idnits 2.17.1 draft-ietf-6man-why64-05.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 : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (September 16, 2014) is 3503 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- ** Obsolete normative reference: RFC 3315 (Obsoleted by RFC 8415) ** Obsolete normative reference: RFC 4941 (Obsoleted by RFC 8981) ** Obsolete normative reference: RFC 5996 (Obsoleted by RFC 7296) -- No information found for draft-odell-8 - is the name correct? == Outdated reference: A later version (-08) exists of draft-ietf-6man-ipv6-address-generation-privacy-01 == Outdated reference: A later version (-08) exists of draft-ietf-opsec-ipv6-host-scanning-04 == Outdated reference: A later version (-82) exists of draft-templin-aerolink-35 -- Obsolete informational reference (is this intentional?): RFC 2629 (Obsoleted by RFC 7749) Summary: 3 errors (**), 0 flaws (~~), 4 warnings (==), 3 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 6MAN B. Carpenter, Ed. 3 Internet-Draft Univ. of Auckland 4 Intended status: Informational T. Chown 5 Expires: March 20, 2015 Univ. of Southampton 6 F. Gont 7 SI6 Networks / UTN-FRH 8 S. Jiang 9 Huawei Technologies Co., Ltd 10 A. Petrescu 11 CEA, LIST 12 A. Yourtchenko 13 cisco 14 September 16, 2014 16 Analysis of the 64-bit Boundary in IPv6 Addressing 17 draft-ietf-6man-why64-05 19 Abstract 21 The IPv6 unicast addressing format includes a separation between the 22 prefix used to route packets to a subnet and the interface identifier 23 used to specify a given interface connected to that subnet. 24 Currently the interface identifier is defined as 64 bits long for 25 almost every case, leaving 64 bits for the subnet prefix. This 26 document describes the advantages of this fixed boundary and analyses 27 the issues that would be involved in treating it as a variable 28 boundary. 30 Status of This Memo 32 This Internet-Draft is submitted in full conformance with the 33 provisions of BCP 78 and BCP 79. 35 Internet-Drafts are working documents of the Internet Engineering 36 Task Force (IETF). Note that other groups may also distribute 37 working documents as Internet-Drafts. The list of current Internet- 38 Drafts is at http://datatracker.ietf.org/drafts/current/. 40 Internet-Drafts are draft documents valid for a maximum of six months 41 and may be updated, replaced, or obsoleted by other documents at any 42 time. It is inappropriate to use Internet-Drafts as reference 43 material or to cite them other than as "work in progress." 45 This Internet-Draft will expire on March 20, 2015. 47 Copyright Notice 49 Copyright (c) 2014 IETF Trust and the persons identified as the 50 document authors. All rights reserved. 52 This document is subject to BCP 78 and the IETF Trust's Legal 53 Provisions Relating to IETF Documents 54 (http://trustee.ietf.org/license-info) in effect on the date of 55 publication of this document. Please review these documents 56 carefully, as they describe your rights and restrictions with respect 57 to this document. Code Components extracted from this document must 58 include Simplified BSD License text as described in Section 4.e of 59 the Trust Legal Provisions and are provided without warranty as 60 described in the Simplified BSD License. 62 Table of Contents 64 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 65 2. Advantages of a fixed identifier length . . . . . . . . . . . 4 66 3. Arguments for shorter identifier lengths . . . . . . . . . . 5 67 3.1. Insufficient address space delegated . . . . . . . . . . 5 68 3.2. Hierarchical addressing . . . . . . . . . . . . . . . . . 6 69 3.3. Audit requirement . . . . . . . . . . . . . . . . . . . . 6 70 3.4. Concerns over ND cache exhaustion . . . . . . . . . . . . 7 71 4. Effects of varying the interface identifier length . . . . . 7 72 4.1. Interaction with IPv6 specifications . . . . . . . . . . 7 73 4.2. Possible failure modes . . . . . . . . . . . . . . . . . 9 74 4.3. Experimental observations . . . . . . . . . . . . . . . . 11 75 4.3.1. Survey of the processing of Neighbor Discovery 76 options with prefixes other than /64 . . . . . . . . 11 77 4.3.2. Other Observations . . . . . . . . . . . . . . . . . 13 78 4.4. Implementation and deployment issues . . . . . . . . . . 14 79 4.5. Privacy issues . . . . . . . . . . . . . . . . . . . . . 15 80 5. Security Considerations . . . . . . . . . . . . . . . . . . . 16 81 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 17 82 7. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 17 83 8. Change log [RFC Editor: Please remove] . . . . . . . . . . . 17 84 9. References . . . . . . . . . . . . . . . . . . . . . . . . . 17 85 9.1. Normative References . . . . . . . . . . . . . . . . . . 17 86 9.2. Informative References . . . . . . . . . . . . . . . . . 21 87 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 23 89 1. Introduction 91 Rather than simply overcoming the IPv4 address shortage by doubling 92 the address size to 64 bits, IPv6 addresses were originally chosen to 93 be 128 bits long to provide flexibility and new possibilities. In 94 particular, the notion of a well-defined interface identifier was 95 added to the IP addressing model. The IPv6 addressing architecture 96 [RFC4291] specifies that a unicast address is divided into n bits of 97 subnet prefix followed by (128-n) bits of interface identifier (IID). 98 The bits in the IID have no meaning and the entire identifier should 99 be treated as an opaque value [RFC7136]. Also, since IPv6 routing is 100 entirely based on variable length prefixes (also known as variable 101 length subnet masks), there is no basic architectural assumption that 102 n has any particular fixed value. All IPv6 routing protocols support 103 prefixes of any length up to /128. 105 The IID is of basic importance in the IPv6 stateless address 106 autoconfiguration (SLAAC) process [RFC4862]. However, it is 107 important to understand that its length is a parameter in the SLAAC 108 process, and it is determined in a separate link-type specific 109 document (see Section 2 of RFC 4862). The SLAAC protocol does not 110 define its length or assume any particular length. Similarly, DHCPv6 111 [RFC3315] does not include a prefix length in its address assignment. 113 The notion of a /64 boundary in the address was introduced after the 114 initial design of IPv6, following a period when it was expected to be 115 at /80. There were two motivations for setting it at /64. One was 116 the original "8+8" proposal [DRAFT-odell] that eventually led to ILNP 117 [RFC6741], which required a fixed point for the split between local 118 and wide-area parts of the address. The other was the expectation 119 that EUI-64 MAC addresses would become widespread in place of 48-bit 120 addresses, coupled with the plan at that time that auto-configured 121 addresses would normally be based on interface identifiers derived 122 from MAC addresses. 124 As a result, RFC 4291 describes a method of forming interface 125 identifiers from IEEE EUI-64 hardware addresses [IEEE802] and this 126 specifies that such interface identifiers are 64 bits long. Various 127 other methods of forming interface identifiers also specify a length 128 of 64 bits. The addressing architecture, as modified by [RFC7136], 129 states that "For all unicast addresses, except those that start with 130 the binary value 000, Interface IDs are required to be 64 bits long. 131 If derived from an IEEE MAC-layer address, they must be constructed 132 in Modified EUI-64 format." The de facto length of almost all IPv6 133 interface identifiers is therefore 64 bits. The only documented 134 exception is in [RFC6164], which standardises 127-bit prefixes for 135 point-to-point links between routers, among other things to avoid a 136 loop condition known as the ping-pong problem. 138 With that exception, and despite the comments above about the routing 139 architecture and the design of SLAAC, using an IID shorter than 64 140 bits and a subnet prefix longer than 64 bits is outside the current 141 IPv6 specifications, so results may vary. 143 The question is often asked why the subnet prefix boundary is set 144 rigidly at /64. The first purpose of this document is to explain the 145 advantages of the fixed IID length. Its second purpose is to analyse 146 in some detail the effects of hypothetically varying the IID length. 147 The fixed length limits the practical length of a routing prefix to 148 64 bits, whereas architecturally, and from the point of view of 149 routing protocols, it could be any value up to /128, as for host 150 routes. Whatever the length of the IID, the longest match is done on 151 the concatenation of prefix and IID. Here, we mainly discuss the 152 question of a shorter IID, which would allow a longer subnet prefix. 153 The document makes no proposal for a change to the IID length. 155 The following three sections describe in turn the advantages of the 156 fixed length IID, some arguments for shorter lengths, and the 157 expected effects of varying the length. 159 2. Advantages of a fixed identifier length 161 As mentioned in Section 1, the existence of an IID of a given length 162 is a necessary part of IPv6 stateless address autoconfiguration 163 (SLAAC) [RFC4862]. This length is normally the same for all nodes on 164 a given link that is running SLAAC. Even though this length is a 165 parameter for SLAAC, determined separately for the link layer media 166 type of each interface, a globally fixed IID length for all link 167 layer media is the simplest solution, and is consistent with the 168 principles of Internet host configuration described in [RFC5505]. 170 An interface identifier of significant length, clearly separated from 171 the subnet prefix, makes it possible to limit the traceability of a 172 host computer by varying the identifier. This is discussed further 173 in Section 4.5. 175 An interface identifier of significant length guarantees that there 176 are always enough addresses in any subnet to add one or more real or 177 virtual interfaces. There might be other limits, but IP addressing 178 will never get in the way. 180 The addressing architecture [RFC4291] [RFC7136] sets the IID length 181 at 64 bits for all unicast addresses, and therefore for all media 182 supporting SLAAC. An immediate effect of fixing the IID length at 64 183 bits is, of course, that it fixes the subnet prefix length also at 64 184 bits, regardless of the aggregate prefix assigned to the site 185 concerned, which in accordance with [RFC6177] should be /56 or 186 shorter. This situation has various specific advantages: 188 o Everything is the same. Compared to IPv4, there is no more 189 calculating leaf subnet sizes, no more juggling between subnets, 190 and fewer consequent errors. Network design is therefore simpler 191 and much more straightforward. This is of importance for all 192 types of networks - enterprise, campus, small office, or home 193 networks - and for all types of operator, from professional to 194 consumer. 196 o Adding a subnet is easy - just take another /64 from the pool. No 197 estimates, calculations, consideration or judgement is needed. 199 o Router configurations are homogeneous and easier to understand. 201 o Documentation is easier to write and easier to read; training is 202 easier. 204 The remainder of this document describes arguments that have been 205 made against the current fixed IID length and analyses the effects of 206 a possible change. However, the consensus of the IETF is that the 207 benefits of keeping the length fixed at 64 bits, and the practical 208 difficulties of changing it, outweigh the arguments for change. 210 3. Arguments for shorter identifier lengths 212 In this section we describe arguments for scenarios where shorter 213 IIDs, implying prefixes longer than /64, have been used or proposed. 215 3.1. Insufficient address space delegated 217 A site may not be delegated a sufficiently generous prefix from which 218 to allocate a /64 prefix to all of its internal subnets. In this 219 case the site may either determine that it does not have enough 220 address space to number all its network elements and thus, at the 221 very best, be only partially operational, or it may choose to use 222 internal prefixes longer than /64 to allow multiple subnets and the 223 hosts within them to be configured with addresses. 225 In this case, the site might choose, for example, to use a /80 per 226 subnet, in combination with hosts using either manually configured 227 addressing or DHCPv6 [RFC3315]. 229 Scenarios that have been suggested where an insufficient prefix might 230 be delegated include home or small office networks, vehicles, 231 building services and transportation services (road signs, etc.). It 232 should be noted that the homenet architecture text 233 [I-D.ietf-homenet-arch] states that a CPE should consider the lack of 234 sufficient address space to be an error condition, rather than using 235 prefixes longer than /64 internally. 237 Another scenario occasionally suggested is one where the Internet 238 address registries actually begin to run out of IPv6 prefix space, 239 such that operators can no longer assign reasonable prefixes to users 240 in accordance with [RFC6177]. It is sometimes suggested that 241 assigning a prefix such as /48 or /56 to every user site (including 242 the smallest) as recommended by [RFC6177] is wasteful. In fact, the 243 currently released unicast address space, 2000::/3, contains 35 244 trillion /48 prefixes ((2**45 = 35,184,372,088,832), of which only a 245 small fraction have been allocated. Allowing for a conservative 246 estimate of allocation efficiency, i.e., an HD-ratio of 0.94 247 [RFC4692], approximately 5 trillion /48 prefixes can be allocated. 248 Even with a relaxed HD-ratio of 0.89, approximately one trillion /48 249 prefixes can be allocated. Furthermore, with only 2000::/3 currently 250 committed for unicast addressing, we still have approximately 85% of 251 the address space in reserve. Thus there is no objective risk of 252 prefix depletion by assigning /48 or /56 prefixes even to the 253 smallest sites. 255 3.2. Hierarchical addressing 257 Some operators have argued that more prefix bits are needed to allow 258 an aggregated hierarchical addressing scheme within a campus or 259 corporate network. However, if a campus or enterprise gets a /48 260 prefix (or shorter), then that already provides 16 bits for 261 hierarchical allocation. In any case, flat IGP routing is widely and 262 successfully used within rather large networks, with hundreds of 263 routers and thousands of end systems. Therefore there is no 264 objective need for additional prefix bits to support hierarchy and 265 aggregation within enterprises. 267 3.3. Audit requirement 269 Some network operators wish to know and audit which nodes are active 270 on a network, especially those that are allowed to communicate off 271 link or off site. They may also wish to limit the total number of 272 active addresses and sessions that can be sourced from a particular 273 host, LAN or site, in order to prevent potential resource depletion 274 attacks or other problems spreading beyond a certain scope of 275 control. It has been argued that this type of control would be 276 easier if only long network prefixes with relatively small numbers of 277 possible hosts per network were used, reducing the discovery problem. 278 However, such sites most typically operate using DHCPv6, which means 279 that all legitimate hosts are automatically known to the DHCPv6 280 servers, which is sufficient for audit purposes. Such hosts could, 281 if desired, be limited to a small range of IID values without 282 changing the /64 subnet length. Any hosts inadvertently obtaining 283 addresses via SLAAC can be audited through Neighbor Discovery logs. 285 3.4. Concerns over ND cache exhaustion 287 A site may be concerned that it is open to neighbour discovery (ND) 288 cache exhaustion attacks [RFC3756], whereby an attacker sends a large 289 number of messages in rapid succession to a series of (most likely 290 inactive) host addresses within a specific subnet. Such an attack 291 attempts to fill a router's ND cache with ND requests pending 292 completion, in so doing denying correct operation to active devices 293 on the network. 295 One potential way to mitigate this attack would be to consider using 296 a /120 prefix, thus limiting the number of addresses in the subnet to 297 be similar to an IPv4 /24 prefix, which should not cause any concerns 298 for ND cache exhaustion. Note that the prefix does need to be quite 299 long for this scenario to be valid. The number of theoretically 300 possible ND cache slots on the segment needs to be of the same order 301 of magnitude as the actual number of hosts. Thus small increases 302 from the /64 prefix length do not have a noticeable impact: even 2^32 303 potential entries, a factor of two billion decrease compared to 2^64, 304 is still more than enough to exhaust the memory on current routers. 305 Given that SLAAC assumes a 64 bit network boundary, in such an 306 approach hosts would likely need to use DHCPv6, or be manually 307 configured with addresses. 309 It should be noted that several other mitigations of the ND cache 310 attack are described in [RFC6583], and that limiting the size of the 311 cache and the number of incomplete entries allowed would also defeat 312 the attack. For the specific case of a point-to-point link between 313 routers, this attack is indeed mitigated by a /127 prefix [RFC6164]. 315 4. Effects of varying the interface identifier length 317 This section of the document analyses the impact and effects of 318 varying the length of an IPv6 unicast IID by reducing it to less than 319 64 bits. 321 4.1. Interaction with IPv6 specifications 323 The precise 64-bit length of the Interface ID is widely mentioned in 324 numerous RFCs describing various aspects of IPv6. It is not 325 straightforward to distinguish cases where this has normative impact 326 or affects interoperability. This section aims to identify 327 specifications that contain an explicit reference to the 64-bit 328 length. Regardless of implementation issues, the RFCs themselves 329 would all need to be updated if the 64-bit rule was changed, even if 330 the updates were small, which would involve considerable time and 331 effort. 333 First and foremost, the RFCs describing the architectural aspects of 334 IPv6 addressing explicitly state, refer and repeat this apparently 335 immutable value: Addressing Architecture [RFC4291], IPv6 Address 336 Assignment to End Sites [RFC6177], Reserved Interface Identifiers 337 [RFC5453], ILNP Node Identifiers [RFC6741]. Customer Edge routers 338 impose /64 for their interfaces [RFC7084]. The IPv6 Subnet Model 339 [RFC5942] points out that the assumption of a /64 prefix length is a 340 potential implementation error. 342 Numerous IPv6-over-foo documents make mandatory statements with 343 respect to the 64-bit length of the Interface ID to be used during 344 the Stateless Autoconfiguration. These documents include [RFC2464] 345 (Ethernet), [RFC2467] (FDDI), [RFC2470] (Token Ring), [RFC2492] 346 (ATM), [RFC2497] (ARCnet), [RFC2590] (Frame Relay), [RFC3146] (IEEE 347 1394), [RFC4338] (Fibre Channel), [RFC4944] (IEEE 802.15.4), 348 [RFC5072] (PPP), [RFC5121] [RFC5692] (IEEE 802.16), [RFC2529] 349 (6over4), [RFC5214] (ISATAP), [I-D.templin-aerolink] (AERO), 350 [I-D.ietf-6lowpan-btle], [I-D.ietf-6man-6lobac], 351 [I-D.brandt-6man-lowpanz]. 353 To a lesser extent, the address configuration RFCs themselves may in 354 some ways assume the 64-bit length of an Interface ID (e.g, SLAAC for 355 the link-local addresses, DHCPv6 for the potentially assigned EUI- 356 64-based IP addresses, Optimistic Duplicate Address Detection 357 [RFC4429] which computes 64-bit-based collision probabilities). 359 The MLDv1 [RFC2710] and MLDv2 [RFC3810] protocols mandate that all 360 queries be sent with a link-local source address, with the exception 361 of MLD messages sent using the unspecified address when the link- 362 local address is tentative [RFC3590]. At the time of publication of 363 RFC 2710, the IPv6 addressing architecture specified link-local 364 addresses with 64-bit interface identifiers. MLDv2 explicitly 365 specifies the use of the fe80::/64 link-local prefix, and bases the 366 querier election algorithm on the link-local subnet prefix of length 367 /64. 369 The IPv6 Flow Label Specification [RFC6437] gives an example of a 370 20-bit hash function generation which relies on splitting an IPv6 371 address in two equally-sized 64bit-length parts. 373 The basic transition mechanisms [RFC4213] refer to IIDs of length 64 374 for link-local addresses, and other transition mechanisms such as 375 Teredo [RFC4380] assume the use of IIDs of length 64. Similar 376 assumptions are found in 6to4 [RFC3056] and 6rd [RFC5969]. 377 Translation-based transition mechanisms such as NAT64 and NPTv6 have 378 some dependency on prefix length, discussed below. 380 The proposed method [RFC7278] of extending an assigned /64 prefix 381 from a smartphone's cellular interface to its WiFi link relies on 382 prefix length, and implicitly on the length of the Interface ID, to 383 be valued at 64. 385 The CGA and HBA specifications rely on the 64-bit identifier length 386 (see below), as do the Privacy extensions [RFC4941] and some examples 387 in IKEv2bis [RFC5996]. 389 464XLAT [RFC6877] explicitly mentions acquiring /64 prefixes. 390 However, it also discusses the possibility of using the interface 391 address on the device as the endpoint for the traffic, thus 392 potentially removing this dependency. 394 [RFC2526] reserves a number of subnet anycast addresses by reserving 395 some anycast IIDs. An anycast IID so reserved cannot be less than 7 396 bits long. This means that a subnet prefix length longer than /121 397 is not possible, and a subnet of exactly /121 would be useless since 398 all its identifiers are reserved. It also means that half of a /120 399 is reserved for anycast. This could of course be fixed in the way 400 described for /127 in [RFC6164], i.e., avoiding the use of anycast 401 within a /120 subnet. Note that support for "on-link anycast" is a 402 standard IPv6 neighbor discovery capability [RFC4861][RFC7094], and 403 therefore applications and their developers would expect it to be 404 available. 406 The Mobile IP home network models [RFC4887] rely heavily on the /64 407 subnet length and assume a 64-bit IID. 409 While preparing this document, it was noted that many other IPv6 410 specifications refer to mandatory alignment on 64-bit boundaries, 411 64-bit data structures, 64-bit counters in MIBs, 64-bit sequence 412 numbers and cookies in security, etc. Finally, the number "64" may 413 be considered "magic" in some RFCs, e.g., 64k limits in DNS and 414 Base64 encodings in MIME. None of this has any influence on the 415 length of the IID, but might confuse a careless reader. 417 4.2. Possible failure modes 419 This section discusses several specific aspects of IPv6 where we can 420 expect operational failures with subnet prefixes other than /64. 422 o Router implementations: Router implementors might interpret IETF 423 standards such as [RFC6164] and [RFC7136] to indicate that 424 prefixes between /65 and /126 inclusive for unicast packets on- 425 the-wire are invalid, and operational practices that utilize 426 prefix lengths in this range may fail on some devices, as 427 discussed in Section 4.3.2. 429 o Multicast: [RFC3306] defines a method for generating IPv6 430 multicast group addresses based on unicast prefixes. This method 431 assumes a longest prefix of 64 bits. If a longer prefix is used, 432 there is no way to generate a specific multicast group address 433 using this method. In such cases the administrator would need to 434 use an "artificial" prefix from within their allocation (a /64 or 435 shorter) from which to generate the group address. This prefix 436 would not correspond to a real subnet. 438 Similarly [RFC3956], which specifies Embedded-RP, allowing IPv6 439 multicast rendezvous point addresses to be embedded in the 440 multicast group address, would also fail, as the scheme assumes a 441 maximum prefix length of 64 bits. 443 o CGA: The Cryptographically Generated Address format (CGA, 444 [RFC3972]) is heavily based on a /64 interface identifier. 445 [RFC3972] has defined a detailed algorithm showing how to generate 446 a 64-bit interface identifier from a public key and a 64-bit 447 subnet prefix. Changing the /64 boundary would certainly 448 invalidate the current CGA definition. However, CGA might benefit 449 in a redefined version if more bits are used for interface 450 identifier (which means shorter prefix length). For now, 59 bits 451 are used for cryptographic purposes. The more bits are available, 452 the stronger CGA could be. Conversely, longer prefixes would 453 weaken CGA. 455 o NAT64: Both stateless [RFC6052] NAT64 and stateful NAT64 [RFC6146] 456 are flexible for the prefix length. [RFC6052] has defined 457 multiple address formats for NAT64. In Section 2 "IPv4-Embedded 458 IPv6 Prefix and Format" of [RFC6052], the network-specific prefix 459 could be one of /32, /40, /48, /56, /64 and /96. The remaining 460 part of the IPv6 address is constructed by a 32-bit IPv4 address, 461 a 8-bit u byte and a variable length suffix (there is no u byte 462 and suffix in the case of 96-bit Well-Known Prefix). NAT64 is 463 therefore OK with a subnet boundary out to /96, but not longer. 465 o NPTv6: IPv6-to-IPv6 Network Prefix Translation [RFC6296] is also 466 bound to /64 boundary. NPTv6 maps a /64 prefix to another /64 467 prefix. When the NPTv6 Translator is configured with a /48 or 468 shorter prefix, the 64-bit interface identifier is kept unmodified 469 during translation. However, the /64 boundary might be changed as 470 long as the "inside" and "outside" prefixes have the same length. 472 o ILNP: Identifier-Locator Network Protocol (ILNP) [RFC6741] is 473 designed around the /64 boundary, since it relies on locally 474 unique 64-bit node identifiers (in the interface identifier 475 field). While a re-design to use longer prefixes is not 476 inconceivable, this would need major changes to the existing 477 specification for the IPv6 version of ILNP. 479 o shim6: The Multihoming Shim Protocol for IPv6 (shim6) [RFC5533] in 480 its insecure form treats IPv6 address as opaque 128-bit objects. 481 However, to secure the protocol against spoofing, it is essential 482 to either use CGAs (see above) or Hash-Based Addresses (HBA) 483 [RFC5535]. Like CGAs, HBAs are generated using a procedure that 484 assumes a 64-bit identifier. Therefore, in effect, secure shim6 485 is affected by the /64 boundary exactly like CGAs. 487 o Duplicate address risk: If SLAAC was modified to work with shorter 488 IIDs, the statistical risk of hosts choosing the same pseudo- 489 random identifier [RFC7217] would increase correspondingly. The 490 practical impact of this would range from slight to dramatic, 491 depending on how much the IID length was reduced. In particular, 492 a /120 prefix would imply an 8 bit IID and address collisions 493 would be highly probable. 495 o The link-local prefix: While RFC 4862 is careful not to define any 496 specific length of link-local prefix within fe80::/10, the 497 addressing architecture [RFC4291] does define the link-local IID 498 length to be 64 bits. If different hosts on a link used IIDs of 499 different lengths to form a link-local address, there is potential 500 for confusion and unpredictable results. Typically today the 501 choice of 64 bits for the link-local IID length is hard-coded per 502 interface, in accordance with the relevant IPv6-over-foo 503 specification, and systems behave as if the link local prefix was 504 actually fe80::/64. There might be no way to change this except 505 conceivably by manual configuration, which will be impossible if 506 the host concerned has no local user interface. 508 It goes without saying that if prefixes longer than /64 are to be 509 used, all hosts must be capable of generating IIDs shorter than 64 510 bits, in order to follow the auto-configuration procedure correctly 511 [RFC4862]. 513 4.3. Experimental observations 515 4.3.1. Survey of the processing of Neighbor Discovery options with 516 prefixes other than /64 518 This section provides a survey of the processing of Neighbor 519 Discovery options which include prefixes that are different than /64. 521 The behavior of nodes was assessed with respect to the following 522 options: 524 o PIO-A: Prefix Information Option (PIO) [RFC4861] with the A bit 525 set. 527 o PIO-L: Prefix Information Option (PIO) [RFC4861] with the L bit 528 set. 530 o PIO-AL: Prefix Information Option (PIO) [RFC4861] with both the A 531 and L bits set. 533 o RIO: Route Information Option (RIO) [RFC4191]. 535 In the tables below, the following notation is used: 537 NOT-SUP: 538 This option is not supported (i.e., it is ignored no matter the 539 prefix length used). 541 LOCAL: 542 The corresponding prefix is considered "on-link". 544 ROUTE 545 The corresponding route is added to the IPv6 routing table. 547 IGNORE: 548 The Option is ignored as an error. 550 +--------------------+--------+-------+--------+---------+ 551 | Operating System | PIO-A | PIO-L | PIO-AL | RIO | 552 +--------------------+--------+-------+--------+---------+ 553 | FreeBSD 9.0 | IGNORE | LOCAL | LOCAL | NOT-SUP | 554 +--------------------+--------+-------+--------+---------+ 555 | Linux 3.0.0-15 | IGNORE | LOCAL | LOCAL | NOT-SUP | 556 +--------------------+--------+-------+--------+---------+ 557 | Linux-current | IGNORE | LOCAL | LOCAL | NOT-SUP | 558 +--------------------+--------+-------+--------+---------+ 559 | NetBSD 5.1 | IGNORE | LOCAL | LOCAL | NOT-SUP | 560 +--------------------+--------+-------+--------+---------+ 561 | OpenBSD-current | IGNORE | LOCAL | LOCAL | NOT-SUP | 562 +--------------------+--------+-------+--------+---------+ 563 | Win XP SP2 | IGNORE | LOCAL | LOCAL | ROUTE | 564 +--------------------+--------+-------+--------+---------+ 565 | Win 7 Home Premium | IGNORE | LOCAL | LOCAL | ROUTE | 566 +--------------------+--------+-------+--------+---------+ 568 Table 1: Processing of ND options with prefixes longer than /64 569 +--------------------+--------+-------+--------+---------+ 570 | Operating System | PIO-A | PIO-L | PIO-AL | RIO | 571 +--------------------+--------+-------+--------+---------+ 572 | FreeBSD 9.0 | IGNORE | LOCAL | LOCAL | NOT-SUP | 573 +--------------------+--------+-------+--------+---------+ 574 | Linux 3.0.0-15 | IGNORE | LOCAL | LOCAL | NOT-SUP | 575 +--------------------+--------+-------+--------+---------+ 576 | Linux-current | IGNORE | LOCAL | LOCAL | NOT-SUP | 577 +--------------------+--------+-------+--------+---------+ 578 | NetBSD 5.1 | IGNORE | LOCAL | LOCAL | NOT-SUP | 579 +--------------------+--------+-------+--------+---------+ 580 | OpenBSD-current | IGNORE | LOCAL | LOCAL | NOT-SUP | 581 +--------------------+--------+-------+--------+---------+ 582 | Win XP SP2 | IGNORE | LOCAL | LOCAL | ROUTE | 583 +--------------------+--------+-------+--------+---------+ 584 | Win 7 Home Premium | IGNORE | LOCAL | LOCAL | ROUTE | 585 +--------------------+--------+-------+--------+---------+ 587 Table 2: Processing of ND options with prefixes shorter than /64 589 The results obtained can be summarized as follows: 591 o the "A" bit in the Prefix Information Options is honored only if 592 the prefix length is 64. At least for the case where the IID 593 length is defined to be 64 bits in the corresponding link-type- 594 specific document, which is the case for all currently published 595 such documents, this is consistent with [RFC4862], which defines 596 the case where the sum of the advertised prefix length and the IID 597 length does not equal 128 as an error condition. 599 o the "L" bit in the Prefix Information Options is honored for any 600 arbitrary prefix length (whether shorter or longer than /64). 602 o nodes that support the Route Information Option, allow such routes 603 to be specified with prefixes of any arbitrary length (whether 604 shorter or longer than /64) 606 4.3.2. Other Observations 608 Participants in the V6OPS working group have indicated that some 609 forwarding devices have been shown to work correctly with long 610 prefixes such as /80 or /96. Indeed, it is to be expected that 611 longest prefix match based forwarding will work for any prefix 612 length, and no reports of this completely failing have been noted. 613 Also, DHCPv6 is in widespread use without any dependency on the /64 614 boundary. Reportedly, there are deployments of /120 subnets 615 configured using DHCPv6. 617 There have been definite reports that some routers have a performance 618 drop-off or even resource exhaustion for prefixes longer than /64, 619 due to design issues. In particular, some routing chip designs 620 allocate much less space for longer prefixes than for prefixes up to 621 /64, for the sake of savings in memory, power and lookup latency. 622 Some devices need special-case code to handle point-to-point links 623 according to [RFC6164]. 625 It has been reported that at least one type of switch has a content- 626 addressable memory limited to 144 bits, which is indeed a typical 627 value for commodity components [TCAM]. This means that packet 628 filters or access control lists cannot be defined based on 128-bit 629 addresses and two 16-bit port numbers; the longest prefix that could 630 be used in such a filter is a /112. 632 4.4. Implementation and deployment issues 634 From an early stage, implementations and deployments of IPv6 assumed 635 the /64 subnet length, even though routing was based on prefixes of 636 any length. As shown above, this became anchored in many 637 specifications (Section 4.1) and in important aspects of 638 implementations commonly used in local area networks (Section 4.3). 639 In fact, a programmer might be lulled into assuming a comfortable 640 rule of thumb that subnet prefixes are always /64 and an IID is 641 always of length 64. Apart from the limited evidence in 642 Section 4.3.1, we cannot tell without code inspections or tests 643 whether existing stacks are able to handle a flexible IID length, or 644 whether they would require modification to do so. A conforming 645 implementation of an IPv6-over-foo that specifies a 64 bit IID for 646 foo links will of course only support 64. But in a well designed 647 stack, the IP layer itself will treat that 64 as a parameter, so 648 changing the IID length in the IPv6-over-foo code should be all that 649 is necessary. 651 The main practical consequence of the existing specifications is that 652 deployments in which longer subnet prefixes are used cannot make use 653 of SLAAC-configured addresses, and require either manually configured 654 addresses or DHCPv6. To reverse this argument, if it was considered 655 desirable to allow auto-configured addresses with subnet prefixes 656 longer than /64, all of the specifications identified above as 657 depending on /64 would have to be modified, with due regard to 658 interoperability with unmodified stacks. In fact [RFC7217] allows 659 for this possibility. Then modified stacks would have to be 660 developed and deployed. It might be the case that some stacks 661 contain dependencies on the /64 boundary which are not directly 662 implied by the specifications, and any such hidden dependencies would 663 also need to be found and removed. 665 At least one DHCPv6 client unconditionally installs a /64 prefix as 666 on-link when it configures an interface with an address, although 667 some specific operating system vendors seem to change this default 668 behavior by tweaking a client-side script. This is in clear 669 violation of the IPv6 subnet model [RFC5942]. The motivation for 670 this choice is that if there is no router on the link, the hosts 671 would fail to communicate with each other using the configured 672 addresses because the "on-link assumption" was removed in [RFC4861]. 673 This is not really about the magic number of 64, but an 674 implementation may sometimes pick an arbitrary value of prefix length 675 due to the removal of the on-link assumption, and the value chosen 676 will most likely be 64. 678 Typical IP Address Management (IPAM) tools treat /64 as the default 679 subnet length, but allow users to specify longer subnet prefixes if 680 desired. Clearly, all IPAM tools and network management systems 681 would need to be checked in detail. 683 Finally, IPv6 is already deployed at many sites, with a large number 684 of staff trained on the basis of the existing standards, supported by 685 documentation and tools based on those standards. Numerous existing 686 middlebox devices are also based on those standards. These people, 687 documents, tools and devices represent a very large investment that 688 would be seriously impacted by a change in the /64 boundary. 690 4.5. Privacy issues 692 The length of the interface identifier has implications for privacy 693 [I-D.ietf-6man-ipv6-address-generation-privacy]. In any case in 694 which the value of the identifier is intended to be hard to guess, 695 whether or not it is cryptographically generated, it is apparent that 696 more bits are better. For example, if there are only 20 bits to be 697 guessed, at most just over a million guesses are needed, today well 698 within the capacity of a low cost attack mechanism. It is hard to 699 state in general how many bits are enough to protect privacy, since 700 this depends on the resources available to the attacker, but it seems 701 clear that a privacy solution needs to resist an attack requiring 702 billions rather than millions of guesses. Trillions would be better, 703 suggesting that at least 40 bits should be available. Thus we can 704 argue that subnet prefixes longer than say /80 might raise privacy 705 concerns by making the IID guessable. 707 A prefix long enough to limit the number of addresses comparably to 708 an IPv4 subnet, such as /120, would create exactly the same situation 709 for privacy as IPv4 except for the absence of NAT. In particular, a 710 host would be forced to pick a new IID when roaming to a new network, 711 to avoid collisions. As mentioned earlier, it is likely that SLAAC 712 will not be used on such a subnet. 714 5. Security Considerations 716 In addition to the privacy issues mentioned in Section 4.5, and the 717 issues mentioned with CGAs and HBAs in Section 4.2, the length of the 718 subnet prefix affects the matter of defence against scanning attacks 719 [I-D.ietf-opsec-ipv6-host-scanning]. Assuming the attacker has 720 discovered or guessed the prefix length, a longer prefix reduces the 721 space that the attacker needs to scan, e.g., to only 256 addresses if 722 the prefix is /120. On the other hand, if the attacker has not 723 discovered the prefix length and assumes it to be /64, routers can 724 trivially discard attack packets that do not fall within an actual 725 subnet. 727 However, assume that an attacker finds one valid address A and 728 assumes that it is within a long prefix such as a /120. The attacker 729 then starts a scanning attack by scanning "outwards" from A, by 730 trying A+1, A-1, A+2, A-2, etc. This attacker will easily find all 731 hosts in any subnet with a long prefix, because they will have 732 addresses close to A. We therefore conclude that any prefix 733 containing densely packed valid addresses is vulnerable to a scanning 734 attack, without the attacker needing to guess the prefix length. 735 Therefore, to preserve IPv6's advantage over IPv4 in resisting 736 scanning attacks, it is important that subnet prefixes are short 737 enough to allow sparse allocation of identifiers within each subnet. 738 The considerations are similar to those for privacy, and we can again 739 argue that prefixes longer than say /80 might significantly increase 740 vulnerability. Ironically, this argument is exactly converse to the 741 argument for longer prefixes to resist an ND cache attack, as 742 described in Section 3.4. 744 Denial of service attacks related to Neighbor Discovery are discussed 745 in Section 3.4 and in [RFC6583]. One of the mitigations suggested by 746 that document is "sizing subnets to reflect the number of addresses 747 actually in use", but the fact that this greatly simplifies scanning 748 attacks is not noted. For further discussion of scanning attacks, 749 see [I-D.ietf-opsec-ipv6-host-scanning]. 751 Note that, although not known at the time of writing, there might be 752 other resource exhaustion attacks available, similar in nature to the 753 ND cache attack. We cannot exclude that such attacks might be 754 exacerbated by sparsely populated subnets such as a /64. It should 755 also be noted that this analysis assumes a conventional deployment 756 model with a significant number of end-systems located in a single 757 LAN broadcast domain. Other deployment models might lead to 758 different conclusions. 760 6. IANA Considerations 762 This document requests no action by IANA. 764 7. Acknowledgements 766 This document was inspired by a vigorous discussion on the V6OPS 767 working group mailing list with at least 20 participants. Later, 768 valuable comments were received from Ran Atkinson, Fred Baker, Steven 769 Blake, Lorenzo Colitti, David Farmer, Bill Fenner, Ray Hunter, 770 Paraskevi Iliadou, Jen Linkova, Philip Matthews, Matthew Petach, 771 Scott Schmit, Tatuya Jinmei, Fred Templin, Ole Troan, Stig Venaas, 772 and numerous other participants in the 6MAN working group. An 773 extremely detailed review by Mark Smith was especially helpful. 775 This document was produced using the xml2rfc tool [RFC2629]. 777 8. Change log [RFC Editor: Please remove] 779 draft-ietf-6man-why64-05: Area Director review comments, 2014-09-16. 781 draft-ietf-6man-why64-04: fixed reference error, 2014-09-10. 783 draft-ietf-6man-why64-03: fixed nits, 2014-08-27. 785 draft-ietf-6man-why64-02: responded to WGLC reviews and comments, 786 2014-08-16. 788 draft-ietf-6man-why64-01: language improvements, added TCAM 789 reference, 2014-05-07. 791 draft-ietf-6man-why64-00: WG adoption, WG comments, including major 792 text reorganisation: 3 main sections describe advantages of fixed 793 length IID, arguments for shorter lengths, and expected effects of 794 varying the length, 2014-04-11. 796 draft-carpenter-6man-why64-01: WG comments, added experimental 797 results, implementation/deployment text, 2014-02-06. 799 draft-carpenter-6man-why64-00: original version, 2014-01-06. 801 9. References 803 9.1. Normative References 805 [RFC2464] Crawford, M., "Transmission of IPv6 Packets over Ethernet 806 Networks", RFC 2464, December 1998. 808 [RFC2467] Crawford, M., "Transmission of IPv6 Packets over FDDI 809 Networks", RFC 2467, December 1998. 811 [RFC2470] Crawford, M., Narten, T., and S. Thomas, "Transmission of 812 IPv6 Packets over Token Ring Networks", RFC 2470, December 813 1998. 815 [RFC2492] Armitage, G., Schulter, P., and M. Jork, "IPv6 over ATM 816 Networks", RFC 2492, January 1999. 818 [RFC2497] Souvatzis, I., "Transmission of IPv6 Packets over ARCnet 819 Networks", RFC 2497, January 1999. 821 [RFC2526] Johnson, D. and S. Deering, "Reserved IPv6 Subnet Anycast 822 Addresses", RFC 2526, March 1999. 824 [RFC2529] Carpenter, B. and C. Jung, "Transmission of IPv6 over IPv4 825 Domains without Explicit Tunnels", RFC 2529, March 1999. 827 [RFC2590] Conta, A., Malis, A., and M. Mueller, "Transmission of 828 IPv6 Packets over Frame Relay Networks Specification", RFC 829 2590, May 1999. 831 [RFC2710] Deering, S., Fenner, W., and B. Haberman, "Multicast 832 Listener Discovery (MLD) for IPv6", RFC 2710, October 833 1999. 835 [RFC3056] Carpenter, B. and K. Moore, "Connection of IPv6 Domains 836 via IPv4 Clouds", RFC 3056, February 2001. 838 [RFC3146] Fujisawa, K. and A. Onoe, "Transmission of IPv6 Packets 839 over IEEE 1394 Networks", RFC 3146, October 2001. 841 [RFC3306] Haberman, B. and D. Thaler, "Unicast-Prefix-based IPv6 842 Multicast Addresses", RFC 3306, August 2002. 844 [RFC3315] Droms, R., Bound, J., Volz, B., Lemon, T., Perkins, C., 845 and M. Carney, "Dynamic Host Configuration Protocol for 846 IPv6 (DHCPv6)", RFC 3315, July 2003. 848 [RFC3590] Haberman, B., "Source Address Selection for the Multicast 849 Listener Discovery (MLD) Protocol", RFC 3590, September 850 2003. 852 [RFC3810] Vida, R. and L. Costa, "Multicast Listener Discovery 853 Version 2 (MLDv2) for IPv6", RFC 3810, June 2004. 855 [RFC3956] Savola, P. and B. Haberman, "Embedding the Rendezvous 856 Point (RP) Address in an IPv6 Multicast Address", RFC 857 3956, November 2004. 859 [RFC3972] Aura, T., "Cryptographically Generated Addresses (CGA)", 860 RFC 3972, March 2005. 862 [RFC4191] Draves, R. and D. Thaler, "Default Router Preferences and 863 More-Specific Routes", RFC 4191, November 2005. 865 [RFC4213] Nordmark, E. and R. Gilligan, "Basic Transition Mechanisms 866 for IPv6 Hosts and Routers", RFC 4213, October 2005. 868 [RFC4291] Hinden, R. and S. Deering, "IP Version 6 Addressing 869 Architecture", RFC 4291, February 2006. 871 [RFC4338] DeSanti, C., Carlson, C., and R. Nixon, "Transmission of 872 IPv6, IPv4, and Address Resolution Protocol (ARP) Packets 873 over Fibre Channel", RFC 4338, January 2006. 875 [RFC4380] Huitema, C., "Teredo: Tunneling IPv6 over UDP through 876 Network Address Translations (NATs)", RFC 4380, February 877 2006. 879 [RFC4429] Moore, N., "Optimistic Duplicate Address Detection (DAD) 880 for IPv6", RFC 4429, April 2006. 882 [RFC4861] Narten, T., Nordmark, E., Simpson, W., and H. Soliman, 883 "Neighbor Discovery for IP version 6 (IPv6)", RFC 4861, 884 September 2007. 886 [RFC4862] Thomson, S., Narten, T., and T. Jinmei, "IPv6 Stateless 887 Address Autoconfiguration", RFC 4862, September 2007. 889 [RFC4941] Narten, T., Draves, R., and S. Krishnan, "Privacy 890 Extensions for Stateless Address Autoconfiguration in 891 IPv6", RFC 4941, September 2007. 893 [RFC4944] Montenegro, G., Kushalnagar, N., Hui, J., and D. Culler, 894 "Transmission of IPv6 Packets over IEEE 802.15.4 895 Networks", RFC 4944, September 2007. 897 [RFC5072] Varada, S., Haskins, D., and E. Allen, "IP Version 6 over 898 PPP", RFC 5072, September 2007. 900 [RFC5121] Patil, B., Xia, F., Sarikaya, B., Choi, JH., and S. 901 Madanapalli, "Transmission of IPv6 via the IPv6 902 Convergence Sublayer over IEEE 802.16 Networks", RFC 5121, 903 February 2008. 905 [RFC5214] Templin, F., Gleeson, T., and D. Thaler, "Intra-Site 906 Automatic Tunnel Addressing Protocol (ISATAP)", RFC 5214, 907 March 2008. 909 [RFC5453] Krishnan, S., "Reserved IPv6 Interface Identifiers", RFC 910 5453, February 2009. 912 [RFC5533] Nordmark, E. and M. Bagnulo, "Shim6: Level 3 Multihoming 913 Shim Protocol for IPv6", RFC 5533, June 2009. 915 [RFC5535] Bagnulo, M., "Hash-Based Addresses (HBA)", RFC 5535, June 916 2009. 918 [RFC5692] Jeon, H., Jeong, S., and M. Riegel, "Transmission of IP 919 over Ethernet over IEEE 802.16 Networks", RFC 5692, 920 October 2009. 922 [RFC5942] Singh, H., Beebee, W., and E. Nordmark, "IPv6 Subnet 923 Model: The Relationship between Links and Subnet 924 Prefixes", RFC 5942, July 2010. 926 [RFC5969] Townsley, W. and O. Troan, "IPv6 Rapid Deployment on IPv4 927 Infrastructures (6rd) -- Protocol Specification", RFC 928 5969, August 2010. 930 [RFC5996] Kaufman, C., Hoffman, P., Nir, Y., and P. Eronen, 931 "Internet Key Exchange Protocol Version 2 (IKEv2)", RFC 932 5996, September 2010. 934 [RFC6052] Bao, C., Huitema, C., Bagnulo, M., Boucadair, M., and X. 935 Li, "IPv6 Addressing of IPv4/IPv6 Translators", RFC 6052, 936 October 2010. 938 [RFC6146] Bagnulo, M., Matthews, P., and I. van Beijnum, "Stateful 939 NAT64: Network Address and Protocol Translation from IPv6 940 Clients to IPv4 Servers", RFC 6146, April 2011. 942 [RFC6164] Kohno, M., Nitzan, B., Bush, R., Matsuzaki, Y., Colitti, 943 L., and T. Narten, "Using 127-Bit IPv6 Prefixes on Inter- 944 Router Links", RFC 6164, April 2011. 946 [RFC6177] Narten, T., Huston, G., and L. Roberts, "IPv6 Address 947 Assignment to End Sites", BCP 157, RFC 6177, March 2011. 949 [RFC6296] Wasserman, M. and F. Baker, "IPv6-to-IPv6 Network Prefix 950 Translation", RFC 6296, June 2011. 952 [RFC6437] Amante, S., Carpenter, B., Jiang, S., and J. Rajahalme, 953 "IPv6 Flow Label Specification", RFC 6437, November 2011. 955 [RFC7084] Singh, H., Beebee, W., Donley, C., and B. Stark, "Basic 956 Requirements for IPv6 Customer Edge Routers", RFC 7084, 957 November 2013. 959 [RFC7136] Carpenter, B. and S. Jiang, "Significance of IPv6 960 Interface Identifiers", RFC 7136, February 2014. 962 9.2. Informative References 964 [DRAFT-odell] 965 O'Dell, M., "8+8 - An Alternate Addressing Architecture 966 for IPv6", draft-odell-8+8.00 (work in progress), October 967 1996. 969 [I-D.brandt-6man-lowpanz] 970 Brandt, A. and J. Buron, "Transmission of IPv6 packets 971 over ITU-T G.9959 Networks", draft-brandt-6man-lowpanz-02 972 (work in progress), June 2013. 974 [I-D.ietf-6lowpan-btle] 975 Nieminen, J., Savolainen, T., Isomaki, M., Patil, B., 976 Shelby, Z., and C. Gomez, "Transmission of IPv6 Packets 977 over BLUETOOTH Low Energy", draft-ietf-6lowpan-btle-12 978 (work in progress), February 2013. 980 [I-D.ietf-6man-6lobac] 981 Lynn, K., Martocci, J., Neilson, C., and S. Donaldson, 982 "Transmission of IPv6 over MS/TP Networks", draft-ietf- 983 6man-6lobac-01 (work in progress), March 2012. 985 [I-D.ietf-6man-ipv6-address-generation-privacy] 986 Cooper, A., Gont, F., and D. Thaler, "Privacy 987 Considerations for IPv6 Address Generation Mechanisms", 988 draft-ietf-6man-ipv6-address-generation-privacy-01 (work 989 in progress), February 2014. 991 [I-D.ietf-homenet-arch] 992 Chown, T., Arkko, J., Brandt, A., Troan, O., and J. Weil, 993 "IPv6 Home Networking Architecture Principles", draft- 994 ietf-homenet-arch-17 (work in progress), July 2014. 996 [I-D.ietf-opsec-ipv6-host-scanning] 997 Gont, F. and T. Chown, "Network Reconnaissance in IPv6 998 Networks", draft-ietf-opsec-ipv6-host-scanning-04 (work in 999 progress), June 2014. 1001 [I-D.templin-aerolink] 1002 Templin, F., "Transmission of IP Packets over AERO Links", 1003 draft-templin-aerolink-35 (work in progress), September 1004 2014. 1006 [IEEE802] IEEE, "IEEE Standard for Local and Metropolitan Area 1007 Networks: Overview and Architecture", IEEE Std 802-2001 1008 (R2007), 2007. 1010 [RFC2629] Rose, M., "Writing I-Ds and RFCs using XML", RFC 2629, 1011 June 1999. 1013 [RFC3756] Nikander, P., Kempf, J., and E. Nordmark, "IPv6 Neighbor 1014 Discovery (ND) Trust Models and Threats", RFC 3756, May 1015 2004. 1017 [RFC4692] Huston, G., "Considerations on the IPv6 Host Density 1018 Metric", RFC 4692, October 2006. 1020 [RFC4887] Thubert, P., Wakikawa, R., and V. Devarapalli, "Network 1021 Mobility Home Network Models", RFC 4887, July 2007. 1023 [RFC5505] Aboba, B., Thaler, D., Andersson, L., and S. Cheshire, 1024 "Principles of Internet Host Configuration", RFC 5505, May 1025 2009. 1027 [RFC6583] Gashinsky, I., Jaeggli, J., and W. Kumari, "Operational 1028 Neighbor Discovery Problems", RFC 6583, March 2012. 1030 [RFC6741] Atkinson,, RJ., "Identifier-Locator Network Protocol 1031 (ILNP) Engineering Considerations", RFC 6741, November 1032 2012. 1034 [RFC6877] Mawatari, M., Kawashima, M., and C. Byrne, "464XLAT: 1035 Combination of Stateful and Stateless Translation", RFC 1036 6877, April 2013. 1038 [RFC7094] McPherson, D., Oran, D., Thaler, D., and E. Osterweil, 1039 "Architectural Considerations of IP Anycast", RFC 7094, 1040 January 2014. 1042 [RFC7217] Gont, F., "A Method for Generating Semantically Opaque 1043 Interface Identifiers with IPv6 Stateless Address 1044 Autoconfiguration (SLAAC)", RFC 7217, April 2014. 1046 [RFC7278] Byrne, C., Drown, D., and A. Vizdal, "Extending an IPv6 1047 /64 Prefix from a Third Generation Partnership Project 1048 (3GPP) Mobile Interface to a LAN Link", RFC 7278, June 1049 2014. 1051 [TCAM] Meiners, C., Liu, A., and E. Torng, "Algorithmic 1052 Approaches to Redesigning TCAM-Based Systems", ACM 1053 SIGMETRICS'08 467-468, 2008. 1055 Authors' Addresses 1057 Brian Carpenter (editor) 1058 Department of Computer Science 1059 University of Auckland 1060 PB 92019 1061 Auckland 1142 1062 New Zealand 1064 Email: brian.e.carpenter@gmail.com 1066 Tim Chown 1067 University of Southampton 1068 Southampton, Hampshire SO17 1BJ 1069 United Kingdom 1071 Email: tjc@ecs.soton.ac.uk 1073 Fernando Gont 1074 SI6 Networks / UTN-FRH 1075 Evaristo Carriego 2644 1076 Haedo, Provincia de Buenos Aires 1706 1077 Argentina 1079 Email: fgont@si6networks.com 1080 Sheng Jiang 1081 Huawei Technologies Co., Ltd 1082 Q14, Huawei Campus 1083 No.156 Beiqing Road 1084 Hai-Dian District, Beijing 100095 1085 P.R. China 1087 Email: jiangsheng@huawei.com 1089 Alexandru Petrescu 1090 CEA, LIST 1091 CEA Saclay 1092 Gif-sur-Yvette, Ile-de-France 91190 1093 France 1095 Email: Alexandru.Petrescu@cea.fr 1097 Andrew Yourtchenko 1098 cisco 1099 7a de Kleetlaan 1100 Diegem 1830 1101 Belgium 1103 Email: ayourtch@cisco.com