP address has the IPv4 address embedded, based on a 515 predefined structure policy that identifies them as an ISATAP 516 address. 518 [IPv6 Prefix (64 bits)][0000:5EFE][IPv4 address] 520 When using subnet prefix length longer then 64 bits it is good 521 engineering practice that the portion of the IPv6 prefix from bit 65 522 to the end of the host-id does not match with the well-known ISATAP 523 [0000:5EFE] address when assigning an IPv6 address to a non-ISATAP 524 interface. 526 In its actual definition there is no multicast support on ISATAP. 528 3.3.4. /126 addresses 530 The 126 bit subnet prefixes are typically used for point-to-point 531 links similar to a the IPv4 address conservative /30 allocation for 532 point-to-point links. The usage of this subnet address length does 533 not lead to any additional considerations other than the ones 534 discussed earlier in this section, particularly those related to the 535 "u" and "g" bits. 537 3.3.5. /127 addresses 539 The usage of the /127 addresses, the equivalent of IPv4's RFC3021 [5] 540 is not valid and should be strongly discouraged as documented in 541 RFC3627 [15]. 543 3.3.6. /128 addresses 545 The 128 bit address prefix may be used in those situations where we 546 know that one, and only one address is sufficient. Example usage 547 would be the off-link loopback address of a network device. 549 When choosing a 128 bit prefix, it is recommended to take the "u" and 550 "g" bits into consideration and to make sure that there is no overlap 551 with either the following well-known addresses: 552 o Subnet Router Anycast Address 553 o Reserved Subnet Anycast Address 554 o Addresses used by Embedded-RP 555 o ISATAP Addresses 557 4. Allocation of the IID of an IPv6 Address 559 In order to have a complete IPv6 address, an interface must be 560 associated a prefix and an Interface Identifier (IID). Section 3 of 561 this document analyzed the prefix selection considerations. This 562 section discusses the elements that should be considered when 563 assigning the IID portion of the IPv6 address. 565 There are various ways to allocate an IPv6 address to a device or 566 interface. The option with the least amount of caveats for the 567 network administrator is that of EUI-64 [2] based addresses. For the 568 manual or dynamic options, the overlap with well known IPv6 addresses 569 should be avoided. 571 4.1. Automatic EUI-64 Format Option 573 When using this method the network administrator has to allocate a 574 valid 64 bit subnet prefix. The EUI-64 [2] allocation procedure can 575 from that moment onward assign the remaining 64 IID bits in a 576 stateless manner. All the considerations for selecting a valid IID 577 have been incorporated in the EUI-64 methodology. 579 4.2. Using Privacy Extensions 581 The main purpose of IIDs generated based on RFC3041 [6] is to provide 582 privacy to the entity using this address. While there are no 583 particular constraints in the usage of these addresses as defined in 584 [6] there are some implications to be aware of when using privacy 585 addresses as documented in section 4 of RFC3041 [6] 587 4.3. Manual/Dynamic Assignment Option 589 This section discusses those IID allocations that are not implemented 590 through stateless address configuration (Section 4.1). They are 591 applicable regardless of the prefix length used on the link. It is 592 out of scope for this section to discuss the various assignment 593 methods (e.g. manual configuration, DHCPv6, etc). 595 In this situation the actual allocation is done by human intervention 596 and consideration needs to be given to the complete IPv6 address so 597 that it does not result in overlaps with any of the well known IPv6 598 addresses: 599 o Subnet Router Anycast Address 600 o Reserved Subnet Anycast Address 601 o Addresses used by Embedded-RP 602 o ISATAP Addresses 604 When using an address assigned by human intervention it is 605 recommended to choose IPv6 addresses which are not obvious to guess 606 and/or avoid any IPv6 addresses that embed IPv4 addresses used in the 607 current infrastructure. Following these two recommendations will 608 make it more difficult for malicious third parties to guess targets 609 for attack, and thus reduce security threats to a certain extent. 611 5. IANA Considerations 613 There are no extra IANA consideration for this document. 615 6. Security Considerations 617 This IPv6 addressing document does not have any direct impact on 618 Internet infrastructure security. 620 7. Acknowledgements 622 Constructive feedback and contributions have been received from Marla 623 Azinger, Stig Venaas, Pekka Savola, John Spence, Patrick Grossetete, 624 Carlos Garcia Braschi, Brian Carpenter, Mark Smith, Janos Mohacsi, 625 Jim Bound, Fred Templin and Ginny Listman. 627 8. References 629 8.1. Normative References 631 8.2. Informative References 633 [1] Rekhter, Y., Moskowitz, R., Karrenberg, D., Groot, G., and E. 634 Lear, "Address Allocation for Private Internets", BCP 5, 635 RFC 1918, February 1996. 637 [2] Thomson, S. and T. Narten, "IPv6 Stateless Address 638 Autoconfiguration", RFC 2462, December 1998. 640 [3] Hinden, R., Fink, R., and J. Postel, "IPv6 Testing Address 641 Allocation", RFC 2471, December 1998. 643 [4] Johnson, D. and S. Deering, "Reserved IPv6 Subnet Anycast 644 Addresses", RFC 2526, March 1999. 646 [5] Retana, A., White, R., Fuller, V., and D. McPherson, "Using 31- 647 Bit Prefixes on IPv4 Point-to-Point Links", RFC 3021, 648 December 2000. 650 [6] Narten, T. and R. Draves, "Privacy Extensions for Stateless 651 Address Autoconfiguration in IPv6", RFC 3041, January 2001. 653 [7] Durand, A., Fasano, P., Guardini, I., and D. Lento, "IPv6 654 Tunnel Broker", RFC 3053, January 2001. 656 [8] Carpenter, B. and K. Moore, "Connection of IPv6 Domains via 657 IPv4 Clouds", RFC 3056, February 2001. 659 [9] IAB and IESG, "IAB/IESG Recommendations on IPv6 Address 660 Allocations to Sites", RFC 3177, September 2001. 662 [10] Durand, A. and C. Huitema, "The H-Density Ratio for Address 663 Assignment Efficiency An Update on the H ratio", RFC 3194, 664 November 2001. 666 [11] Droms, R., Bound, J., Volz, B., Lemon, T., Perkins, C., and M. 667 Carney, "Dynamic Host Configuration Protocol for IPv6 668 (DHCPv6)", RFC 3315, July 2003. 670 [12] Draves, R., "Default Address Selection for Internet Protocol 671 version 6 (IPv6)", RFC 3484, February 2003. 673 [13] Blanchet, M., "A Flexible Method for Managing the Assignment of 674 Bits of an IPv6 Address Block", RFC 3531, April 2003. 676 [14] Hinden, R., Deering, S., and E. Nordmark, "IPv6 Global Unicast 677 Address Format", RFC 3587, August 2003. 679 [15] Savola, P., "Use of /127 Prefix Length Between Routers 680 Considered Harmful", RFC 3627, September 2003. 682 [16] Troan, O. and R. Droms, "IPv6 Prefix Options for Dynamic Host 683 Configuration Protocol (DHCP) version 6", RFC 3633, 684 December 2003. 686 [17] Fink, R. and R. Hinden, "6bone (IPv6 Testing Address 687 Allocation) Phaseout", RFC 3701, March 2004. 689 [18] Droms, R., "Stateless Dynamic Host Configuration Protocol 690 (DHCP) Service for IPv6", RFC 3736, April 2004. 692 [19] Huitema, C. and B. Carpenter, "Deprecating Site Local 693 Addresses", RFC 3879, September 2004. 695 [20] Savola, P. and B. Haberman, "Embedding the Rendezvous Point 696 (RP) Address in an IPv6 Multicast Address", RFC 3956, 697 November 2004. 699 [21] Baker, F., Lear, E., and R. Droms, "Procedures for Renumbering 700 an IPv6 Network without a Flag Day", RFC 4192, September 2005. 702 [22] Hinden, R. and B. Haberman, "Unique Local IPv6 Unicast 703 Addresses", RFC 4193, October 2005. 705 [23] Templin, F., Gleeson, T., Talwar, M., and D. Thaler, "Intra- 706 Site Automatic Tunnel Addressing Protocol (ISATAP)", RFC 4214, 707 October 2005. 709 [24] Nordmark, E. and T. Li, "Threats Relating to IPv6 Multihoming 710 Solutions", RFC 4218, October 2005. 712 [25] Lear, E., "Things Multihoming in IPv6 (MULTI6) Developers 713 Should Think About", RFC 4219, October 2005. 715 [26] Hinden, R. and S. Deering, "IP Version 6 Addressing 716 Architecture", RFC 4291, February 2006. 718 [27] Chown, T., Venaas, S., and C. Strauf, "Dynamic Host 719 Configuration Protocol (DHCP): IPv4 and IPv6 Dual-Stack 720 Issues", RFC 4477, May 2006. 722 [28] De Clercq, J., Ooms, D., Prevost, S., and F. Le Faucheur, 723 "Connecting IPv6 Islands over IPv4 MPLS Using IPv6 Provider 724 Edge Routers (6PE)", RFC 4798, February 2007. 726 [29] ARIN, "http://www.arin.net/policy/nrpm.html#six54". 728 [30] Chown, T., "IPv6 Implications for TCP/UDP Port Scanning 729 (draft-ietf-v6ops-scanning-implications-03.txt)", March 2007. 731 [31] APNIC, ARIN, RIPE NCC, "IPv6 Address Allocation and Assignment 732 Policy (www.ripe.net/ripe/docs/ipv6policy.html)", January 2003. 734 [32] Chown, T., Thompson, M., Ford, A., and S. Venaas, "Things to 735 think about when Renumbering an IPv6 network 736 (draft-chown-v6ops-renumber-thinkabout-05.txt)", March 2007. 738 [33] "List of Internet-Drafts relevant to the Multi6-WG 739 (http://ops.ietf.org/multi6/draft-list.html )". 741 Appendix A. Case Studies 743 This appendix contains two case studies for IPv6 addressing schemas 744 that have been based on the statements and considerations of this 745 draft. These case studies illustrate how this draft has been used in 746 two specific network scenarios. The case studies may serve as basic 747 considerations for an administrator who designs the IPv6 addressing 748 schema for an enterprise or ISP network, but are not intended to 749 serve as general design proposal for every kind of IPv6 network. All 750 subnet sizes used in this appendix are for practical visualization 751 and do not dictate RIR policy. 753 A.1. Enterprise Considerations 755 In this section we consider a case study of a campus network that is 756 deploying IPv6 in parallel with existing IPv4 protocols in a dual- 757 stack environment. The specific example is the University of 758 Southampton (UK), focusing on a large department within that network. 759 The deployment currently spans around 1,000 hosts and over 1,500 760 users. 762 A.1.1. Obtaining general IPv6 network prefixes 764 In the case of a campus network, the site will typically take its 765 connectivity from its National Research and Education Network (NREN). 766 Southampton connects to JANET, the UK academic network, via its local 767 regional network LeNSE. JANET currently has a /32 allocation from 768 RIPE NCC. The current recommended practice is for sites to receive a 769 /48 allocation, and on this basis Southampton has received such a 770 prefix for its own use. The regional network also uses its own 771 allocation from the NREN provider. 773 No ULA addressing is used on site. The campus is not multihomed 774 (JANET is the sole provider), nor does it expect to change service 775 provider, and thus does not plan to use ULAs for the (perceived) 776 benefit of easing network renumbering. Indeed, the campus has 777 renumbered following the aforementioned renumbering procedure [21] on 778 two occasions, and this has proven adequate (with provisos documented 779 in [32]. We also do not see any need to deploy ULAs for in or out of 780 band network management; there are enough IPv6 prefixes available in 781 the site allocation for the infrastructure. In some cases, use of 782 private IP address space in IPv4 creates problems, so we believe that 783 the availability of ample global IPv6 address space for 784 infrastructure may be a benefit for many sites. 786 No 6bone addressing is used on site any more. We note that since the 787 6bone phaseout of June 2006 [17] most transit ISPs have begun 788 filtering attempted use of such prefixes. 790 Southampton does participate in global and organization scope IPv6 791 multicast networks. Multicast address allocations are not discussed 792 here as they are not in scope for the document. We note that IPv6 793 has advantages for multicast group address allocation. In IPv4 a 794 site needs to use techniques like GLOP to pick a globally unique 795 multicast group to use. This is problematic if the site does not use 796 BGP and have an ASN. In IPv6 unicast-prefix-based IPv6 multicast 797 addresses empower a site to pick a globally unique group address 798 based on its unicast own site or link prefix. Embedded RP is also in 799 use, is seen as a potential advantage for IPv6 and multicast, and has 800 been tested successfully across providers between sites (including 801 paths to/from the US and UK). 803 A.1.2. Forming an address (subnet) allocation plan 805 The campus has a /16 prefix for IPv4 use; in principle 256 subnets of 806 256 addresses. In reality the subnetting is muddier, because of 807 concerns of IPv4 address conservation; subnets are sized to the hosts 808 within them, e.g. a /26 IPv4 prefix is used if a subnet has 35 hosts 809 in it. While this is efficient, it increases management burden when 810 physical deployments change, and IPv4 subnets require resizing (up or 811 down), even with DHCP in use. 813 The /48 IPv6 prefix is considerably larger than the IPv4 allocation 814 already in place at the site. It is loosely equivalent to a 'Class 815 A' IPv4 prefix in that it has 2^16 (over 65,000) subnets, but has an 816 effectively unlimited subnet address size (2^64) compared to 256 in 817 the IPv4 equivalent. The increased subnet size means that /64 IPv6 818 prefixes can be used on all subnets, without any requirement to 819 resize them at a later date. The increased subnet volume allows 820 subnets to be allocated more generously to schools and departments in 821 the campus. While address conservation is still important, it is no 822 longer an impediment on network management. Rather, address (subnet) 823 allocation is more about embracing the available address space and 824 planning for future expansion. 826 In a dual-stack network, we choose to deploy our IP subnets 827 congruently for IPv4 and IPv6. This is because the systems are still 828 in the same administrative domains and the same geography. We do not 829 expect to have IPv6-only subnets in production use for a while yet, 830 outside our test beds and our early Mobile IPv6 trials. With 831 congruent addressing, our firewall policies are also aligned for IPv4 832 and IPv6 traffic at our site border. 834 The subnet allocation plan required a division of the address space 835 per school or department. Here a /56 was allocated to the school 836 level of the university; there are around 30 schools currently. A 837 /56 of IPv6 address space equates to 256 /64 size subnet allocations. 838 Further /56 allocations were made for central IT infrastructure, for 839 the network infrastructure and the server side systems. 841 A.1.3. Other considerations 843 The network uses a Demilitarized Zone (DMZ) topology for some level 844 of protection of 'public' systems. Again, this topology is congruent 845 with the IPv4 network. 847 There are no specific transition methods deployed internally to the 848 campus; everything is using the conventional dual-stack approach. 849 There is no use of ISATAP [23] for example. 851 For the Mobile IPv6 early trials, we have allocated one prefix for 852 Home Agent (HA) use. We have not yet considered in detail how Mobile 853 IPv6 usage may grow, and whether more or even every subnet will 854 require HA support. 856 The university operates a tunnel broker [7] service on behalf of 857 UKERNA for JANET sites. This uses separate address space from JANET, 858 not our university site allocation. 860 A.1.4. Node configuration considerations 862 We currently use stateless autoconfiguration on most subnets for IPv6 863 hosts. There is no DHCPv6 service deployed yet, beyond tests of 864 early code releases. We plan to deploy DHCPv6 for address assignment 865 when robust client and server code is available (at the time of 866 writing the potential for this looks good, e.g. via the ISC 867 implementation). We also are seeking a common integrated DHCP/DNS 868 management platform, even if the servers themselves are not co- 869 located, including integrated DHCPv4 and DHCPv6 server configuration, 870 as discussed in [27]. Currently we add client statelessly 871 autoconfigured addresses to the DNS manually, though dynamic DNS is 872 an option. Our administrators would prefer the use of DHCP because 873 they believe it gives them more management control. 875 Regarding the implications of the larger IPv6 subnet address space on 876 scanning attacks [30], we note that all our hosts are dual-stack, and 877 thus are potentially exposed over both protocols anyway. We publish 878 all addresses in DNS, and do not operate a two faced DNS. 880 We have internal usage of RFC3041 privacy addresses [6] currently 881 (certain platforms currently ship with it on by default), but may 882 wish to administratively disable this (perhaps via DHCP) to ease 883 management complexity. However, we need to determine the feasibility 884 of this on all systems, e.g. for guests on wireless LAN or other 885 user-maintained systems. Network management and monitoring should be 886 simpler without RFC3041 in operation, in terms of identifying which 887 physical hosts are using which addresses. We note that RFC3041 is 888 only an issue for outbound connections, and that there is potential 889 to assign privacy addresses via DHCPv6. 891 We manually configure server addresses to avoid address changes on a 892 change of network adaptor. With IPv6 you can choose to pick ::53 for 893 a DNS server, or can pick 'random' addresses for obfuscation, though 894 that's not an issue for publicly advertised addresses (dns, mx, web, 895 etc). 897 A.2. Service Provider Considerations 899 In this section an IPv6 addressing schema is sketched that could 900 serve as an example for an Internet Service Provider. 902 Sub-section A.2.1 starts with some thoughts regarding objective 903 requirements of such an addressing schema and derives a few general 904 thumb rules that have to be kept in mind when designing an ISP IPv6 905 addressing plan. 907 Sub-section A.2.2 illustrates these findings of A.2.1 with an 908 exemplary IPv6 addressing schema for an MPLS-based ISP offering 909 Internet Services as well as Network Access services to several 910 millions of customers. 912 A.2.1. Investigation of objective Requirements for an IPv6 addressing 913 schema of a Service Provider 915 The first step of the IPv6 addressing plan design for a Service 916 provider should identify all technical, operational, political and 917 business requirements that have to be satisfied by the services 918 supported by this addressing schema. 920 According to the different technical constraints and business models 921 as well as the different weights of these requirements (from the 922 point of view of the corresponding Service Provider) it is very 923 likely that different addressing schemas will be developed and 924 deployed by different ISPs. Nevertheless the addressing schema of 925 sub-section A.2.2 is one possible example. 927 For this document it is assumed that our exemplary ISP has to fulfill 928 several roles for its customers as there are: 930 o Local Internet Registry 931 o Network Access Provider 932 o Internet Service Provider 934 A.2.1.1. Requirements for an IPv6 addressing schema from the LIR 935 perspective of the Service Provider 937 In their role as LIR the Service Providers have to care about the 938 policy constraints of the RIRs and the standards of the IETF 939 regarding IPv6 addressing. In this context, the following basic 940 requirements and recommendations have to be considered and should be 941 satisfied by the IPv6 address allocation plan of a Service Provider: 942 o As recommended in RFC 3177 [9] and in several RIR policies 943 "Common" customers sites (normally private customers) should 944 receive a /48 prefix from the aggregate of the Service Provider. 946 (Note: The addressing plan must be flexible enough and take into 947 account the possible change of the minimum allocation size for end 948 users currently under definition by the RIRs.) 949 o "Big customers" (like big enterprises, governmental agencies etc.) 950 may receive shorter prefixes according to their needs when this 951 need could be documented and justified to the RIR. 952 o The IPv6 address allocation schema has to be able to meet the HD- 953 ratio that is proposed for IPv6. This requirement corresponds to 954 the demand for an efficient usage of the IPv6 address aggregate by 955 the Service Provider. (Note: The currently valid IPv6 HD-ratio of 956 0.94 means an effective usage of about 31% of a /20 prefix of the 957 Service Provider on the basis of /48 assignments.) 958 o All assignments to customers have to be documented and stored into 959 a database that can also be queried by the RIR. 960 o The LIR has to make available means for supporting the reverse DNS 961 mapping of the customer prefixes. 963 A.2.1.2. IPv6 addressing schema requirements from the ISP perspective 964 of the Service Provider 966 From ISP perspective the following basic requirements could be 967 identified: 968 o The IPv6 address allocation schema must be able to realize a 969 maximal aggregation of all IPv6 address delegations to customers 970 into the address aggregate of the Service Provider. Only this 971 provider aggregate will be routed and injected into the global 972 routing table (DFZ). This strong aggregation keeps the routing 973 tables of the DFZ small and eases filtering and access control 974 very much. 975 o The IPv6 addressing schema of the SP should contain optimal 976 flexibility since the infrastructure of the SP will change over 977 the time with new customers, transport technologies and business 978 cases. The requirement of optimal flexibility is contrary to the 979 requirements of strong IPv6 address aggregation and efficient 980 address usage, but at this point each SP has to decide which of 981 these requirements to prioritize. 982 o Keeping the multilevel network hierarchy of an ISP in mind, due to 983 addressing efficiency reasons not all hierarchy levels can and 984 should be mapped into the IPv6 addressing schema of an ISP. 985 Sometimes it is much better to implement a more "flat" addressing 986 for the ISP network than to loose big chunks of the IPv6 address 987 aggregate in addressing each level of network hierarchy. (Note: 988 In special cases it is even recommendable for really "small" ISPs 989 to design and implement a totally flat IPv6 addressing schema 990 without any level of hierarchy.) 991 o Besides that a decoupling of provider network addressing and 992 customer addressing is recommended. (Note: A strong aggregation 993 e.g. on POP, aggregation router or Label Edge Router (LER) level 994 limits the numbers of customer routes that are visible within the 995 ISP network but brings also down the efficiency of the IPv6 996 addressing schema. That's why each ISP has to decide how many 997 internal aggregation levels it wants to deploy.) 999 A.2.1.3. IPv6 addressing schema requirements from the Network Access 1000 provider perspective of the Service Provider 1002 As already done for the LIR and the ISP roles of the SP it is also 1003 necessary to identify requirements that come from its Network Access 1004 Provider role. Some of the basic requirements are: 1005 o The IPv6 addressing schema of the SP must be chosen in a way that 1006 it can handle new requirements that are triggered from customer 1007 side. This can be for instance the growing needs of the customers 1008 regarding IPv6 addresses as well as customer driven modifications 1009 within the access network topology (e.g. when the customer moves 1010 from one point of network attachment (POP) to another). (See 1011 section A.2.3.4 "Changing Point of Network Attachment".) 1012 o For each IPv6 address assignment to customers a "buffer zone" 1013 should be reserved that allows the customer to grow in its 1014 addressing range without renumbering or assignment of additional 1015 prefixes. 1016 o The IPv6 addressing schema of the SP must deal with multiple- 1017 attachments of a single customer to the SP network infrastructure 1018 (i.e. multi-homed network access with the same SP). 1020 These few requirements are only part of all the requirements a 1021 Service Provider has to investigate and keep in mind during the 1022 definition phase of its addressing architecture. Each SP will most 1023 likely add more constraints to this list. 1025 A.2.1.4. A few thumb rules for designing an IPv6 ISP addressing 1026 architecture 1028 As outcome of the above enumeration of requirements regarding an ISP 1029 IPv6 addressing plan the following design "thumb rules" have been 1030 derived: 1031 o No "One size fits all". Each ISP must develop its own IPv6 1032 address allocation schema depending on its concrete business 1033 needs. It is not practicable to design one addressing plan that 1034 fits for all kinds of ISPs (Small / big, Routed / MPLS-based, 1035 access / transit, LIR / No-LIR, etc.). 1036 o The levels of IPv6 address aggregation within the ISP addressing 1037 schema should strongly correspond to the implemented network 1038 structure and their number should be minimized because of 1039 efficiency reasons. It is assumed that the SPs own infrastructure 1040 will be addressed in a fairly flat way whereas the part of the 1041 customer addressing architecture should contain several levels of 1042 aggregation. 1043 o Keep the number of IPv6 customer routes inside your network as 1044 small as necessary. A totally flat customer IPv6 addressing 1045 architecture without any intermediate aggregation level will lead 1046 to lots of customer routes inside the SP network. A fair trade- 1047 off between address aggregation levels (and hence the size of the 1048 internal routing table of the SP) and address conservation of the 1049 addressing architecture has to be found. 1050 o The ISP IPv6 addressing schema should provide maximal flexibility. 1051 This has to be realized for supporting different sizes of customer 1052 IPv6 address aggregates ("big" customers vs. "small" customers) as 1053 well as to allow future growing rates (e.g. of customer 1054 aggregates) and possible topological or infrastructural changes. 1055 o A limited number of aggregation levels and sizes of customer 1056 aggregates will ease the management of the addressing schema. 1057 This has to be weighed against the previous "thumb rule" - 1058 flexibility. 1060 A.2.2. Exemplary IPv6 address allocation plan for a Service Provider 1062 In this example, the Service Provider is assumed to operate an MPLS 1063 based backbone and implements 6PE [28] to provide IPv6 backbone 1064 transport between the different locations (POPs) of a fully dual- 1065 stacked network access and aggregation area. 1067 Besides that it is assumed that the Service Provider: 1068 o has received a /20 from its RIR 1069 o operates its own LIR 1070 o has to address its own IPv6 infrastructure 1071 o delegates prefixes from this aggregate to its customers 1073 This addressing schema should illustrate how the /20 IPv6 prefix of 1074 the SP can be used to address the SP-own infrastructure and to 1075 delegate IPv6 prefixes to its customers following the above mentioned 1076 requirements and thumb rules as far as possible. 1078 The below figure summarizes the device types in a SP network and the 1079 typical network design of a MPLS-based service provider. The network 1080 hierarchy of the SP has to be taken into account for the design of an 1081 IPv6 addressing schema and defines its basic shape and the various 1082 levels of aggregation. 1084 +------------------------------------------------------------------+ 1085 | LSRs of the MPLS Backbone of the SP | 1086 +------------------------------------------------------------------+ 1087 | | | | | 1088 | | | | | 1089 +-----+ +-----+ +--------+ +--------+ +--------+ 1090 | LER | | LER | | LER-BB | | LER-BB | | LER-BB | 1091 +-----+ +-----+ +--------+ +--------+ +--------+ 1092 | | | | | | / | | | 1093 | | | | | | / | | | 1094 | | | | +------+ +------+ +------+ | | 1095 | | | | |BB-RAR| |BB-RAR| | AG | | | 1096 | | | | +------+ +------+ +------+ | | 1097 | | | | | | | | | | | | 1098 | | | | | | | | | | | | 1099 | | | | | | | | +-----+ +-----+ +-----+ +-----+ 1100 | | | | | | | | | RAR | | RAR | | RAR | | RAR | 1101 | | | | | | | | +-----+ +-----+ +-----+ +-----+ 1102 | | | | | | | | | | | | | | | | 1103 | | | | | | | | | | | | | | | | 1104 +-------------------------------------------------------------------+ 1105 | Customer networks | 1106 +-------------------------------------------------------------------+ 1107 Figure: Exemplary Service Provider Network 1109 LSR ... Label Switch Router 1110 LER ... Label Edge Router 1111 LER-BB ... Broadband Label Edge Router 1112 RAR ... Remote Access Router 1113 BB-RAR ... Broadband Remote Access Router 1114 AG ... Aggregation Router 1116 Basic design decisions for the exemplary Service Provider IPv6 1117 address plan regarding customer prefixes take into consideration: 1118 o The prefixes assigned to all customers behind the same LER (e.g. 1119 LER or LER-BB) are aggregated under one LER prefix. This ensures 1120 that the number of labels that have to be used for 6PE is limited 1121 and hence provides a strong MPLS label conservation. 1122 o The /20 prefix of the SP is separated into 3 different pools that 1123 are used to allocate IPv6 prefixes to the customers of the SP: 1124 * A pool (e.g. /24) for satisfying the addressing needs of really 1125 "big" customers (as defined in A.2.2.1 sub-section A.) that 1126 need IPv6 prefixes larger than /48 (e.g. /32). These customers 1127 are assumed to be connected to several POPs of the access 1128 network, so that this customer prefix will be visible in each 1129 of these POPs. 1131 * A pool (e.g. /24) for the LERs with direct customer connections 1132 (e.g. dedicated line access) and without an additional 1133 aggregation area between the customer and the LER. (These LERs 1134 are mostly connected to a limited number of customers because 1135 of the limited number of interfaces/ports.) 1136 * A larger pool (e.g. 14*/24) for LERs (e.g. LER-BB) that serve 1137 a high number of customers that are normally connected via some 1138 kind of aggregation network (e.g. DSL customers behind a BB- 1139 RAR or Dial-In customers behind a RAR). 1140 * The IPv6 address delegation within each Pool (end customer 1141 delegation or also the aggregates that are dedicated to the 1142 LERs itself) should be chosen with an additional buffer zone of 1143 100% - 300% for future growth. I.e. 1 or 2 additional prefix 1144 bits should be reserved according to the expected future growth 1145 rate of the corresponding customer / the corresponding network 1146 device aggregate. 1148 A.2.2.1. Defining an IPv6 address allocation plan for customers of the 1149 Service Provider 1151 A.2.2.1.1. 'Big' customers 1153 SP's "big" customers receive their prefix from the /24 IPv6 address 1154 aggregate that has been reserved for their "big" customers. A 1155 customer is considered as "big" customer if it has a very complex 1156 network infrastructure and/or huge IPv6 address needs (e.g. because 1157 of very large customer numbers) and/or several uplinks to different 1158 POPs of the SP network. 1160 The assigned IPv6 address prefixes can have a prefix length in the 1161 range 32-48 and for each assignment a 100 or 300% future growing zone 1162 is marked as "reserved" for this customer. This means for instance 1163 that with a delegation of a /34 to a customer the corresponding /32 1164 prefix (which contains this /34) is reserved for the customers future 1165 usage. 1167 The prefixes for the "big" customers can be chosen from the 1168 corresponding "big customer" pool by either using an equidistant 1169 algorithm or using mechanisms similar to the Sparse Allocation 1170 Algorithm (SAA) [31]. 1172 A.2.2.1.2. 'Common' customers 1174 All customers that are not "big" customers are considered as "common" 1175 customers. They represent the majority of customers hence they 1176 receive a /48 out of the IPv6 customer address pool of the LER where 1177 they are directly connected or aggregated. 1179 Again a 100 - 300% future growing IPv6 address range is reserved for 1180 each customer, so that a "common" customer receives a /48 allocation 1181 but has a /47 or /46 reserved. 1183 (Note: If it is obvious that the likelyhood of needing a /47 or /46 1184 in the future is very small for a "common" customer, than no growing 1185 buffer should be reserved for it and only a /48 will be assigned 1186 without any growing buffer.) 1188 In the network access scenarios where the customer is directly 1189 connected to the LER the customer prefix is directly taken out of the 1190 customer IPv6 address aggregate (e.g. /38) of the corresponding LER. 1192 In all other cases (e.g. the customer is attached to a RAR that is 1193 themselves aggregated to an AG or to a LER) at least 2 different 1194 approaches are possible. 1196 1) Mapping of Aggregation Network Hierarchy into Customer IPv6 1197 Addressing Schema. The aggregation network hierarchy could be mapped 1198 into the design of the customer prefix pools of each network level in 1199 order to achieve a maximal aggregation at the LER level as well as at 1200 the intermediate levels. (Example: Customer - /48, RAR - /38, AG - 1201 /32, LER-BB - /30). At each network level an adequate growing zone 1202 should be reserved. (Note: This approach requires of course some 1203 "fine tuning" of the addressing schema based on a very good knowledge 1204 of the Service Provider network topology including actual growing 1205 ranges and rates.) 1207 When the IPv6 customer address pool of a LER (or another device of 1208 the aggregation network - AG or RAR) is exhausted, the related LER 1209 (or AG or RAR) prefix is shortened by 1 or 2 bits (e.g. from /38 to 1210 /37 or /36) so that the originally reserved growing zone can be used 1211 for further IPv6 address allocations to customers. In the case where 1212 this growing zone is exhausted as well a new prefix range from the 1213 corresponding pool of the next higher hierarchy level can be 1214 requested. 1216 2) "Flat" Customer IPv6 Addressing Schema. The other option is to 1217 allocate all the customer prefixes directly out of the customer IPv6 1218 address pool of the LER where the customers are attached and 1219 aggregated and to ignore the intermediate aggregation network 1220 infrastructure. This approach leads of course to a higher amount of 1221 customer routes at LER and aggregation network level but takes a 1222 great amount of complexity out of the addressing schema. 1223 Nevertheless the aggregation of the customer prefixes to one prefix 1224 at LER level is realized as required above. 1226 (Note: The handling of (e.g. technically triggered) changes within 1227 the ISP access network is shortly discussed in section A.2.3.5.) 1229 If the actual observed growing rates show that the reserved growing 1230 zones are not needed than these growing areas can be freed and used 1231 for assignments for prefix pools to other devices at the same level 1232 of the network hierarchy. 1234 A.2.2.2. Defining an IPv6 address allocation plan for the Service 1235 Provider Network Infrastructure 1237 For the IPv6 addressing of SPs own network infrastructure a /32 (or 1238 /40) from the "big" customers address pool can be chosen. 1240 This SP infrastructure prefix is used to code the network 1241 infrastructure of the SP by assigning a /48 to every POP/location and 1242 using for instance a /56 for coding the corresponding router within 1243 this POP. Each SP internal link behind a router interface could be 1244 coded using a /64 prefix. (Note: While it is suggested to choose a 1245 /48 for addressing the POP/location of the SP network it is left to 1246 each SP to decide what prefix length to assign to the routers and 1247 links within this POP.) 1249 The IIDs of the router interfaces may be generated by using EUI-64 or 1250 through plain manual configuration e.g. for coding additional network 1251 or operational information into the IID. 1253 It is assumed that again 100 - 300% growing zones for each level of 1254 network hierarchy and additional prefix bits may be assigned to POPs 1255 and/or routers if needed. 1257 Loopback interfaces of routers may be chosen from the first /64 of 1258 the /56 router prefix (in the example above). 1260 (Note: The /32 prefix that has been chosen for addressing SPs own 1261 IPv6 network infrastructure gives enough place to code additional 1262 functionalities like security levels or private and test 1263 infrastructure although such approaches haven't been considered in 1264 more detail for the above described SP until now.) 1266 Point-to-point links to customers (e.g. PPP links, dedicated line 1267 etc.) may be addressed using /126 prefixes out of the first /64 of 1268 the access routers that could be reserved for this reason. 1270 A.2.3. Additional Remarks 1271 A.2.3.1. ULA 1273 From the actual view point of SP there is no compelling reason why 1274 ULAs should be used from a SP. Look at section 2.2. 1276 ULAs could be used inside the SP network in order to have an 1277 additional "site-local scoped" IPv6 address for SPs own 1278 infrastructure for instance for network management reasons and maybe 1279 also in order to have an addressing schema that couldn't be reached 1280 from outside the SP network. 1282 In the case when ULAs are used it is possible to map the proposed 1283 internal IPv6 addressing of SPs own network infrastructure as 1284 described in A.2.2.2 above directly to the ULA addressing schema by 1285 substituting the /48 POP prefix with a /48 ULA site prefix. 1287 A.2.3.2. Multicast 1289 IPv6 Multicast-related addressing issues are out of the scope of this 1290 document. 1292 A.2.3.3. POP Multi-homing 1294 POP (or better LER) Multi-homing of customers with the same SP can be 1295 realized within the proposed IPv6 addressing schema of the SP by 1296 assigning multiple LER-dependent prefixes to this customer (i.e. 1297 considering each customer location as a single-standing customer) or 1298 by choosing a customer prefix out of the pool of "big" customers. 1299 The second solution has the disadvantage that in every LER where the 1300 customer is attached this prefix will appear inside the IGP routing 1301 table requiring an explicit MPLS label. 1303 (Note: The described negative POP/LER Multi-homing effects to the 1304 addressing architecture in the SP access network are not tackled by 1305 implementing the Shim6 Site Multi-homing approach since this approach 1306 targets only on a mechanism for dealing with multiple prefixes in end 1307 systems -- the SP will nevertheless have unaggregated customer 1308 prefixes in its internal routing tables.) 1310 A.2.3.4. Changing Point of Network Attachement 1312 In the possible case that a customer has to change its point of 1313 network attachment to another POP/LER within the ISP access network 1314 two different approaches can be applied assuming that the customer 1315 uses PA addresses out of the SP aggregate: 1317 1.) The customer has to renumber its network with an adequate 1318 customer prefix out of the aggregate of the corresponding LER/RAR of 1319 its new network attachement. To minimise the administrative burden 1320 for the customer the prefix should be of the same size as the former. 1321 This conserves the IPv6 address aggregation within the SP network 1322 (and the MPLS label space) but adds additional burden to the 1323 customer. Hence this approach will most likely only be chosen in the 1324 case of "small customers" with temporary addressing needs and/or 1325 prefix delegation with address auto-configuration. 1327 2.) The customer does not need to renumber its network and keeps its 1328 address aggregate. 1330 This apporach leads to additional more-specific routing entries 1331 within the IGP routing table of the LER and will hence consume 1332 additional MPLS labels - but it is totally transparent to the 1333 customer. Because this results in additional administrative effort 1334 and will stress the router resources (label space, memory) of the ISP 1335 this solution will only be offered to the most valuable customers of 1336 an ISP (like e.g. "big customers" or "enterprise customers"). 1338 Nevertheless the ISP has again to find a fair trade-off between 1339 customer renumbering and sub-optimal address aggregation (i.e. the 1340 generation of additional more-specific routing entries within the IGP 1341 and the waste of MPLS Label space). 1343 A.2.3.5. Restructuring of SP (access) network and Renumbering 1345 A technically triggered restructuring of the SP (access) network (for 1346 instance because of split of equipment or installation of new 1347 equipment) should not lead to a customer network renumbering. This 1348 challenge should be handled in advance by an intelligent network 1349 design and IPv6 address planing. 1351 In the worst case the customer network renumbering could be avoided 1352 through the implementation of more specific customer routes. (Note: 1353 Since this kind of network restructuring will mostly happen within 1354 the access network (at the level) below the LER, the LER aggregation 1355 level will not be harmed and the more-specific routes will not 1356 consume additional MPLS label space.) 1358 A.2.3.6. Extensions needed for the later IPv6 migration phases 1360 The proposed IPv6 addressing schema for a SP needs some slight 1361 enhancements / modifications for the later phases of IPv6 1362 integration, for instance in the case when the whole MPLS backbone 1363 infrastructure (LDP, IGP etc.) is realized over IPv6 transport and an 1364 IPv6 addressing of the LSRs is needed. Other changes may be 1365 necessary as well but should not be explained at this point. 1367 Authors' Addresses 1369 Gunter Van de Velde 1370 Cisco Systems 1371 De Kleetlaan 6a 1372 Diegem 1831 1373 Belgium 1375 Phone: +32 2704 5473 1376 Email: gunter@cisco.com 1378 Ciprian Popoviciu 1379 Cisco Systems 1380 7025-6 Kit Creek Road 1381 Research Triangle Park, North Carolina PO Box 14987 1382 USA 1384 Phone: +1 919 392-3723 1385 Email: cpopovic@cisco.com 1387 Tim Chown 1388 University of Southampton 1389 Highfield 1390 Southampton, SO17 1BJ 1391 United Kingdom 1393 Phone: +44 23 8059 3257 1394 Email: tjc@ecs.soton.ac.uk 1396 Olaf Bonness 1397 T-Systems Enterprise Services GmbH 1398 Goslarer Ufer 35 1399 Berlin, 10589 1400 Germany 1402 Phone: +49 30 3497 3124 1403 Email: Olaf.Bonness@t-systems.com 1404 Christian Hahn 1405 T-Systems Enterprise Services GmbH 1406 Goslarer Ufer 35 1407 Berlin, 10589 1408 Germany 1410 Phone: +49 30 3497 3164 1411 Email: HahnC@t-systems.com 1413 Full Copyright Statement 1415 Copyright (C) The IETF Trust (2007). 1417 This document is subject to the rights, licenses and restrictions 1418 contained in BCP 78, and except as set forth therein, the authors 1419 retain all their rights. 1421 This document and the information contained herein are provided on an 1422 "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS 1423 OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND 1424 THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS 1425 OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF 1426 THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED 1427 WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. 1429 Intellectual Property 1431 The IETF takes no position regarding the validity or scope of any 1432 Intellectual Property Rights or other rights that might be claimed to 1433 pertain to the implementation or use of the technology described in 1434 this document or the extent to which any license under such rights 1435 might or might not be available; nor does it represent that it has 1436 made any independent effort to identify any such rights. Information 1437 on the procedures with respect to rights in RFC documents can be 1438 found in BCP 78 and BCP 79. 1440 Copies of IPR disclosures made to the IETF Secretariat and any 1441 assurances of licenses to be made available, or the result of an 1442 attempt made to obtain a general license or permission for the use of 1443 such proprietary rights by implementers or users of this 1444 specification can be obtained from the IETF on-line IPR repository at 1445 http://www.ietf.org/ipr. 1447 The IETF invites any interested party to bring to its attention any 1448 copyrights, patents or patent applications, or other proprietary 1449 rights that may cover technology that may be required to implement 1450 this standard. Please address the information to the IETF at 1451 ietf-ipr@ietf.org. 1453 Acknowledgment 1455 Funding for the RFC Editor function is provided by the IETF 1456 Administrative Support Activity (IASA).