idnits 2.17.1 draft-mrw-nat66-13.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 abstract seems to contain references ([RFC2119]), which it shouldn't. Please replace those with straight textual mentions of the documents in question. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == Line 1352 has weird spacing: '...d short inner...' == Line 1354 has weird spacing: '...d short outer...' == Line 1356 has weird spacing: '...d short inner...' == Line 1357 has weird spacing: '...d short datag...' == Line 1358 has weird spacing: '...ed char chec...' == (12 more instances...) -- The document date (April 21, 2011) is 4751 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: Experimental ---------------------------------------------------------------------------- -- Looks like a reference, but probably isn't: '3' on line 1588 -- Looks like a reference, but probably isn't: '1' on line 1587 -- Looks like a reference, but probably isn't: '0' on line 1587 -- Looks like a reference, but probably isn't: '2' on line 1588 -- Looks like a reference, but probably isn't: '4' on line 1588 -- Looks like a reference, but probably isn't: '5' on line 1588 -- Looks like a reference, but probably isn't: '6' on line 1589 -- Looks like a reference, but probably isn't: '7' on line 1589 -- Obsolete informational reference (is this intentional?): RFC 2629 (Obsoleted by RFC 7749) -- Obsolete informational reference (is this intentional?): RFC 3484 (Obsoleted by RFC 6724) -- Obsolete informational reference (is this intentional?): RFC 5245 (Obsoleted by RFC 8445, RFC 8839) -- Obsolete informational reference (is this intentional?): RFC 5389 (Obsoleted by RFC 8489) -- Obsolete informational reference (is this intentional?): RFC 5766 (Obsoleted by RFC 8656) -- Obsolete informational reference (is this intentional?): RFC 5996 (Obsoleted by RFC 7296) Summary: 1 error (**), 0 flaws (~~), 7 warnings (==), 16 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group M. Wasserman 3 Internet-Draft Painless Security 4 Intended status: Experimental F. Baker 5 Expires: October 23, 2011 Cisco Systems 6 April 21, 2011 8 IPv6-to-IPv6 Network Prefix Translation 9 draft-mrw-nat66-13 11 Abstract 13 This document describes a stateless, transport-agnostic IPv6-to-IPv6 14 Network Prefix Translation (NPTv6) function that provides the address 15 independence benefit associated with IPv4-to-IPv4 NAT (NAPT44), and 16 in addition provides a 1:1 relationship between addresses in the 17 "inside" and "outside" prefixes, preserving end to end reachability 18 at the network layer. 20 Requirements Terminology 22 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 23 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 24 document are to be interpreted as described in RFC 2119 [RFC2119]. 26 Status of this Memo 28 This Internet-Draft is submitted in full conformance with the 29 provisions of BCP 78 and BCP 79. 31 Internet-Drafts are working documents of the Internet Engineering 32 Task Force (IETF). Note that other groups may also distribute 33 working documents as Internet-Drafts. The list of current Internet- 34 Drafts is at http://datatracker.ietf.org/drafts/current/. 36 Internet-Drafts are draft documents valid for a maximum of six months 37 and may be updated, replaced, or obsoleted by other documents at any 38 time. It is inappropriate to use Internet-Drafts as reference 39 material or to cite them other than as "work in progress." 41 This Internet-Draft will expire on October 23, 2011. 43 Copyright Notice 45 Copyright (c) 2011 IETF Trust and the persons identified as the 46 document authors. All rights reserved. 48 This document is subject to BCP 78 and the IETF Trust's Legal 49 Provisions Relating to IETF Documents 50 (http://trustee.ietf.org/license-info) in effect on the date of 51 publication of this document. Please review these documents 52 carefully, as they describe your rights and restrictions with respect 53 to this document. Code Components extracted from this document must 54 include Simplified BSD License text as described in Section 4.e of 55 the Trust Legal Provisions and are provided without warranty as 56 described in the Simplified BSD License. 58 Table of Contents 60 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4 61 1.1. What is Address Independence? . . . . . . . . . . . . . . 5 62 1.2. NPTv6 Applicability . . . . . . . . . . . . . . . . . . . 6 63 2. NPTv6 Overview . . . . . . . . . . . . . . . . . . . . . . . . 7 64 2.1. NPTv6: the simplest case . . . . . . . . . . . . . . . . 8 65 2.2. NPTv6 between peer networks . . . . . . . . . . . . . . . 9 66 2.3. NPTv6 redundancy and load-sharing . . . . . . . . . . . . 9 67 2.4. NPTv6 multihoming . . . . . . . . . . . . . . . . . . . . 10 68 2.5. Mapping with No Per-Flow State . . . . . . . . . . . . . 10 69 2.6. Checksum-Neutral Mapping . . . . . . . . . . . . . . . . 11 70 3. NPTv6 Algorithmic Specification . . . . . . . . . . . . . . . 11 71 3.1. NPTv6 configuration calculations . . . . . . . . . . . . 12 72 3.2. NPTv6 translation, internal network to external 73 network . . . . . . . . . . . . . . . . . . . . . . . . . 12 74 3.3. NPTv6 translation, external network to internal 75 network . . . . . . . . . . . . . . . . . . . . . . . . . 13 76 3.4. NPTv6 with a /48 or shorter prefix . . . . . . . . . . . 13 77 3.5. NPTv6 with a /49 or longer prefix . . . . . . . . . . . . 13 78 3.6. /48 Prefix Mapping Example . . . . . . . . . . . . . . . 14 79 3.7. Address Mapping for Longer Prefixes . . . . . . . . . . . 14 80 4. Implications of Network Address Translator Behavioral 81 Requirements . . . . . . . . . . . . . . . . . . . . . . . . . 15 82 4.1. Prefix configuration and generation . . . . . . . . . . . 15 83 4.2. Subnet numbering . . . . . . . . . . . . . . . . . . . . 15 84 4.3. NAT Behavioral Requirements . . . . . . . . . . . . . . . 16 85 5. Implications for Applications . . . . . . . . . . . . . . . . 16 86 5.1. Recommendation for network planners considering use 87 of NPTv6 Translation . . . . . . . . . . . . . . . . . . 18 88 5.2. Recommendations for application writers . . . . . . . . . 18 89 5.3. Recommendation for future work . . . . . . . . . . . . . 18 90 6. A Note on Port Mapping . . . . . . . . . . . . . . . . . . . . 19 91 7. Security Considerations . . . . . . . . . . . . . . . . . . . 19 92 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 20 93 9. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 20 94 10. Change Log . . . . . . . . . . . . . . . . . . . . . . . . . . 20 95 10.1. Changes Between draft-mrw-behave-nat66-00 and -01 . . . . 20 96 10.2. Changes between *behave-nat66-01 and -02 . . . . . . . . 21 97 10.3. Changes between *nat66-00 and *nat66-01 . . . . . . . . . 21 98 10.4. Changes between *nat66-01 and *nat66-02 . . . . . . . . . 21 99 10.5. Changes between *nat66-02 and *nat66-03 . . . . . . . . . 22 100 10.6. Changes between *nat66-03 and *nat66-04 . . . . . . . . . 22 101 10.7. Changes between *nat66-04 and *nat66-05 . . . . . . . . . 22 102 10.8. Changes between *nat66-05 and *nat66-06 . . . . . . . . . 23 103 10.9. Changes between *nat66-06 and *nat66-07 . . . . . . . . . 23 104 10.10. Changes between *nat66-07 and *nat66-08 . . . . . . . . . 23 105 10.11. Changes up to *nat66-10 . . . . . . . . . . . . . . . . . 23 106 10.12. Changes up to *nat66-11 and -12 . . . . . . . . . . . . . 23 107 10.13. Changes for *nat66-13 . . . . . . . . . . . . . . . . . . 23 108 11. References . . . . . . . . . . . . . . . . . . . . . . . . . . 23 109 11.1. Normative References . . . . . . . . . . . . . . . . . . 23 110 11.2. Informative References . . . . . . . . . . . . . . . . . 24 111 Appendix A. Why GSE? . . . . . . . . . . . . . . . . . . . . . . 26 112 Appendix B. Verification code . . . . . . . . . . . . . . . . . . 28 113 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 35 115 1. Introduction 117 This document describes a stateless IPv6-to-IPv6 Network Prefix 118 Translation (NPTv6) function, designed to provide address 119 independence to the edge network. It is transport-agnostic with 120 respect to transports that don't checksum the IP header, such as 121 SCTP, and to transports that use the TCP/UDP/DCCP pseudo-header and 122 checksum [RFC1071]. 124 Note that, for reasons discussed in [RFC2993] and Section 5, the IETF 125 does not generally recommend the use of Network Address Technology. 127 This has several ramifications: 129 o Any security benefit that NAPT44 might offer is not present in 130 NPTv6, necessitating the use of a firewall to obtain those 131 benefits if desired. An example of such a firewall is described 132 in [RFC6092]. 134 o End to end reachability is preserved, although the address used 135 "inside" the edge network differs from the address used "outside" 136 the edge network. This has implications for application referrals 137 and other uses of Internet layer addresses. 139 o If there are multiple identically-configured prefix translators 140 between two networks, there is no need for them to exchange 141 dynamic state, as there is no dynamic state - the algorithmic 142 translation will be identical across each of them. The network 143 can therefore asymmetrically route, load-share, and fail-over 144 among them without issue. 146 o Since translation is 1:1 at the network layer, there is no need to 147 modify port numbers or other transport parameters. 149 o TCP sessions that authenticate peers using the TCP Authentication 150 Option [RFC5925] cannot have their addresses translated, as the 151 addresses are used in the calculation of the Message 152 Authentication Code. This consideration applies in general to any 153 UNilateral Self-Address Fixing (UNSAF) [RFC3424] Protocol, which 154 the IAB recommends against the deployment of in an environment 155 that changes Internet addresses. 157 o Applications using the Internet Key Exchange Protocol Version 2 158 (IKEv2) [RFC5996] should, at least in theory, detect the presence 159 of the translator; while no NAT traversal solution is required, 160 [RFC5996] would require such sessions to use UDP. 162 1.1. What is Address Independence? 164 For the purposes of this document, IPv6 Address Independence consists 165 of the following set of properties: 167 From the perspective of the edge network: 169 * The IPv6 addresses used inside the local network (for 170 interfaces, access lists, and logs) do not need to be 171 renumbered if the global prefix(es) assigned for use by the 172 edge network are changed. 174 * The IPv6 addresses used inside the edge network (for 175 interfaces, access lists, and logs) or within other upstream 176 networks (such as when multihoming) do not need to be 177 renumbered when a site adds, drops, or changes upstream 178 networks. 180 * It is not necessary for an administration to convince an 181 upstream network to route its internal IPv6 prefixes, or for it 182 to advertise prefixes derived from other upstream networks into 183 it. 185 * Unless it wants to optimize routing between multiple upstream 186 networks in the process of multihoming, there is therefore no 187 need for a BGP exchange with the upstream network. 189 From the perspective of the upstream network: 191 * IPv6 addresses used by the edge network are guaranteed to have 192 a provider-allocated prefix, eliminating the need and concern 193 for BCP 38 [RFC2827] ingress filtering and the advertisement of 194 customer-specific prefixes. 196 Thus, address independence has ramifications for the edge network, 197 networks it directly connects with (especially its upstream 198 networks), and for the Internet as a whole. The desire for address 199 independence has been a primary driver for IPv4 NAT deployment in 200 medium to large-sized enterprise networks, including NAT deployments 201 in enterprises that have plenty of IPv4 provider independent address 202 space (from IPv4 "swamp space"). It has also been a driver for edge 203 networks to become members of Regional Internet Registry (RIR) 204 communities, seeking to obtain BGP Autonomous System Numbers and 205 provider independent prefixes, and as a result has been one of the 206 drivers of the explosion of the IPv4 route table. Service providers 207 have stated that the lack of address independence from their 208 customers has been a negative incentive to deployment, due to the 209 impact of customer routing expected in their networks. 211 The Local Network Protection [RFC4864] document discusses a related 212 concept called "Address Autonomy" as a benefit of NAPT44. [RFC4864] 213 indicates that address autonomy can be achieved by the simultaneous 214 use of global addresses on all nodes within a site that need external 215 connectivity, and Unique Local Addresses (ULAs) [RFC4193] for all 216 internal communication. However, this solution fails to meet the 217 requirement for address independence, because if an ISP renumbering 218 event occurs, all of the hosts, routers, DHCP servers, ACLs, 219 firewalls and other internal systems that are configured with global 220 addresses from the ISP will need to be renumbered before global 221 connectivity is fully restored. 223 The use of IPv6 Provider Independent (PI) addresses has also been 224 suggested as a means to fulfill the address independence requirement. 225 However, this solution requires that an enterprise qualify to receive 226 a PI assignment and persuade their ISP to install specific routes for 227 the enterprise's PI addresses. There are a number of practical 228 issues with this approach, especially if there is a desire to route 229 to a number of geographically and topologically diverse set of sites, 230 which can sometimes involve coordinating with several ISPs to route 231 portions of a single PI prefix. These problems have caused numerous 232 enterprises with plenty of IPv4 swamp space to choose to use IPv4 NAT 233 for part, or substantially all, of their internal network instead of 234 using their provider independent address space. 236 1.2. NPTv6 Applicability 238 NPTv6 provides a simple and compelling solution to meet the Address 239 Independence requirement in IPv6. The address independence benefit 240 stems directly from the translation function of the network prefix 241 translator. To avoid as many of the issues associated with NAPT44 as 242 possible, NPTv6 is defined to include a two-way, checksum-neutral, 243 algorithmic translation function, and nothing else. 245 The fact that NPTv6 does not map ports and is checksum-neutral avoids 246 the need for an NPTv6 Translator to re-write transport layer headers. 247 This makes it feasible to deploy new or improved transport layer 248 protocols without upgrading NPTv6 Translators. Similarly, since 249 NPTv6 does not re-write transport layer headers, NPTv6 will not 250 interfere with encryption of the full IP payload in many cases. 252 The default NPTv6 address mapping mechanism is purely algorithmic, so 253 NPTv6 translators do not need to maintain per-node or per-connection 254 state, allowing deployment of more robust and adaptive networks than 255 can be deployed using NAPT44. Since the default NPTv6 mapping can be 256 performed in either direction, it does not interfere with inbound 257 connection establishment, thus allowing internal nodes to participate 258 in direct Peer-to-Peer applications without the application layer 259 overhead one finds in many IPv4 Peer-to-Peer applications. 261 Although NPTv6 compares favorably to NAPT44 in several ways, it does 262 not eliminate all of the architectural problems associated with IPv4 263 NAT, as described in [RFC2993]. NPTv6 involves modifying IP headers 264 in transit, so it is not compatible with security mechanisms, such as 265 the IPsec Authentication Header, that provide integrity protection 266 for the IP header. NPTv6 may interfere with the use of application 267 protocols that transmit IP addresses in the application-specific 268 portion of the IP datagram. These applications currently require 269 application layer gateways (ALGs) to work correctly through NAPT44 270 devices, and similar ALGs may be required for these applications to 271 work through NPTv6 Translators. The use of separate internal and 272 external prefixes creates complexity for DNS deployment, due to the 273 desire for internal nodes to communicate with other internal nodes 274 using internal addresses, while external nodes need to obtain 275 external addresses to communicate with the same nodes. This 276 frequently results in the deployment of "split DNS", which may add 277 complexity to network configuration. 279 The choice of address within the edge network bears consideration. 280 One could use a ULA, which maximizes address independence. That 281 could also be considered a misuse of the ULA; if the expectation is 282 that a ULA prevents access to a system from outside the range of the 283 ULA, NPTv6 overrides that. On the other hand, the administration is 284 aware that it has made that choice, and could if it desired deploy a 285 second ULA for the purpose of privacy; the only prefix that will be 286 translated is one that has an NPTv6 Translator configured to 287 translate to or from it. Also, using any other global scope address 288 format makes one either obtain a PI prefix or be at the mercy of the 289 agency from which it was allocated. 291 There are significant technical impacts associated with the 292 deployment of any prefix translation mechanism, including NPTv6, and 293 we strongly encourage anyone who is considering the implementation or 294 deployment of NPTv6 to read [RFC4864] and [RFC5902], and to carefully 295 consider the alternatives described in that document, some of which 296 may cause fewer problems than NPTv6. 298 2. NPTv6 Overview 300 NPTv6 may be implemented in an IPv6 router to map one IPv6 address 301 prefix to another IPv6 prefix as each IPv6 datagram transits the 302 router. A router that implements an NPTv6 prefix translation 303 function is referred to as an NPTv6 Translator. 305 2.1. NPTv6: the simplest case 307 In its simplest form, an NPTv6 Translator interconnects two network 308 links, one of which is an "internal" network link attached to a leaf 309 network within a single administrative domain, and the other of which 310 is an "external" network with connectivity to the global Internet. 311 All of the hosts on the internal network will use addresses from a 312 single, locally-routed prefix, and those addresses will be translated 313 to/from addresses in a globally-routable prefix as IP datagrams 314 transit the NPTv6 Translator. The lengths of these two prefixes will 315 be functionally the same; if they differ, the longer of the two will 316 limit the ability to use subnets in the shorter. 318 External Network: Prefix = 2001:0DB8:0001:/48 319 -------------------------------------- 320 | 321 | 322 +-------------+ 323 | NPTv6 | 324 | Translator | 325 +-------------+ 326 | 327 | 328 -------------------------------------- 329 Internal Network: Prefix = FD01:0203:0405:/48 331 Figure 1: A simple translator 333 Figure 1 shows an NPTv6 Translator attached to two networks. In this 334 example, the internal network uses IPv6 Unique Local Addresses (ULAs) 335 [RFC4193] to represent the internal IPv6 nodes, and the external 336 network uses globally routable IPv6 addresses to represent the same 337 nodes. 339 When an NPTv6 Translator forwards datagrams in the "outbound" 340 direction, from the internal network to the external network, NPTv6 341 overwrites the IPv6 source prefix (in the IPv6 header) with a 342 corresponding external prefix. When datagrams are forwarded in the 343 "inbound" direction, from the external network to the internal 344 network, the IPv6 destination prefix is overwritten with a 345 corresponding internal prefix. Using the prefixes shown in the 346 diagram above, as an IP datagram passes through the NPTv6 Translator 347 in the outbound direction, the source prefix (FD01:0203:0405:/48) 348 will be overwritten with the external prefix (2001:0DB8:0001:/48). 349 In an inbound datagram, the destination prefix (2001:0DB8:0001:/48) 350 will be overwritten with the internal prefix (FD01:0203:0405:/48). 351 In both cases, it is the local IPv6 prefix that is overwritten; the 352 remote IPv6 prefix remains unchanged. Nodes on the internal network 353 are said to be "behind" the NPTv6 Translator. 355 2.2. NPTv6 between peer networks 357 NPTv6 can also be used between two private networks. In these cases, 358 both networks may use ULA prefixes, with each subnet in one network 359 mapped into a corresponding subnet in the other network, and vice 360 versa. Or, each network may use ULA prefixes for internal 361 addressing, and global unicast addresses on the other network. 363 Internal Prefix = FD01:4444:5555:/48 364 -------------------------------------- 365 V | External Prefix 366 V | 2001:0DB8:6666:/48 367 V +---------+ ^ 368 V | NPTv6 | ^ 369 V | Device | ^ 370 V +---------+ ^ 371 External Prefix | ^ 372 2001:0DB8:0001:/48 | ^ 373 -------------------------------------- 374 Internal Prefix = FD01:0203:0405:/48 376 Figure 2: Flow of Information in Translation 378 2.3. NPTv6 redundancy and load-sharing 380 In some cases, more than one NPTv6 Translator may be attached to a 381 network, as shown in Figure 3. In such cases, NPTv6 Translators are 382 configured with the same internal and external prefixes. Since there 383 is only one translation, even though there are multiple translators, 384 they map only one external address (prefix and IID) to the internal 385 address. 387 External Network: Prefix = 2001:0DB8:0001:/48 388 -------------------------------------- 389 | | 390 | | 391 +-------------+ +-------------+ 392 | NPTv6 | | NPTv6 | 393 | Translator | | Translator | 394 | #1 | | #2 | 395 +-------------+ +-------------+ 396 | | 397 | | 398 -------------------------------------- 399 Internal Network: Prefix = FD01:0203:0405:/48 400 Figure 3: Parallel Translators 402 2.4. NPTv6 multihoming 404 External Network #1: External Network #2: 405 Prefix = 2001:0DB8:0001:/48 Prefix = 2001:0DB8:5555:/48 406 --------------------------- -------------------------- 407 | | 408 | | 409 +-------------+ +-------------+ 410 | NPTv6 | | NPTv6 | 411 | Translator | | Translator | 412 | #1 | | #2 | 413 +-------------+ +-------------+ 414 | | 415 | | 416 -------------------------------------- 417 Internal Network: Prefix = FD01:0203:0405:/48 419 Figure 4: Parallel Translators with different upstream networks 421 When multihoming, NPTv6 Translators are attached to an internal 422 network, as shown in Figure 4, but connected to different external 423 networks. In such cases, NPTv6 Translators are configured with the 424 same internal prefix, but different external prefixes. Since there 425 are multiple translations, they map multiple external addresses 426 (prefix and IID) to the common internal address. A system within the 427 edge network is unable to determine which external address it is 428 using apart from services such as STUN [RFC5389]. 430 Multihoming in this sense has one negative feature as compared with 431 multihoming with a provider independent address; when routes change 432 between NPTv6 Translators, since the upstream network changes, the 433 translated prefix can change. This would cause sessions and 434 referrals dependent on it to fail as well. This is not expected to 435 be a major issue, however, in networks where routing is generally 436 stable. 438 2.5. Mapping with No Per-Flow State 440 When NPTv6 is used as described in this document, no per-node or per- 441 flow state is maintained in the NPTv6 Translator. Both inbound and 442 outbound datagrams are translated algorithmically, using only 443 information found in the IPv6 header. Due to this property, NPTv6's 444 two-way, algorithmic address mapping can support both outbound and 445 inbound connection establishment without the need for state-priming 446 or rendezvous mechanisms, or the maintenance of mapping state. This 447 is a significant improvement over NAPT44 devices, but it also has 448 significant security implications which are described in Section 7. 450 2.6. Checksum-Neutral Mapping 452 When a change is made to one of the IP header fields in the IPv6 453 pseudo-header checksum (such as one of the IP addresses), the 454 checksum field in the transport layer header may become invalid. 455 Fortunately, an incremental change in the area covered by the 456 Internet standard checksum [RFC1071] will result in a well-defined 457 change to the checksum value [RFC1624]. So, a checksum change caused 458 by modifying part of the area covered by the checksum can be 459 corrected by making a complementary change to a different 16-bit 460 field covered by the same checksum. 462 The NPTv6 mapping mechanisms described in this document are checksum- 463 neutral, which means that they result in IP headers that will 464 generate the same IPv6 pseudo-header checksum when the checksum is 465 calculated using the standard Internet checksum algorithm [RFC1071]. 466 Any changes that are made during translation of the IPv6 prefix are 467 offset by changes to other parts of the IPv6 address. This results 468 in transport layers that use the Internet checksum (such as TCP and 469 UDP) calculating the same IPv6 pseudo header checksum for both the 470 internal and external forms of the same datagram, which avoids the 471 need for the NPTv6 Translator to modify those transport layer headers 472 to correct the checksum value. 474 The outgoing checksum correction is achieved by making a change to a 475 16 bit section of the source address that is not used for routing in 476 the external network. Due to the nature of checksum arithmetic, when 477 the corresponding correction is applied to the same bits of 478 destination address of the inbound packet, the DA is returned to the 479 correct internal value. 481 As noted in Section 4.2, this mapping results in an edge network 482 using a /48 external prefix to be unable to use subnet 0xFFFF. 484 3. NPTv6 Algorithmic Specification 486 The [RFC4291] IPv6 Address is reproduced for clarity in Figure 5. 488 0 15 16 31 32 47 48 63 64 79 80 95 96 111 112 127 489 +-------+-------+-------+-------+-------+-------+-------+-------+ 490 | Routing Prefix | Subnet| Interface Identifier (IID) | 491 +-------+-------+-------+-------+-------+-------+-------+-------+ 493 Figure 5: Enumeration of the IPv6 Address [RFC4291] 495 3.1. NPTv6 configuration calculations 497 When an NPTv6 Translation function is configured, it is configured 498 with 500 o one or more "internal" interfaces with their "internal" routing 501 domain prefixes, and 503 o one or more "external" interfaces with their "external" routing 504 domain prefixes. 506 In the simple case, there is one of each. If a single router 507 provides NPTv6 translation services between a multiplicity of domains 508 (as might be true when multihoming), each internal/external pair must 509 be thought of as a separate NPTv6 Translator from the perspective of 510 this specification. 512 When an NPTv6 Translator is configured, the translation function 513 first ensures that the internal and external prefixes are the same 514 length, if necessary by extending the shorter of the two with zeroes. 515 These two prefixes will be used in the prefix translation function 516 described in Section 3.2 and Section 3.3. 518 They are then zero-extended to /64, for the purposes of a 519 calculation. The translation function calculates the ones-complement 520 sum of the 16 bit words of the /64 external prefix and the /64 521 internal prefix. It then calculates the difference between these 522 values: internal minus external. This value, called the 523 "adjustment", is effectively constant for the lifetime of the NPTv6 524 Translator configuration, and used in per-datagram processing. 526 3.2. NPTv6 translation, internal network to external network 528 When a datagram passes through the NPTv6 Translator from an internal 529 to an external network, its IPv6 Source Address is changed in two 530 ways: 532 o If the internal subnet number has no mapping, such as being 0xFFFF 533 or simply not mapped, discard the datagram. This SHOULD result in 534 an ICMP Destination Unreachable. 536 o The internal prefix is overwritten with the external prefix, in 537 effect subtracting the difference between the two checksums (the 538 adjustment) from the pseudo-header's checksum, and 540 o A 16-bit word of the address has the adjustment added to it using 541 one's complement arithmetic. If the result is 0xFFFF, it is 542 overwritten as zero. The choice of word is as specified in 543 Section 3.4 or Section 3.5 as appropriate. 545 3.3. NPTv6 translation, external network to internal network 547 When a datagram passes through the NPTv6 Translator from an external 548 to an internal network, its IPv6 Destination Address is changed in 549 two ways: 551 o The external prefix is overwritten with the internal prefix, in 552 effect adding the difference between the two checksums (the 553 adjustment) to the pseudoheader's checksum, and 555 o A 16-bit word of the address has the adjustment subtracted from it 556 (bitwise inverted and added to it) it using one's complement 557 arithmetic. If the result is 0xFFFF, it is overwritten as zero. 558 The choice of word is as specified in Section 3.4 or Section 3.5 559 as appropriate. 561 3.4. NPTv6 with a /48 or shorter prefix 563 When an NPTv6 Translator is configured with internal and external 564 prefixes that are 48 bits in length (a /48) or shorter, the 565 adjustment MUST be added to or subtracted from bits 48..63 of the 566 address. 568 This mapping results in no modification of the Interface Identifier 569 (IID), which is held in the lower half of the IPv6 address, so it 570 will not interfere with future protocols that may use unique IIDs for 571 node identification. 573 NPTv6 Translator implementations MUST implement the /48 mapping. 575 3.5. NPTv6 with a /49 or longer prefix 577 When an NPTv6 Translator is configured with internal and external 578 prefixes that are longer than 48 bits in length (such as a /52, /56, 579 or /60), the adjustment must be added to or subtracted from one of 580 the words in bits 64..79, 80..95, 96..111, or 112..127 of the 581 address. While the choice of word is immaterial as long as it is 582 consistent, for consistency's sake, these words MUST be inspected in 583 that sequence, and the first that is not initially 0xFFFF chosen. 585 NPTv6 Translator implementations SHOULD implement the mapping for 586 longer prefixes. 588 3.6. /48 Prefix Mapping Example 590 For the network shown in Figure 1, the Internal Prefix is FD01:0203: 591 0405:/48, and the External Prefix is 2001:0DB8:0001:/48. 593 If a node with internal address FD01:0203:0405:0001::1234 sends an 594 outbound datagram through the NPTv6 Translator, the resulting 595 external address will be 2001:0DB8:0001:D550::1234. The resulting 596 address is obtained by calculating the checksum of both the internal 597 and external 48-bit prefixes, subtracting the internal prefix from 598 the external prefix using one's complement arithmetic to calculate 599 the "adjustment", and adding the adjustment to the 16-bit subnet 600 field (in this case 0x0001). 602 To show the work: 604 The one's complement checksum of FD01:0203:0405 is 0xFCF5. The one's 605 complement checksum of 2001:0DB8:0001 is 0xD245. Using one's 606 complement arithmetic, 0xD245 - 0xFCF5 = 0xD54F. The subnet in the 607 original datagram is 0x0001. Using one's complement arithmetic, 608 0x0001 + 0xD54F = 0xD550. Since 0xD550 != 0xFFFF, it is not changed 609 to 0x0000. 611 So, the value 0xD550 is written in the 16-bit subnet area, resulting 612 in a mapped external address of 2001:0DB8:0001:D550::1234. 614 When a response datagram is received, it will contain the destination 615 address 2001:0DB8:0001:D550::0001, which will be mapped using the 616 inverse mapping algorithm, back to FD01:0203:0405:0001::1234. 618 In this case, the difference between the two prefixes will be 619 calculated as follows: 621 Using one's complement arithmetic, 0xFCF5 - 0xD245 = 0x2AB0. The 622 subnet in the original datagram = 0xD550. Using one's complement 623 arithmetic, 0xD550 + 0x2AB0 = 0x0001. Since 0x0001 != 0xFFFF, it is 624 not changed to 0x0000. 626 So the value 0x0001 is written into the subnet field, and the 627 internal value of the subnet field is properly restored. 629 3.7. Address Mapping for Longer Prefixes 631 If the prefix being mapped is longer than 48 bits, the algorithm is 632 slightly more complex. A common case will be that the internal and 633 external prefixes are of different length. In such a case, the 634 shorter prefix is zero-extended to the length of the longer as 635 described in Section 3.1 for the purposes of overwriting the prefix. 637 Then, they are both zero-extended to 64 bits to facilitate one's 638 complement arithmetic. The "adjustment" is calculated using those 64 639 bit prefixes. 641 For example if the internal prefix is a /48 ULA and the external 642 prefix is a /56 provider-allocated prefix, the ULA becomes a /56 with 643 zeros in bits 48..55. For purposes of one's complement arithmetic, 644 they are then both zero-extended to 64 bits. A side-effect of this 645 is that a subset of the subnets possible in the shorter prefix are 646 untranslatable. While the security value of this is debatable, the 647 administration may choose to use them for subnets that it knows need 648 no external accessibility. 650 We then find the first word in the IID that does not have the value 651 0xFFFF, trying bits 64..79, and then 80..95, 96..111, and finally 652 112..127. We perform the same calculation (with the same proof of 653 correctness) as in Section 3.6, but applying it to that word. 655 Although any 16-bit portion of an IPv6 IID could contain 0xFFFF, an 656 IID of all-ones is a reserved anycast identifier that should not be 657 used on the network [RFC2526]. If an NPTv6 Translator discovers a 658 datagram with an IID of all-zeros while performing address mapping, 659 that datagram MUST be dropped, and an ICMPv6 Parameter Problem error 660 SHOULD be generated [RFC4443]. 662 Note: this mechanism does involve modification of the IID; it may not 663 be compatible with future mechanisms that use unique IIDs for node 664 identification. 666 4. Implications of Network Address Translator Behavioral Requirements 668 4.1. Prefix configuration and generation 670 NPTv6 Translators MUST support manual configuration of internal and 671 external prefixes, and MUST NOT place any restrictions on those 672 prefixes except that they be valid IPv6 unicast prefixes as described 673 in [RFC4291]. They MAY also support random generation of ULA 674 addresses on command. Since the most common place anticipated for 675 the implementation of an NPTv6 Translator is a CPE router, the reader 676 is urged to consider the requirements of 677 [I-D.ietf-v6ops-ipv6-cpe-router]. 679 4.2. Subnet numbering 681 For reasons detailed in Appendix B, a network using NPTv6 Translation 682 and a /48 external prefix MUST NOT use the value 0xFFFF to designate 683 a subnet that it expects to be translated. 685 4.3. NAT Behavioral Requirements 687 NPTv6 Translators MUST support hairpinning behavior, as defined in 688 the NAT Behavioral Requirements for UDP document [RFC4787]. This 689 means that when an NPTv6 Translator receives a datagram on the 690 internal interface that has a destination address that matches the 691 site's external prefix, it will translate the datagram and forward it 692 internally. This allows internal nodes to reach other internal nodes 693 using their external, global addresses when necessary. 695 Conceptually, the datagram leaves the domain (is translated as 696 described in Section 3.2), and returns (is again translated as 697 described in Section 3.3). As a result, the datagram exchange will 698 be through the NPTv6 Translator in both directions for the lifetime 699 of the session. The alternative would be to require the NPTv6 700 Translator to drop the datagram, forcing the sender to use the 701 correct internal prefix for its peer. Performing only the external- 702 to-internal translation results in the datagram being sent from the 703 untranslated internal address of the source to the translated and 704 therefore internal address of its peer, which would enable the 705 session to bypass the NPTv6 Translator for future datagrams. It 706 would also mean that the original sender would be unlikely to 707 recognize the response when it arrived. 709 Because NPTv6 does not perform port mapping and uses a one-to-one, 710 reversible mapping algorithm, none of the other NAT behavioral 711 requirements apply to NPTv6. 713 5. Implications for Applications 715 NPTv6 Translation does not create several of the problems known to 716 exist with other kinds of NATs and discussed in [RFC2993]. In 717 particular: NPTv6 Translation is stateless, so a "reset" or brief 718 outage of an NPTv6 Translator does not break connections that 719 traverse the translation function, and if multiple NPTv6 Translators 720 exist between the same two networks, load can shift or be dynamically 721 load-shared among them. Also, an NPTv6 Translator does not aggregate 722 traffic for several hosts/interfaces behind a lesser number of 723 external addresses, so there is no inherent expectation for an NPTv6 724 Translator to block new inbound flows from external hosts, and no 725 issue with a filter or blacklist associated with one prefix within 726 the domain affecting another. A firewall can of course be used in 727 conjunction with NPTv6 Translator; this would allow the network 728 administrator more flexibility to specify security policy than would 729 be possible with a traditional NAT. 731 However, NPTv6 Translation does create difficulties for some kinds of 732 applications. Some examples include: 734 o An application instance "behind" an NPTv6 Translator will see a 735 different address for its connections than its peers "outside" the 736 NPTv6 Translator. 738 o An application instance "outside" an NPTv6 Translator will see a 739 different address for its connections than any peer "inside" an 740 NPTv6 Translator. 742 o An application instance wishing to establish communication with a 743 peer "behind" an NPTv6 Translator may need to use a different 744 address to reach that peer depending on whether the instance is 745 behind the same NPTv6 Translator or external to it. Since an 746 NPTv6 Translator implements hairpinning (Section 4.3), it suffices 747 for applications to always use their external addresses. However, 748 this creates inefficiencies in the local network and may also 749 complicate implementation of the NPTv6 Translator. [RFC3484] also 750 would prefer the private address in such a case in order to reduce 751 those inefficiencies. 753 o An application instance which moves from a realm "behind" an NPTv6 754 Translator to a realm that is "outside" the network, or vice 755 versa, may find that it is no longer able to reach its peers at 756 the same addresses it was previously able to use. 758 o An application instance which is intermittently communicating with 759 a peer that moves from behind an NPTv6 Translator to "outside" of 760 it, or vice versa, may find that it is no longer able to reach 761 that peer at the same address that it had previously used. 763 Many, but not all, of the applications which are adversely affected 764 by NPTv6 Translation are those that do "referrals" - where an 765 application instance passes its own addresses, and/or addresses of 766 its peers, to other peers. (Some believe referrals are inherently 767 undesirable; others believe that they are necessary in some 768 circumstances. A discussion of the merits of referrals, or lack 769 thereof, is beyond the scope of this document.) 771 To some extent, the incidence of these difficulties can be reduced by 772 DNS hacks that attempt to expose addresses "behind" an NPTv6 773 Translator only to hosts which are also behind the same NPTv6 774 Translator; and perhaps also, to expose only the "internal" addresses 775 of hosts behind the NPTv6 Translator to other hosts behind the same 776 NPTv6 Translator. However, this cannot be a complete solution. A 777 full discussion of these issues is out of scope for this document, 778 but briefly: (a) reliance on DNS to solve this problem depends on 779 hosts always making queries from DNS servers in the same realm as 780 they are (or on DNS interception proxies, which create their own 781 problems), and on mobile hosts/applications not caching those 782 results; (b) reliance on DNS to solve this problem depends on network 783 administrators on all networks using such applications to reliably 784 and accurately maintain current DNS entries for every host using 785 those applications; and (c) reliance on DNS to solve this problem 786 depends on applications always using DNS names, even though they 787 often must run in environments where DNS names are not reliably 788 maintained for every host. Other issues are that there is often no 789 single distinguished name for a host, no reliable way for a host to 790 determine what DNS names are associated with it, and which names are 791 appropriate to use in which contexts. 793 5.1. Recommendation for network planners considering use of NPTv6 794 Translation 796 In light of the above, network planners considering the use of NPTv6 797 translation should carefully consider the kinds of applications that 798 they will need to run in the future, and determine whether the 799 address stability and provider independence benefits are consistent 800 with their application requirements. 802 5.2. Recommendations for application writers 804 Several mechanisms (e.g. STUN [RFC5389], TURN [RFC5766], ICE 805 [RFC5245]) have been used with traditional IPv4 NAT to circumvent 806 some of the limitations of such devices. Similar mechanisms could 807 also be applied to circumvent some of the issues with NPTv6 808 Translator. However, all of these require the assistance of an 809 external server or a function co-located with the translator that can 810 tell an "internal" host what its "external" addresses are. 812 5.3. Recommendation for future work 814 It might be desirable to define a general mechanism which would allow 815 hosts within a translation domain to determine their external 816 addresses and/or request that inbound traffic be permitted. If such 817 a mechanism were to be defined, it would ideally be general enough to 818 also accommodate other types of NAT likely to be encountered by IPV6 819 applications - in particular, IPv4/IPv6 Translation 820 [I-D.ietf-behave-v6v4-framework] [I-D.ietf-behave-dns64] 821 [I-D.ietf-behave-v6v4-xlate] [I-D.ietf-behave-v6v4-xlate-stateful] 822 [RFC6052]. For this and other reasons, such a mechanism is beyond 823 the scope of this document. 825 6. A Note on Port Mapping 827 In addition to overwriting IP addresses when datagrams are forwarded, 828 NAPT44 devices overwrite the source port number in outbound traffic, 829 and the destination port number in inbound traffic. This mechanism 830 is called "port mapping". 832 The major benefit of port mapping is that it allows multiple 833 computers to share a single IPv4 address. A large number of internal 834 IPv4 addresses (typically from one of the [RFC1918] private address 835 spaces) can be mapped into a single external, globally routable IPv4 836 address, with the local port number used to identify which internal 837 node should receive each inbound datagram. This address 838 amplification feature is not generally foreseen as a necessity at 839 this time. 841 Since port mapping requires re-writing a portion of the transport 842 layer header, it requires NAPT44 devices to be aware of all of the 843 transport protocols that they forward, thus stifling the development 844 of new and improved transport protocols and preventing the use of 845 IPsec encryption. Modifying the transport layer header is 846 incompatible with security mechanisms that encrypt the full IP 847 payload, and restricts the NAPT44 to forwarding transport layers that 848 use weak checksum algorithms that are easily recalculated in routers. 850 Since there is significant detriment caused by modifying transport 851 layer headers and very little, if any, benefit to the use of port 852 mapping in IPv6, NPTv6 Translators that comply with this 853 specification MUST NOT perform port mapping. 855 7. Security Considerations 857 When NPTv6 is deployed using either of the two-way, algorithmic 858 mappings defined in the document, it allows direct inbound 859 connections to internal nodes. While this can be viewed as a benefit 860 of NPTv6 vs. NAPT44, it does open internal nodes to attacks that 861 would be more difficult in a NAPT44 network. Although this situation 862 is not substantially worse, from a security standpoint, than running 863 IPv6 with no NAT, some enterprises may assume that an NPTv6 864 Translator will offer similar protection to a NAPT44 device. 866 The port mapping mechanism in NAPT44 implementations requires that 867 state be created in both directions. This has lead to an industry- 868 wide perception that NAT functionality is the same as a stateful 869 firewall. It is not. The translation function of the NAT only 870 creates dynamic state in one direction and has no policy. For this 871 reason, it is RECOMMENDED that NPTv6 Translators also implement 872 firewall functionality such as described in [RFC6092], with 873 appropriate configuration options including turning it on or off. 875 When [RFC4864] talks about randomizing the subnet identifier, the 876 idea is to make it harder for worms to guess a valid subnet 877 identifier at an advertised network prefix. This should not be 878 interpreted as endorsing concealing the subnet identifier behind the 879 obfuscating function of a translator such as NPTv6. [RFC4864] 880 specifically talks about how to obtain the desired properties of 881 concealment without using a translator. Topology hiding when using 882 NAT is often ineffective in environments where the topology is 883 visible in application layer messaging protocols such as DNS, SIP, 884 SMTP, etc. If the information were not available through the 885 application layer, [RFC2993] would not be valid. 887 Due to the potential interactions with IKEv2/IPsec NAT traversal, it 888 would be valuable to test interactions of NPTv6 with various aspects 889 of current-day IKEv2/IPsec NAT traversal. 891 8. IANA Considerations 893 This document has no IANA considerations. 895 9. Acknowledgements 897 The checksum-neutral algorithmic address mapping described in this 898 document is based on e-mail written by Iljtsch van Beijnum. 900 The following people provided advice or review comments that 901 substantially improved this document: Allison Mankin, Christian 902 Huitema, Dave Thaler, Ed Jankiewicz, Eric Kline, Iljtsch van Beijnum, 903 Jari Arkko, Keith Moore, Mark Townsley, Merike Kaeo, Ralph Droms, 904 Remi Despres, Steve Blake, and Tony Hain. 906 This document was written using the xml2rfc tool described in RFC 907 2629 [RFC2629]. 909 10. Change Log 911 This section should be removed by the RFC Editor. 913 10.1. Changes Between draft-mrw-behave-nat66-00 and -01 915 There were several minor changes made between the *behave-nat66-00 916 and -01 versions of this draft: 918 o Added Fred Baker as a co-author. 920 o Minor arithmetic corrections. 922 o Added AH to paragraph on NAT security issues. 924 o Added additional NAT topologies to overview (diagrams TBD). 926 10.2. Changes between *behave-nat66-01 and -02 928 There were further changes made between *behave-nat66-01 and -02: 930 o Removed topology hiding mechanism. 932 o Added diagrams. 934 o Made minor updates based on mailing list feedback. 936 o Added discussion of IPv6 SAF document. 938 o Added applicability section. 940 o Added discussion of Address Independence requirement. 942 o Added hairpinning requirement and discussion of applicability of 943 other NAT behavioral requirements. 945 10.3. Changes between *nat66-00 and *nat66-01 947 There were further changes made between nat66-01 and nat66-02: 949 o Added mapping for prefixes longer than /48. 951 o Change draft name to remove reference to the behave WG. 953 o Resolved various open issues and fixed typos. 955 10.4. Changes between *nat66-01 and *nat66-02 957 o Change the acronym "NAT66" to "NPTv6", so people don't read "NAT" 958 and MEGO. 960 o Change the term used to refer to the function from "NAT66 device" 961 to "NPTv6 Translator". It's not a "device" function, it's a 962 function that is applied between two interfaces. Consider a 963 router with two upstreams and two legs in the local network; it 964 will not translate between the local legs, but will translate to 965 and from each upstream, and be configured differently for each of 966 the two ISPs. 968 o Comment specifically on the security aspects. 970 o Comment specifically on the application issues raised on this 971 list. 973 o Comment specifically on multihoming, load-sharing, and asymmetric 974 routing. 976 o Spell out the hairpinning requirement and its implications. 978 o Spell out the service provider side of Address Independence. 980 o 00 focuses on the edge's view 982 o Detail the algorithm in a manner clearer to the implementor (I 983 think) 985 o Spell out the case for GSE-style DMZs between the edge and the 986 transit network, which is about the implications for the global 987 routing table. 989 o Refer to [RFC6092] as a CPE firewall description. 991 10.5. Changes between *nat66-02 and *nat66-03 993 o Added an appendix on Verification code 995 o Various minor markups in response to Ralph Droms 997 10.6. Changes between *nat66-03 and *nat66-04 999 o Markups in response to Christian Huitema, mostly surrounding the 1000 issue of subnet 0xFFFF. 1002 o Refer to [I-D.ietf-v6ops-ipv6-cpe-router] for CPE router 1003 requirements. 1005 10.7. Changes between *nat66-04 and *nat66-05 1007 o Update statistics in appendix A per BGP report of 17 December 2010 1009 o Update security considerations using text supplied by Merike Kaeo. 1011 10.8. Changes between *nat66-05 and *nat66-06 1013 o restore a code snippet inadvertently removed in version -05 1015 10.9. Changes between *nat66-06 and *nat66-07 1017 o Changed requested status to experimental 1019 o Incorporated comments from Eric Kline 1021 10.10. Changes between *nat66-07 and *nat66-08 1023 The section on Application Considerations was expanded after 1024 discussion with Keith Moore. 1026 10.11. Changes up to *nat66-10 1028 Address review comments during IETF Last Call and the Transport 1029 Directorate Review. 1031 10.12. Changes up to *nat66-11 and -12 1033 Address Dave Thaler's comments, mostly editorial, bit also addressing 1034 UNSAF protocols like the TCP Authentication Option. 1036 10.13. Changes for *nat66-13 1038 o Inserted a sentence to make Jari happy. 1040 o Inserted a paragraph suggested by Stewart Bryant. 1042 o normalized the terms "packet" and "datagram", for consistency. 1044 11. References 1046 11.1. Normative References 1048 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1049 Requirement Levels", BCP 14, RFC 2119, March 1997. 1051 [RFC2526] Johnson, D. and S. Deering, "Reserved IPv6 Subnet Anycast 1052 Addresses", RFC 2526, March 1999. 1054 [RFC4193] Hinden, R. and B. Haberman, "Unique Local IPv6 Unicast 1055 Addresses", RFC 4193, October 2005. 1057 [RFC4291] Hinden, R. and S. Deering, "IP Version 6 Addressing 1058 Architecture", RFC 4291, February 2006. 1060 [RFC4443] Conta, A., Deering, S., and M. Gupta, "Internet Control 1061 Message Protocol (ICMPv6) for the Internet Protocol 1062 Version 6 (IPv6) Specification", RFC 4443, March 2006. 1064 [RFC4787] Audet, F. and C. Jennings, "Network Address Translation 1065 (NAT) Behavioral Requirements for Unicast UDP", BCP 127, 1066 RFC 4787, January 2007. 1068 11.2. Informative References 1070 [GSE] O'Dell, M., "GSE - An Alternate Addressing Architecture 1071 for IPv6", February 1997, 1072 . 1074 [I-D.ietf-behave-dns64] 1075 Bagnulo, M., Sullivan, A., Matthews, P., and I. Beijnum, 1076 "DNS64: DNS extensions for Network Address Translation 1077 from IPv6 Clients to IPv4 Servers", 1078 draft-ietf-behave-dns64-11 (work in progress), 1079 October 2010. 1081 [I-D.ietf-behave-v6v4-framework] 1082 Baker, F., Li, X., Bao, C., and K. Yin, "Framework for 1083 IPv4/IPv6 Translation", 1084 draft-ietf-behave-v6v4-framework-10 (work in progress), 1085 August 2010. 1087 [I-D.ietf-behave-v6v4-xlate] 1088 Li, X., Bao, C., and F. Baker, "IP/ICMP Translation 1089 Algorithm", draft-ietf-behave-v6v4-xlate-23 (work in 1090 progress), September 2010. 1092 [I-D.ietf-behave-v6v4-xlate-stateful] 1093 Bagnulo, M., Matthews, P., and I. Beijnum, "Stateful 1094 NAT64: Network Address and Protocol Translation from IPv6 1095 Clients to IPv4 Servers", 1096 draft-ietf-behave-v6v4-xlate-stateful-12 (work in 1097 progress), July 2010. 1099 [I-D.ietf-v6ops-ipv6-cpe-router] 1100 Singh, H., Beebee, W., Donley, C., Stark, B., and O. 1101 Troan, "Basic Requirements for IPv6 Customer Edge 1102 Routers", draft-ietf-v6ops-ipv6-cpe-router-09 (work in 1103 progress), December 2010. 1105 [NIST] NIST, "Draft NIST Framework and Roadmap for Smart Grid 1106 Interoperability, Release 1.0", September 2009. 1108 [RFC1071] Braden, R., Borman, D., Partridge, C., and W. Plummer, 1109 "Computing the Internet checksum", RFC 1071, 1110 September 1988. 1112 [RFC1624] Rijsinghani, A., "Computation of the Internet Checksum via 1113 Incremental Update", RFC 1624, May 1994. 1115 [RFC1918] Rekhter, Y., Moskowitz, R., Karrenberg, D., Groot, G., and 1116 E. Lear, "Address Allocation for Private Internets", 1117 BCP 5, RFC 1918, February 1996. 1119 [RFC2629] Rose, M., "Writing I-Ds and RFCs using XML", RFC 2629, 1120 June 1999. 1122 [RFC2827] Ferguson, P. and D. Senie, "Network Ingress Filtering: 1123 Defeating Denial of Service Attacks which employ IP Source 1124 Address Spoofing", BCP 38, RFC 2827, May 2000. 1126 [RFC2993] Hain, T., "Architectural Implications of NAT", RFC 2993, 1127 November 2000. 1129 [RFC3424] Daigle, L. and IAB, "IAB Considerations for UNilateral 1130 Self-Address Fixing (UNSAF) Across Network Address 1131 Translation", RFC 3424, November 2002. 1133 [RFC3484] Draves, R., "Default Address Selection for Internet 1134 Protocol version 6 (IPv6)", RFC 3484, February 2003. 1136 [RFC4864] Van de Velde, G., Hain, T., Droms, R., Carpenter, B., and 1137 E. Klein, "Local Network Protection for IPv6", RFC 4864, 1138 May 2007. 1140 [RFC5245] Rosenberg, J., "Interactive Connectivity Establishment 1141 (ICE): A Protocol for Network Address Translator (NAT) 1142 Traversal for Offer/Answer Protocols", RFC 5245, 1143 April 2010. 1145 [RFC5389] Rosenberg, J., Mahy, R., Matthews, P., and D. Wing, 1146 "Session Traversal Utilities for NAT (STUN)", RFC 5389, 1147 October 2008. 1149 [RFC5766] Mahy, R., Matthews, P., and J. Rosenberg, "Traversal Using 1150 Relays around NAT (TURN): Relay Extensions to Session 1151 Traversal Utilities for NAT (STUN)", RFC 5766, April 2010. 1153 [RFC5902] Thaler, D., Zhang, L., and G. Lebovitz, "IAB Thoughts on 1154 IPv6 Network Address Translation", RFC 5902, July 2010. 1156 [RFC5925] Touch, J., Mankin, A., and R. Bonica, "The TCP 1157 Authentication Option", RFC 5925, June 2010. 1159 [RFC5996] Kaufman, C., Hoffman, P., Nir, Y., and P. Eronen, 1160 "Internet Key Exchange Protocol Version 2 (IKEv2)", 1161 RFC 5996, September 2010. 1163 [RFC6052] Bao, C., Huitema, C., Bagnulo, M., Boucadair, M., and X. 1164 Li, "IPv6 Addressing of IPv4/IPv6 Translators", RFC 6052, 1165 October 2010. 1167 [RFC6092] Woodyatt, J., "Recommended Simple Security Capabilities in 1168 Customer Premises Equipment (CPE) for Providing 1169 Residential IPv6 Internet Service", RFC 6092, 1170 January 2011. 1172 Appendix A. Why GSE? 1174 For the purpose of this discussion, let us over-simplify the 1175 Internet's structure by distinguishing between two broad classes of 1176 networks: transit and edge. A "transit network", in this context, is 1177 a network that provides connectivity services to other networks. Its 1178 AS number may show up in a non-final position in BGP AS paths, or in 1179 the case of mobile and residential broadband networks, it may offer 1180 network services to smaller networks that can't justify RIR 1181 membership. An "edge network", in contrast, is any network that is 1182 not a transit network; it is the ultimate customer, and while it 1183 provides internal connectivity for its own use, it is in other 1184 respects a consumer of transit services. In terms of routing, a 1185 network in the transit domain generally needs some way to make 1186 choices about how it routes to other networks; an edge network is 1187 generally quite satisfied with a simple default route. 1189 The [GSE] proposal, and as a result this proposal (which is similar 1190 to GSE in most respects and inspired by it), responds directly to 1191 current concerns in the RIR communities. Edge networks are used to 1192 an environment in IPv4 in which their addressing is disjoint from 1193 that of their upstream transit networks; it is either provider 1194 independent, or a network prefix translator makes their external 1195 address distinct from their internal address, and they like the 1196 distinction. In IPv6, there is a mantra that edge network addresses 1197 should be derived from their upstream, and if they have multiple 1198 upstreams, edge networks are expected to design their networks to use 1199 all of those prefixes equivalently. They see this as unnecessary and 1200 unwanted operational complexity, and are as a result pushing very 1201 hard in the RIR communities for provider independent addressing. 1203 Widespread use of provider independent addressing has a natural and 1204 perhaps unavoidable side-effect that is likely to be very expensive 1205 in the long term. It means that the routing table will enumerate the 1206 networks at the edge of the transit domain, the edge networks, rather 1207 than enumerating the transit domain. Per the BGP Update Report of 17 1208 December 2010, there are currently over 36,000 Autonomous Systems 1209 being advertised in BGP, of which over 15,000 advertise only one 1210 prefix. There are in the neighborhood of 5000 AS's that show up in a 1211 non-final position in AS paths, and perhaps another 5000 networks 1212 whose AS numbers are terminal in more than one AS path. In other 1213 words, we have prefixes for some 36,000 transit and edge networks in 1214 the route table now, many of which arguably need an Autonomous System 1215 number only for multihoming. Current estimates suggest that we could 1216 easily see that be on the order of 10,000,000 within fifteen years. 1217 However, the vast majority of networks (2/3) having the tools 1218 necessary to multihome are not visibly doing so, and would be well 1219 served by any solution that gives them address independence without 1220 the overhead of RIR membership and BGP routing. 1222 Current growth estimates suggest that we could easily see that be on 1223 the order of 10,000,000 within fifteen years. Tens of thousands of 1224 entries in the route table is very survivable; while our protocols 1225 and computers will likely do quite well with tens of millions of 1226 routes, the heat produced and power consumed by those routers, and 1227 the inevitable impact on the cost of those routers, is not a good 1228 outcome. To avoid having a massive and unscalable route table, we 1229 need to find a way that is politically acceptable and returns us to 1230 enumerating the transit domain, not the edge. 1232 There have been a number of proposals. As described, shim6 moves the 1233 complexity to the edge, and the edge is rebelling. Geographic 1234 addressing in essence forces ISPs to "own" geographic territory from 1235 a routing perspective, as otherwise there is no clue in the address 1236 as to what network a datagram should be delivered to in order to 1237 reach it. Metropolitan Addressing can imply regulatory authority, 1238 and even if it is implemented using internet exchange consortia, 1239 visits a great deal of complexity on the transit networks that 1240 directly serve the edge. The one that is likely to be most 1241 acceptable is any proposal that enables an edge network to be 1242 operationally independent of its upstreams, with no obligation to 1243 renumber when it adds, drops, or changes ISPs, and with no additional 1244 burden placed either on the ISP or the edge network as a result. 1245 From an application perspective, an additional operational 1246 requirement in the words of Roadmap for the Smart Grid [NIST], is 1247 that 1248 "...the Network should enable an application in a particular 1249 domain to communicate with an application in any other domain in 1250 the information network, with proper management control over who 1251 and where applications can be interconnected." 1253 In other words, the structure of the network should allow for and 1254 enable appropriate access control, but the structure of the network 1255 should not inherently limit access. 1257 The GSE model, by statelessly translating the prefix between an edge 1258 network and its upstream transit network, accomplishes that with a 1259 minimum of fuss and bother. Stated in the simplest terms, it enables 1260 the edge network to behave as if it has a provider independent prefix 1261 from a multihoming and renumbering perspective without the overhead 1262 of RIR membership or maintaining BGP connectivity, and it enables the 1263 transit networks to aggressively aggregate what are from their 1264 perspective provider-allocated customer prefixes, to maintain a 1265 rational-sized routing table. 1267 Appendix B. Verification code 1269 This non-normative appendix is presented as a proof of concept. It 1270 is in no sense optimized; for example, one's complement arithmetic is 1271 implemented in portable subroutines, where operational 1272 implementations might use one's complement arithmetic instructions 1273 through a pragma; such implementations probably need to explicitly 1274 force 0xFFFF to 0x0000, as the instruction will not. The original 1275 purpose of the code was to verify whether or not it was necessary to 1276 suppress 0xFFFF by overwriting with zero, and whether predicted 1277 issues with subnet numbering were real. 1279 The point is to 1281 o demonstrate that if one or the other representation of zero is not 1282 used in the word the checksum is updated in, the program maps 1283 inner and outer addresses in a manner that is, mathematically, 1:1 1284 and onto (each inner address maps to a unique outer address, and 1285 that outer address maps back to exactly the same inner address), 1286 and 1288 o give guidance on the suppression of 0xFFFF checksums. 1290 In short, in one's complement arithmetic, x-x=0, but will take the 1291 negative representation of zero. If 0xFFFF results are forced to the 1292 value 0x0000, as is recommended in [RFC1071], the word the checksum 1293 is adjusted in cannot be initially 0xFFFF, as on the return it will 1294 be forced to 0. If 0xFFFF results are not forced to the value 0x0000 1295 as is recommended in [RFC1071], the word the checksum is adjusted in 1296 cannot be initially 0, as on the return it will be calculated as 1297 0+(~0) = 0xFFFF. We chose to follow [RFC1071]'s recommendations, 1298 which implies a requirement to not use 0xFFFF as a subnet number in 1299 networks with a /48 external prefix. 1301 /* 1302 * Copyright (c) 2010 IETF Trust and the persons identified as 1303 * authors of the code. All rights reserved. Redistribution 1304 * and use in source and binary forms, with or without 1305 * modification, are permitted provided that the following 1306 * conditions are met: 1307 * 1308 * o Redistributions of source code must retain the above 1309 * copyright notice, this list of conditions and the 1310 * following disclaimer. 1311 * 1312 * o Redistributions in binary form must reproduce the above 1313 * copyright notice, this list of conditions and the 1314 * following disclaimer in the documentation and/or other 1315 * materials provided with the distribution. 1316 * 1317 * o Neither the name of Internet Society, IETF or IETF Trust, 1318 * nor the names of specific contributors, may be used to 1319 * endorse or promote products derived from this software 1320 * without specific prior written permission. 1321 * 1322 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND 1323 * CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, 1324 * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF 1325 * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE 1326 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR 1327 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 1328 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 1329 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 1330 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 1331 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 1332 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR 1333 * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 1334 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 1335 */ 1336 #include "stdio.h" 1337 #include "assert.h" 1338 /* 1339 * program to verify the NPTv6 algorithm 1340 * 1341 * argument: 1342 * perform negative zero suppression: boolean 1343 * 1344 * method: 1345 * We specify an internal and an external prefix. The prefix 1346 * length is presumed to be the common length of both, and for 1347 * this is a /48. We perform the three algorithms specified. 1348 * the "datagram" address is in effect the source address 1349 * internal->external and the destination address 1350 * external->internal. 1351 */ 1352 unsigned short inner_init[] = { 1353 0xFD01, 0x0203, 0x0405, 1, 2, 3, 4, 5}; 1354 unsigned short outer_init[] = { 1355 0x2001, 0x0db8, 0x0001, 1, 2, 3, 4, 5}; 1356 unsigned short inner[8]; 1357 unsigned short datagram[8]; 1358 unsigned char checksum[65536] = {0}; 1359 unsigned short outer[8]; 1360 unsigned short adjustment; 1361 unsigned short suppress; 1362 /* 1363 * One's complement sum. 1364 * return number1 + number2 1365 */ 1366 unsigned short 1367 add1(number1, number2) 1368 unsigned short number1; 1369 unsigned short number2; 1370 { 1371 unsigned int result; 1373 result = number1; 1374 result += number2; 1375 if (suppress) { 1376 while (0xFFFF <= result) { 1377 result = result + 1 - 0x10000; 1378 } 1379 } else { 1380 while (0xFFFF < result) { 1381 result = result + 1 - 0x10000; 1382 } 1383 } 1384 return result; 1385 } 1387 /* 1388 * One's complement difference 1389 * return number1 - number2 1390 */ 1392 unsigned short 1393 sub1(number1, number2) 1394 unsigned short number1; 1395 unsigned short number2; 1396 { 1397 return add1(number1, ~number2); 1398 } 1400 /* 1401 * return one's complement sum of an array of numbers 1402 */ 1403 unsigned short 1404 sum1(numbers, count) 1405 unsigned short *numbers; 1406 int count; 1407 { 1408 unsigned int result; 1410 result = *numbers++; 1411 while (--count > 0) { 1412 result += *numbers++; 1413 } 1415 if (suppress) { 1416 while (0xFFFF <= result) { 1417 result = result + 1 - 0x10000; 1418 } 1419 } else { 1420 while (0xFFFF < result) { 1421 result = result + 1 - 0x10000; 1422 } 1423 } 1424 return result; 1425 } 1427 /* 1428 * NPTv6 initialization: section 3.1 assuming section 3.4 1429 * 1430 * create the /48, a source address in internal format, and a 1431 * source address in external format. calculate the adjustment 1432 * if one /48 is overwritten with the other. 1433 */ 1434 void 1435 nptv6_initialization(subnet) 1436 unsigned short subnet; 1437 { 1438 int i; 1439 unsigned short inner48; 1440 unsigned short outer48; 1442 /* initialize the internal and external prefixes. */ 1443 for (i = 0; i < 8; i++) { 1444 inner[i] = inner_init[i]; 1445 outer[i] = outer_init[i]; 1446 } 1447 inner[3] = subnet; 1448 outer[3] = subnet; 1449 /* calculate the checksum adjustment */ 1450 inner48 = sum1(inner, 3); 1451 outer48 = sum1(outer, 3); 1452 adjustment = sub1(inner48, outer48); 1453 } 1455 /* 1456 * NPTv6 datagram from edge to transit: section 3.2 assuming 1457 * section 3.4 1458 * 1459 * overwrite the prefix in the source address with the outer 1460 * prefix, and adjust the checksum 1461 */ 1462 void 1463 nptv6_inner_to_outer() 1464 { 1465 int i; 1467 /* let's get the source address into the datagram */ 1468 for (i = 0; i < 8; i++) { 1469 datagram[i] = inner[i]; 1470 } 1472 /* overwrite the prefix with the outer prefix */ 1473 for (i = 0; i < 3; i++) { 1474 datagram[i] = outer[i]; 1475 } 1477 /* adjust the checksum */ 1478 datagram[3] = add1(datagram[3], adjustment); 1479 } 1481 /* 1482 * NPTv6 datagram from transit to edge:: section 3.3 assuming 1483 * section 3.4 1484 * 1485 * overwrite the prefix in the destination address with the 1486 * inner prefix, and adjust the checksum 1487 */ 1489 void 1490 nptv6_outer_to_inner() 1491 { 1492 int i; 1494 /* overwrite the prefix with the outer prefix */ 1495 for (i = 0; i < 3; i++) { 1496 datagram[i] = inner[i]; 1497 } 1499 /* adjust the checksum */ 1500 datagram[3] = sub1(datagram[3], adjustment); 1501 } 1503 /* 1504 * main program 1505 */ 1506 main(argc, argv) 1507 int argc; 1508 char **argv; 1509 { 1510 unsigned subnet; 1511 int i; 1513 if (argc < 2) { 1514 fprintf(stderr, "usage: nptv6 supression\n"); 1515 assert(0); 1516 } 1517 suppress = atoi(argv[1]); 1518 assert(suppress <= 1); 1520 for (subnet = 0; subnet < 0x10000; subnet++) { 1521 /* section 3.1: initialize the system */ 1522 nptv6_initialization(subnet); 1524 /* section 3.2: take a datagram from inside to outside */ 1525 nptv6_inner_to_outer(); 1527 /* the resulting checksum value should be unique */ 1528 if (checksum[subnet]) { 1529 printf("inner->outer duplicated checksum: " 1530 "inner: %x:%x:%x:%x:%x:%x:%x:%x(%x) " 1531 "calculated: %x:%x:%x:%x:%x:%x:%x:%x(%x)\n", 1532 inner[0], inner[1], inner[2], inner[3], 1533 inner[4], inner[5], inner[6], inner[7], 1534 sum1(inner, 8), datagram[0], datagram[1], 1535 datagram[2], datagram[3], datagram[4], 1536 datagram[5], datagram[6], datagram[7], 1537 sum1(datagram, 8)); 1538 } 1540 checksum[subnet] = 1; 1542 /* 1543 * the resulting checksum should be the same as the inner 1544 * address's checksum 1545 */ 1546 if (sum1(datagram, 8) != sum1(inner, 8)) { 1547 printf("inner->outer incorrect: " 1548 "inner: %x:%x:%x:%x:%x:%x:%x:%x(%x) " 1549 "calculated: %x:%x:%x:%x:%x:%x:%x:%x(%x)\n", 1550 inner[0], inner[1], inner[2], inner[3], 1551 inner[4], inner[5], inner[6], inner[7], 1552 sum1(inner, 8), 1553 datagram[0], datagram[1], datagram[2], datagram[3], 1554 datagram[4], datagram[5], datagram[6], datagram[7], 1555 sum1(datagram, 8)); 1556 } 1558 /* section 3.3: take a datagram from outside to inside */ 1559 nptv6_outer_to_inner(); 1561 /* 1562 * the returning datagram should have the same checksum it 1563 * left with 1564 */ 1565 if (sum1(datagram, 8) != sum1(inner, 8)) { 1566 printf("outer->inner checksum incorrect: " 1567 "calculated: %x:%x:%x:%x:%x:%x:%x:%x(%x) " 1568 "inner: %x:%x:%x:%x:%x:%x:%x:%x(%x)\n", 1569 datagram[0], datagram[1], datagram[2], datagram[3], 1570 datagram[4], datagram[5], datagram[6], datagram[7], 1571 sum1(datagram, 8), inner[0], inner[1], inner[2], 1572 inner[3], inner[4], inner[5], inner[6], inner[7], 1573 sum1(inner, 8)); 1574 } 1576 /* 1577 * and every octet should calculate back to the same inner 1578 * value 1579 */ 1580 for (i = 0; i < 8; i++) { 1581 if (inner[i] != datagram[i]) { 1582 printf("outer->inner different: " 1583 "calculated: %x:%x:%x:%x:%x:%x:%x:%x " 1584 "inner: %x:%x:%x:%x:%x:%x:%x:%x\n", 1585 datagram[0], datagram[1], datagram[2], 1586 datagram[3], datagram[4], datagram[5], 1587 datagram[6], datagram[7], inner[0], inner[1], 1588 inner[2], inner[3], inner[4], inner[5], 1589 inner[6], inner[7]); 1590 break; 1591 } 1592 } 1593 } 1594 } 1596 Authors' Addresses 1598 Margaret Wasserman 1599 Painless Security 1600 North Andover, MA 01845 1601 USA 1603 Phone: +1 781 405 7464 1604 Email: mrw@painless-security.com 1605 URI: http://www.painless-security.com 1607 Fred Baker 1608 Cisco Systems 1609 Santa Barbara, California 93117 1610 USA 1612 Phone: +1-408-526-4257 1613 Email: fred@cisco.com