1 Network Working Group A. Durand 2 Request for Comments: 4472 Comcast 3 Category: Informational J. Ihren 4 Autonomica 5 P. Savola 6 CSC/FUNET 7 April 2006 8 9 10 Operational Considerations and Issues with IPv6 DNS 11 12 Status of This Memo 13 14 This memo provides information for the Internet community. It does 15 not specify an Internet standard of any kind. Distribution of this 16 memo is unlimited. 17 18 Copyright Notice 19 20 Copyright (C) The Internet Society (2006). 21 22 Abstract 23 24 This memo presents operational considerations and issues with IPv6 25 Domain Name System (DNS), including a summary of special IPv6 26 addresses, documentation of known DNS implementation misbehavior, 27 recommendations and considerations on how to perform DNS naming for 28 service provisioning and for DNS resolver IPv6 support, 29 considerations for DNS updates for both the forward and reverse 30 trees, and miscellaneous issues. This memo is aimed to include a 31 summary of information about IPv6 DNS considerations for those who 32 have experience with IPv4 DNS. 33 34 Table of Contents 35 36 1. Introduction ....................................................3 37 1.1. Representing IPv6 Addresses in DNS Records .................3 38 1.2. Independence of DNS Transport and DNS Records ..............4 39 1.3. Avoiding IPv4/IPv6 Name Space Fragmentation ................4 40 1.4. Query Type '*' and A/AAAA Records ..........................4 41 2. DNS Considerations about Special IPv6 Addresses .................5 42 2.1. Limited-Scope Addresses ....................................5 43 2.2. Temporary Addresses ........................................5 44 2.3. 6to4 Addresses .............................................5 45 2.4. Other Transition Mechanisms ................................5 46 3. Observed DNS Implementation Misbehavior .........................6 47 3.1. Misbehavior of DNS Servers and Load-balancers ..............6 48 3.2. Misbehavior of DNS Resolvers ...............................6 49 50 51 52 Durand, et al. Informational [Page 1] 53 RFC 4472 Considerations with IPv6 DNS April 2006 54 55 56 4. Recommendations for Service Provisioning Using DNS ..............7 57 4.1. Use of Service Names instead of Node Names .................7 58 4.2. Separate vs. the Same Service Names for IPv4 and IPv6 ......8 59 4.3. Adding the Records Only When Fully IPv6-enabled ............8 60 4.4. The Use of TTL for IPv4 and IPv6 RRs .......................9 61 4.4.1. TTL with Courtesy Additional Data ...................9 62 4.4.2. TTL with Critical Additional Data ..................10 63 4.5. IPv6 Transport Guidelines for DNS Servers .................10 64 5. Recommendations for DNS Resolver IPv6 Support ..................10 65 5.1. DNS Lookups May Query IPv6 Records Prematurely ............10 66 5.2. Obtaining a List of DNS Recursive Resolvers ...............12 67 5.3. IPv6 Transport Guidelines for Resolvers ...................12 68 6. Considerations about Forward DNS Updating ......................13 69 6.1. Manual or Custom DNS Updates ..............................13 70 6.2. Dynamic DNS ...............................................13 71 7. Considerations about Reverse DNS Updating ......................14 72 7.1. Applicability of Reverse DNS ..............................14 73 7.2. Manual or Custom DNS Updates ..............................15 74 7.3. DDNS with Stateless Address Autoconfiguration .............16 75 7.4. DDNS with DHCP ............................................17 76 7.5. DDNS with Dynamic Prefix Delegation .......................17 77 8. Miscellaneous DNS Considerations ...............................18 78 8.1. NAT-PT with DNS-ALG .......................................18 79 8.2. Renumbering Procedures and Applications' Use of DNS .......18 80 9. Acknowledgements ...............................................19 81 10. Security Considerations .......................................19 82 11. References ....................................................20 83 11.1. Normative References .....................................20 84 11.2. Informative References ...................................22 85 Appendix A. Unique Local Addressing Considerations for DNS ........24 86 Appendix B. Behavior of Additional Data in IPv4/IPv6 87 Environments ..........................................24 88 B.1. Description of Additional Data Scenarios ..................24 89 B.2. Which Additional Data to Keep, If Any? ....................26 90 B.3. Discussion of the Potential Problems ......................27 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 Durand, et al. Informational [Page 2] 108 RFC 4472 Considerations with IPv6 DNS April 2006 109 110 111 1. Introduction 112 113 This memo presents operational considerations and issues with IPv6 114 DNS; it is meant to be an extensive summary and a list of pointers 115 for more information about IPv6 DNS considerations for those with 116 experience with IPv4 DNS. 117 118 The purpose of this document is to give information about various 119 issues and considerations related to DNS operations with IPv6; it is 120 not meant to be a normative specification or standard for IPv6 DNS. 121 122 The first section gives a brief overview of how IPv6 addresses and 123 names are represented in the DNS, how transport protocols and 124 resource records (don't) relate, and what IPv4/IPv6 name space 125 fragmentation means and how to avoid it; all of these are described 126 at more length in other documents. 127 128 The second section summarizes the special IPv6 address types and how 129 they relate to DNS. The third section describes observed DNS 130 implementation misbehaviors that have a varying effect on the use of 131 IPv6 records with DNS. The fourth section lists recommendations and 132 considerations for provisioning services with DNS. The fifth section 133 in turn looks at recommendations and considerations about providing 134 IPv6 support in the resolvers. The sixth and seventh sections 135 describe considerations with forward and reverse DNS updates, 136 respectively. The eighth section introduces several miscellaneous 137 IPv6 issues relating to DNS for which no better place has been found 138 in this memo. Appendix A looks briefly at the requirements for 139 unique local addressing. Appendix B discusses additional data. 140 141 1.1. Representing IPv6 Addresses in DNS Records 142 143 In the forward zones, IPv6 addresses are represented using AAAA 144 records. In the reverse zones, IPv6 address are represented using 145 PTR records in the nibble format under the ip6.arpa. tree. See 146 [RFC3596] for more about IPv6 DNS usage, and [RFC3363] or [RFC3152] 147 for background information. 148 149 In particular, one should note that the use of A6 records in the 150 forward tree or Bitlabels in the reverse tree is not recommended 151 [RFC3363]. Using DNAME records is not recommended in the reverse 152 tree in conjunction with A6 records; the document did not mean to 153 take a stance on any other use of DNAME records [RFC3364]. 154 155 156 157 158 159 160 161 162 Durand, et al. Informational [Page 3] 163 RFC 4472 Considerations with IPv6 DNS April 2006 164 165 166 1.2. Independence of DNS Transport and DNS Records 167 168 DNS has been designed to present a single, globally unique name space 169 [RFC2826]. This property should be maintained, as described here and 170 in Section 1.3. 171 172 The IP version used to transport the DNS queries and responses is 173 independent of the records being queried: AAAA records can be queried 174 over IPv4, and A records over IPv6. The DNS servers must not make 175 any assumptions about what data to return for Answer and Authority 176 sections based on the underlying transport used in a query. 177 178 However, there is some debate whether the addresses in Additional 179 section could be selected or filtered using hints obtained from which 180 transport was being used; this has some obvious problems because in 181 many cases the transport protocol does not correlate with the 182 requests, and because a "bad" answer is in a way worse than no answer 183 at all (consider the case where the client is led to believe that a 184 name received in the additional record does not have any AAAA records 185 at all). 186 187 As stated in [RFC3596]: 188 189 The IP protocol version used for querying resource records is 190 independent of the protocol version of the resource records; e.g., 191 IPv4 transport can be used to query IPv6 records and vice versa. 192 193 1.3. Avoiding IPv4/IPv6 Name Space Fragmentation 194 195 To avoid the DNS name space from fragmenting into parts where some 196 parts of DNS are only visible using IPv4 (or IPv6) transport, the 197 recommendation is to always keep at least one authoritative server 198 IPv4-enabled, and to ensure that recursive DNS servers support IPv4. 199 See DNS IPv6 transport guidelines [RFC3901] for more information. 200 201 1.4. Query Type '*' and A/AAAA Records 202 203 QTYPE=* is typically only used for debugging or management purposes; 204 it is worth keeping in mind that QTYPE=* ("ANY" queries) only return 205 any available RRsets, not *all* the RRsets, because the caches do not 206 necessarily have all the RRsets and have no way of guaranteeing that 207 they have all the RRsets. Therefore, to get both A and AAAA records 208 reliably, two separate queries must be made. 209 210 211 212 213 214 215 216 217 Durand, et al. Informational [Page 4] 218 RFC 4472 Considerations with IPv6 DNS April 2006 219 220 221 2. DNS Considerations about Special IPv6 Addresses 222 223 There are a couple of IPv6 address types that are somewhat special; 224 these are considered here. 225 226 2.1. Limited-Scope Addresses 227 228 The IPv6 addressing architecture [RFC4291] includes two kinds of 229 local-use addresses: link-local (fe80::/10) and site-local 230 (fec0::/10). The site-local addresses have been deprecated [RFC3879] 231 but are discussed with unique local addresses in Appendix A. 232 233 Link-local addresses should never be published in DNS (whether in 234 forward or reverse tree), because they have only local (to the 235 connected link) significance [WIP-DC2005]. 236 237 2.2. Temporary Addresses 238 239 Temporary addresses defined in RFC 3041 [RFC3041] (sometimes called 240 "privacy addresses") use a random number as the interface identifier. 241 Having DNS AAAA records that are updated to always contain the 242 current value of a node's temporary address would defeat the purpose 243 of the mechanism and is not recommended. However, it would still be 244 possible to return a non-identifiable name (e.g., the IPv6 address in 245 hexadecimal format), as described in [RFC3041]. 246 247 2.3. 6to4 Addresses 248 249 6to4 [RFC3056] specifies an automatic tunneling mechanism that maps a 250 public IPv4 address V4ADDR to an IPv6 prefix 2002:V4ADDR::/48. 251 252 If the reverse DNS population would be desirable (see Section 7.1 for 253 applicability), there are a number of possible ways to do so. 254 255 [WIP-H2005] aims to design an autonomous reverse-delegation system 256 that anyone being capable of communicating using a specific 6to4 257 address would be able to set up a reverse delegation to the 258 corresponding 6to4 prefix. This could be deployed by, e.g., Regional 259 Internet Registries (RIRs). This is a practical solution, but may 260 have some scalability concerns. 261 262 2.4. Other Transition Mechanisms 263 264 6to4 is mentioned as a case of an IPv6 transition mechanism requiring 265 special considerations. In general, mechanisms that include a 266 special prefix may need a custom solution; otherwise, for example, 267 when IPv4 address is embedded as the suffix or not embedded at all, 268 special solutions are likely not needed. 269 270 271 272 Durand, et al. Informational [Page 5] 273 RFC 4472 Considerations with IPv6 DNS April 2006 274 275 276 Note that it does not seem feasible to provide reverse DNS with 277 another automatic tunneling mechanism, Teredo [RFC4380]; this is 278 because the IPv6 address is based on the IPv4 address and UDP port of 279 the current Network Address Translation (NAT) mapping, which is 280 likely to be relatively short-lived. 281 282 3. Observed DNS Implementation Misbehavior 283 284 Several classes of misbehavior in DNS servers, load-balancers, and 285 resolvers have been observed. Most of these are rather generic, not 286 only applicable to IPv6 -- but in some cases, the consequences of 287 this misbehavior are extremely severe in IPv6 environments and 288 deserve to be mentioned. 289 290 3.1. Misbehavior of DNS Servers and Load-balancers 291 292 There are several classes of misbehavior in certain DNS servers and 293 load-balancers that have been noticed and documented [RFC4074]: some 294 implementations silently drop queries for unimplemented DNS records 295 types, or provide wrong answers to such queries (instead of a proper 296 negative reply). While typically these issues are not limited to 297 AAAA records, the problems are aggravated by the fact that AAAA 298 records are being queried instead of (mainly) A records. 299 300 The problems are serious because when looking up a DNS name, typical 301 getaddrinfo() implementations, with AF_UNSPEC hint given, first try 302 to query the AAAA records of the name, and after receiving a 303 response, query the A records. This is done in a serial fashion -- 304 if the first query is never responded to (instead of properly 305 returning a negative answer), significant time-outs will occur. 306 307 In consequence, this is an enormous problem for IPv6 deployments, and 308 in some cases, IPv6 support in the software has even been disabled 309 due to these problems. 310 311 The solution is to fix or retire those misbehaving implementations, 312 but that is likely not going to be effective. There are some 313 possible ways to mitigate the problem, e.g., by performing the 314 lookups somewhat in parallel and reducing the time-out as long as at 315 least one answer has been received, but such methods remain to be 316 investigated; slightly more on this is included in Section 5. 317 318 3.2. Misbehavior of DNS Resolvers 319 320 Several classes of misbehavior have also been noticed in DNS 321 resolvers [WIP-LB2005]. However, these do not seem to directly 322 impair IPv6 use, and are only referred to for completeness. 323 324 325 326 327 Durand, et al. Informational [Page 6] 328 RFC 4472 Considerations with IPv6 DNS April 2006 329 330 331 4. Recommendations for Service Provisioning Using DNS 332 333 When names are added in the DNS to facilitate a service, there are 334 several general guidelines to consider to be able to do it as 335 smoothly as possible. 336 337 4.1. Use of Service Names instead of Node Names 338 339 It makes sense to keep information about separate services logically 340 separate in the DNS by using a different DNS hostname for each 341 service. There are several reasons for doing this, for example: 342 343 o It allows more flexibility and ease for migration of (only a part 344 of) services from one node to another, 345 346 o It allows configuring different properties (e.g., Time to Live 347 (TTL)) for each service, and 348 349 o It allows deciding separately for each service whether or not to 350 publish the IPv6 addresses (in cases where some services are more 351 IPv6-ready than others). 352 353 Using SRV records [RFC2782] would avoid these problems. 354 Unfortunately, those are not sufficiently widely used to be 355 applicable in most cases. Hence an operation technique is to use 356 service names instead of node names (or "hostnames"). This 357 operational technique is not specific to IPv6, but required to 358 understand the considerations described in Section 4.2 and 359 Section 4.3. 360 361 For example, assume a node named "pobox.example.com" provides both 362 SMTP and IMAP service. Instead of configuring the MX records to 363 point at "pobox.example.com", and configuring the mail clients to 364 look up the mail via IMAP from "pobox.example.com", one could use, 365 e.g., "smtp.example.com" for SMTP (for both message submission and 366 mail relaying between SMTP servers) and "imap.example.com" for IMAP. 367 Note that in the specific case of SMTP relaying, the server itself 368 must typically also be configured to know all its names to ensure 369 that loops do not occur. DNS can provide a layer of indirection 370 between service names and where the service actually is, and using 371 which addresses. (Obviously, when wanting to reach a specific node, 372 one should use the hostname rather than a service name.) 373 374 375 376 377 378 379 380 381 382 Durand, et al. Informational [Page 7] 383 RFC 4472 Considerations with IPv6 DNS April 2006 384 385 386 4.2. Separate vs. the Same Service Names for IPv4 and IPv6 387 388 The service naming can be achieved in basically two ways: when a 389 service is named "service.example.com" for IPv4, the IPv6-enabled 390 service could either be added to "service.example.com" or added 391 separately under a different name, e.g., in a sub-domain like 392 "service.ipv6.example.com". 393 394 These two methods have different characteristics. Using a different 395 name allows for easier service piloting, minimizing the disturbance 396 to the "regular" users of IPv4 service; however, the service would 397 not be used transparently, without the user/application explicitly 398 finding it and asking for it -- which would be a disadvantage in most 399 cases. When the different name is under a sub-domain, if the 400 services are deployed within a restricted network (e.g., inside an 401 enterprise), it's possible to prefer them transparently, at least to 402 a degree, by modifying the DNS search path; however, this is a 403 suboptimal solution. Using the same service name is the "long-term" 404 solution, but may degrade performance for those clients whose IPv6 405 performance is lower than IPv4, or does not work as well (see 406 Section 4.3 for more). 407 408 In most cases, it makes sense to pilot or test a service using 409 separate service names, and move to the use of the same name when 410 confident enough that the service level will not degrade for the 411 users unaware of IPv6. 412 413 4.3. Adding the Records Only When Fully IPv6-enabled 414 415 The recommendation is that AAAA records for a service should not be 416 added to the DNS until all of following are true: 417 418 1. The address is assigned to the interface on the node. 419 420 2. The address is configured on the interface. 421 422 3. The interface is on a link that is connected to the IPv6 423 infrastructure. 424 425 In addition, if the AAAA record is added for the node, instead of 426 service as recommended, all the services of the node should be IPv6- 427 enabled prior to adding the resource record. 428 429 For example, if an IPv6 node is isolated from an IPv6 perspective 430 (e.g., it is not connected to IPv6 Internet) constraint #3 would mean 431 that it should not have an address in the DNS. 432 433 434 435 436 437 Durand, et al. Informational [Page 8] 438 RFC 4472 Considerations with IPv6 DNS April 2006 439 440 441 Consider the case of two dual-stack nodes, which both are IPv6- 442 enabled, but the server does not have (global) IPv6 connectivity. As 443 the client looks up the server's name, only A records are returned 444 (if the recommendations above are followed), and no IPv6 445 communication, which would have been unsuccessful, is even attempted. 446 447 The issues are not always so black-and-white. Usually, it's 448 important that the service offered using both protocols is of roughly 449 equal quality, using the appropriate metrics for the service (e.g., 450 latency, throughput, low packet loss, general reliability, etc.). 451 This is typically very important especially for interactive or real- 452 time services. In many cases, the quality of IPv6 connectivity may 453 not yet be equal to that of IPv4, at least globally; this has to be 454 taken into consideration when enabling services. 455 456 4.4. The Use of TTL for IPv4 and IPv6 RRs 457 458 The behavior of DNS caching when different TTL values are used for 459 different RRsets of the same name calls for explicit discussion. For 460 example, let's consider two unrelated zone fragments: 461 462 example.com. 300 IN MX foo.example.com. 463 foo.example.com. 300 IN A 192.0.2.1 464 foo.example.com. 100 IN AAAA 2001:db8::1 465 466 ... 467 468 child.example.com. 300 IN NS ns.child.example.com. 469 ns.child.example.com. 300 IN A 192.0.2.1 470 ns.child.example.com. 100 IN AAAA 2001:db8::1 471 472 In the former case, we have "courtesy" additional data; in the 473 latter, we have "critical" additional data. See more extensive 474 background discussion of additional data handling in Appendix B. 475 476 4.4.1. TTL with Courtesy Additional Data 477 478 When a caching resolver asks for the MX record of example.com, it 479 gets back "foo.example.com". It may also get back either one or both 480 of the A and AAAA records in the additional section. The resolver 481 must explicitly query for both A and AAAA records [RFC2821]. 482 483 After 100 seconds, the AAAA record is removed from the cache(s) 484 because its TTL expired. It could be argued to be useful for the 485 caching resolvers to discard the A record when the shorter TTL (in 486 this case, for the AAAA record) expires; this would avoid the 487 situation where there would be a window of 200 seconds when 488 incomplete information is returned from the cache. Further argument 489 490 491 492 Durand, et al. Informational [Page 9] 493 RFC 4472 Considerations with IPv6 DNS April 2006 494 495 496 for discarding is that in the normal operation, the TTL values are so 497 high that very likely the incurred additional queries would not be 498 noticeable, compared to the obtained performance optimization. The 499 behavior in this scenario is unspecified. 500 501 4.4.2. TTL with Critical Additional Data 502 503 The difference to courtesy additional data is that the A/AAAA records 504 served by the parent zone cannot be queried explicitly. Therefore, 505 after 100 seconds the AAAA record is removed from the cache(s), but 506 the A record remains. Queries for the remaining 200 seconds 507 (provided that there are no further queries from the parent that 508 could refresh the caches) only return the A record, leading to a 509 potential operational situation with unreachable servers. 510 511 Similar cache flushing strategies apply in this scenario; the 512 behavior is likewise unspecified. 513 514 4.5. IPv6 Transport Guidelines for DNS Servers 515 516 As described in Section 1.3 and [RFC3901], there should continue to 517 be at least one authoritative IPv4 DNS server for every zone, even if 518 the zone has only IPv6 records. (Note that obviously, having more 519 servers with robust connectivity would be preferable, but this is the 520 minimum recommendation; also see [RFC2182].) 521 522 5. Recommendations for DNS Resolver IPv6 Support 523 524 When IPv6 is enabled on a node, there are several things to consider 525 to ensure that the process is as smooth as possible. 526 527 5.1. DNS Lookups May Query IPv6 Records Prematurely 528 529 The system library that implements the getaddrinfo() function for 530 looking up names is a critical piece when considering the robustness 531 of enabling IPv6; it may come in basically three flavors: 532 533 1. The system library does not know whether IPv6 has been enabled in 534 the kernel of the operating system: it may start looking up AAAA 535 records with getaddrinfo() and AF_UNSPEC hint when the system is 536 upgraded to a system library version that supports IPv6. 537 538 2. The system library might start to perform IPv6 queries with 539 getaddrinfo() only when IPv6 has been enabled in the kernel. 540 However, this does not guarantee that there exists any useful 541 IPv6 connectivity (e.g., the node could be isolated from the 542 other IPv6 networks, only having link-local addresses). 543 544 545 546 547 Durand, et al. Informational [Page 10] 548 RFC 4472 Considerations with IPv6 DNS April 2006 549 550 551 3. The system library might implement a toggle that would apply some 552 heuristics to the "IPv6-readiness" of the node before starting to 553 perform queries; for example, it could check whether only link- 554 local IPv6 address(es) exists, or if at least one global IPv6 555 address exists. 556 557 First, let us consider generic implications of unnecessary queries 558 for AAAA records: when looking up all the records in the DNS, AAAA 559 records are typically tried first, and then A records. These are 560 done in serial, and the A query is not performed until a response is 561 received to the AAAA query. Considering the misbehavior of DNS 562 servers and load-balancers, as described in Section 3.1, the lookup 563 delay for AAAA may incur additional unnecessary latency, and 564 introduce a component of unreliability. 565 566 One option here could be to do the queries partially in parallel; for 567 example, if the final response to the AAAA query is not received in 568 0.5 seconds, start performing the A query while waiting for the 569 result. (Immediate parallelism might not be optimal, at least 570 without information-sharing between the lookup threads, as that would 571 probably lead to duplicate non-cached delegation chain lookups.) 572 573 An additional concern is the address selection, which may, in some 574 circumstances, prefer AAAA records over A records even when the node 575 does not have any IPv6 connectivity [WIP-RDP2004]. In some cases, 576 the implementation may attempt to connect or send a datagram on a 577 physical link [WIP-R2006], incurring very long protocol time-outs, 578 instead of quickly falling back to IPv4. 579 580 Now, we can consider the issues specific to each of the three 581 possibilities: 582 583 In the first case, the node performs a number of completely useless 584 DNS lookups as it will not be able to use the returned AAAA records 585 anyway. (The only exception is where the application desires to know 586 what's in the DNS, but not use the result for communication.) One 587 should be able to disable these unnecessary queries, for both latency 588 and reliability reasons. However, as IPv6 has not been enabled, the 589 connections to IPv6 addresses fail immediately, and if the 590 application is programmed properly, the application can fall 591 gracefully back to IPv4 [RFC4038]. 592 593 The second case is similar to the first, except it happens to a 594 smaller set of nodes when IPv6 has been enabled but connectivity has 595 not been provided yet. Similar considerations apply, with the 596 exception that IPv6 records, when returned, will be actually tried 597 first, which may typically lead to long time-outs. 598 599 600 601 602 Durand, et al. Informational [Page 11] 603 RFC 4472 Considerations with IPv6 DNS April 2006 604 605 606 The third case is a bit more complex: optimizing away the DNS lookups 607 with only link-locals is probably safe (but may be desirable with 608 different lookup services that getaddrinfo() may support), as the 609 link-locals are typically automatically generated when IPv6 is 610 enabled, and do not indicate any form of IPv6 connectivity. That is, 611 performing DNS lookups only when a non-link-local address has been 612 configured on any interface could be beneficial -- this would be an 613 indication that the address has been configured either from a router 614 advertisement, Dynamic Host Configuration Protocol for IPv6 (DHCPv6) 615 [RFC3315], or manually. Each would indicate at least some form of 616 IPv6 connectivity, even though there would not be guarantees of it. 617 618 These issues should be analyzed at more depth, and the fixes found 619 consensus on, perhaps in a separate document. 620 621 5.2. Obtaining a List of DNS Recursive Resolvers 622 623 In scenarios where DHCPv6 is available, a host can discover a list of 624 DNS recursive resolvers through the DHCPv6 "DNS Recursive Name 625 Server" option [RFC3646]. This option can be passed to a host 626 through a subset of DHCPv6 [RFC3736]. 627 628 The IETF is considering the development of alternative mechanisms for 629 obtaining the list of DNS recursive name servers when DHCPv6 is 630 unavailable or inappropriate. No decision about taking on this 631 development work has been reached as of this writing [RFC4339]. 632 633 In scenarios where DHCPv6 is unavailable or inappropriate, mechanisms 634 under consideration for development include the use of [WIP-O2004] 635 and the use of Router Advertisements to convey the information 636 [WIP-J2006]. 637 638 Note that even though IPv6 DNS resolver discovery is a recommended 639 procedure, it is not required for dual-stack nodes in dual-stack 640 networks as IPv6 DNS records can be queried over IPv4 as well as 641 IPv6. Obviously, nodes that are meant to function without manual 642 configuration in IPv6-only networks must implement the DNS resolver 643 discovery function. 644 645 5.3. IPv6 Transport Guidelines for Resolvers 646 647 As described in Section 1.3 and [RFC3901], the recursive resolvers 648 should be IPv4-only or dual-stack to be able to reach any IPv4-only 649 DNS server. Note that this requirement is also fulfilled by an IPv6- 650 only stub resolver pointing to a dual-stack recursive DNS resolver. 651 652 653 654 655 656 657 Durand, et al. Informational [Page 12] 658 RFC 4472 Considerations with IPv6 DNS April 2006 659 660 661 6. Considerations about Forward DNS Updating 662 663 While the topic of how to enable updating the forward DNS, i.e., the 664 mapping from names to the correct new addresses, is not specific to 665 IPv6, it should be considered especially due to the advent of 666 Stateless Address Autoconfiguration [RFC2462]. 667 668 Typically, forward DNS updates are more manageable than doing them in 669 the reverse DNS, because the updater can often be assumed to "own" a 670 certain DNS name -- and we can create a form of security relationship 671 with the DNS name and the node that is allowed to update it to point 672 to a new address. 673 674 A more complex form of DNS updates -- adding a whole new name into a 675 DNS zone, instead of updating an existing name -- is considered out 676 of scope for this memo as it could require zone-wide authentication. 677 Adding a new name in the forward zone is a problem that is still 678 being explored with IPv4, and IPv6 does not seem to add much new in 679 that area. 680 681 6.1. Manual or Custom DNS Updates 682 683 The DNS mappings can also be maintained by hand, in a semi-automatic 684 fashion or by running non-standardized protocols. These are not 685 considered at more length in this memo. 686 687 6.2. Dynamic DNS 688 689 Dynamic DNS updates (DDNS) [RFC2136] [RFC3007] is a standardized 690 mechanism for dynamically updating the DNS. It works equally well 691 with Stateless Address Autoconfiguration (SLAAC), DHCPv6, or manual 692 address configuration. It is important to consider how each of these 693 behave if IP address-based authentication, instead of stronger 694 mechanisms [RFC3007], was used in the updates. 695 696 1. Manual addresses are static and can be configured. 697 698 2. DHCPv6 addresses could be reasonably static or dynamic, depending 699 on the deployment, and could or could not be configured on the 700 DNS server for the long term. 701 702 3. SLAAC addresses are typically stable for a long time, but could 703 require work to be configured and maintained. 704 705 As relying on IP addresses for Dynamic DNS is rather insecure at 706 best, stronger authentication should always be used; however, this 707 requires that the authorization keying will be explicitly configured 708 using unspecified operational methods. 709 710 711 712 Durand, et al. Informational [Page 13] 713 RFC 4472 Considerations with IPv6 DNS April 2006 714 715 716 Note that with DHCP it is also possible that the DHCP server updates 717 the DNS, not the host. The host might only indicate in the DHCP 718 exchange which hostname it would prefer, and the DHCP server would 719 make the appropriate updates. Nonetheless, while this makes setting 720 up a secure channel between the updater and the DNS server easier, it 721 does not help much with "content" security, i.e., whether the 722 hostname was acceptable -- if the DNS server does not include 723 policies, they must be included in the DHCP server (e.g., a regular 724 host should not be able to state that its name is "www.example.com"). 725 DHCP-initiated DDNS updates have been extensively described in 726 [WIP-SV2005], [WIP-S2005a], and [WIP-S2005b]. 727 728 The nodes must somehow be configured with the information about the 729 servers where they will attempt to update their addresses, sufficient 730 security material for authenticating themselves to the server, and 731 the hostname they will be updating. Unless otherwise configured, the 732 first could be obtained by looking up the authoritative name servers 733 for the hostname; the second must be configured explicitly unless one 734 chooses to trust the IP address-based authentication (not a good 735 idea); and lastly, the nodename is typically pre-configured somehow 736 on the node, e.g., at install time. 737 738 Care should be observed when updating the addresses not to use longer 739 TTLs for addresses than are preferred lifetimes for the addresses, so 740 that if the node is renumbered in a managed fashion, the amount of 741 stale DNS information is kept to the minimum. That is, if the 742 preferred lifetime of an address expires, the TTL of the record needs 743 to be modified unless it was already done before the expiration. For 744 better flexibility, the DNS TTL should be much shorter (e.g., a half 745 or a third) than the lifetime of an address; that way, the node can 746 start lowering the DNS TTL if it seems like the address has not been 747 renewed/refreshed in a while. Some discussion on how an 748 administrator could manage the DNS TTL is included in [RFC4192]; this 749 could be applied to (smart) hosts as well. 750 751 7. Considerations about Reverse DNS Updating 752 753 Updating the reverse DNS zone may be difficult because of the split 754 authority over an address. However, first we have to consider the 755 applicability of reverse DNS in the first place. 756 757 7.1. Applicability of Reverse DNS 758 759 Today, some applications use reverse DNS either to look up some hints 760 about the topological information associated with an address (e.g., 761 resolving web server access logs) or (as a weak form of a security 762 check) to get a feel whether the user's network administrator has 763 764 765 766 767 Durand, et al. Informational [Page 14] 768 RFC 4472 Considerations with IPv6 DNS April 2006 769 770 771 "authorized" the use of the address (on the premise that adding a 772 reverse record for an address would signal some form of 773 authorization). 774 775 One additional, maybe slightly more useful usage is ensuring that the 776 reverse and forward DNS contents match (by looking up the pointer to 777 the name by the IP address from the reverse tree, and ensuring that a 778 record under the name in the forward tree points to the IP address) 779 and correspond to a configured name or domain. As a security check, 780 it is typically accompanied by other mechanisms, such as a user/ 781 password login; the main purpose of the reverse+forward DNS check is 782 to weed out the majority of unauthorized users, and if someone 783 managed to bypass the checks, he would still need to authenticate 784 "properly". 785 786 It may also be desirable to store IPsec keying material corresponding 787 to an IP address in the reverse DNS, as justified and described in 788 [RFC4025]. 789 790 It is not clear whether it makes sense to require or recommend that 791 reverse DNS records be updated. In many cases, it would just make 792 more sense to use proper mechanisms for security (or topological 793 information lookup) in the first place. At minimum, the applications 794 that use it as a generic authorization (in the sense that a record 795 exists at all) should be modified as soon as possible to avoid such 796 lookups completely. 797 798 The applicability is discussed at more length in [WIP-S2005c]. 799 800 7.2. Manual or Custom DNS Updates 801 802 Reverse DNS can of course be updated using manual or custom methods. 803 These are not further described here, except for one special case. 804 805 One way to deploy reverse DNS would be to use wildcard records, for 806 example, by configuring one name for a subnet (/64) or a site (/48). 807 As a concrete example, a site (or the site's ISP) could configure the 808 reverses of the prefix 2001:db8:f00::/48 to point to one name using a 809 wildcard record like "*.0.0.f.0.8.b.d.0.1.0.0.2.ip6.arpa. IN PTR 810 site.example.com.". Naturally, such a name could not be verified 811 from the forward DNS, but would at least provide some form of 812 "topological information" or "weak authorization" if that is really 813 considered to be useful. Note that this is not actually updating the 814 DNS as such, as the whole point is to avoid DNS updates completely by 815 manually configuring a generic name. 816 817 818 819 820 821 822 Durand, et al. Informational [Page 15] 823 RFC 4472 Considerations with IPv6 DNS April 2006 824 825 826 7.3. DDNS with Stateless Address Autoconfiguration 827 828 Dynamic reverse DNS with SLAAC is simpler than forward DNS updates in 829 some regard, while being more difficult in another, as described 830 below. 831 832 The address space administrator decides whether or not the hosts are 833 trusted to update their reverse DNS records. If they are trusted and 834 deployed at the same site (e.g., not across the Internet), a simple 835 address-based authorization is typically sufficient (i.e., check that 836 the DNS update is done from the same IP address as the record being 837 updated); stronger security can also be used [RFC3007]. If they 838 aren't allowed to update the reverses, no update can occur. However, 839 such address-based update authorization operationally requires that 840 ingress filtering [RFC3704] has been set up at the border of the site 841 where the updates occur, and as close to the updater as possible. 842 843 Address-based authorization is simpler with reverse DNS (as there is 844 a connection between the record and the address) than with forward 845 DNS. However, when a stronger form of security is used, forward DNS 846 updates are simpler to manage because the host can be assumed to have 847 an association with the domain. Note that the user may roam to 848 different networks and does not necessarily have any association with 849 the owner of that address space. So, assuming a stronger form of 850 authorization for reverse DNS updates than an address association is 851 generally infeasible. 852 853 Moreover, the reverse zones must be cleaned up by an unspecified 854 janitorial process: the node does not typically know a priori that it 855 will be disconnected, and it cannot send a DNS update using the 856 correct source address to remove a record. 857 858 A problem with defining the clean-up process is that it is difficult 859 to ensure that a specific IP address and the corresponding record are 860 no longer being used. Considering the huge address space, and the 861 unlikelihood of collision within 64 bits of the interface 862 identifiers, a process that would remove the record after no traffic 863 has been seen from a node in a long period of time (e.g., a month or 864 year) might be one possible approach. 865 866 To insert or update the record, the node must discover the DNS server 867 to send the update to somehow, similar to as discussed in 868 Section 6.2. One way to automate this is looking up the DNS server 869 authoritative (e.g., through SOA record) for the IP address being 870 updated, but the security material (unless the IP address-based 871 authorization is trusted) must also be established by some other 872 means. 873 874 875 876 877 Durand, et al. Informational [Page 16] 878 RFC 4472 Considerations with IPv6 DNS April 2006 879 880 881 One should note that Cryptographically Generated Addresses (CGAs) 882 [RFC3972] may require a slightly different kind of treatment. CGAs 883 are addresses where the interface identifier is calculated from a 884 public key, a modifier (used as a nonce), the subnet prefix, and 885 other data. Depending on the usage profile, CGAs might or might not 886 be changed periodically due to, e.g., privacy reasons. As the CGA 887 address is not predictable, a reverse record can only reasonably be 888 inserted in the DNS by the node that generates the address. 889 890 7.4. DDNS with DHCP 891 892 With DHCPv4, the reverse DNS name is typically already inserted to 893 the DNS that reflects the name (e.g., "dhcp-67.example.com"). One 894 can assume similar practice may become commonplace with DHCPv6 as 895 well; all such mappings would be pre-configured and would require no 896 updating. 897 898 If a more explicit control is required, similar considerations as 899 with SLAAC apply, except for the fact that typically one must update 900 a reverse DNS record instead of inserting one (if an address 901 assignment policy that reassigns disused addresses is adopted) and 902 updating a record seems like a slightly more difficult thing to 903 secure. However, it is yet uncertain how DHCPv6 is going to be used 904 for address assignment. 905 906 Note that when using DHCP, either the host or the DHCP server could 907 perform the DNS updates; see the implications in Section 6.2. 908 909 If disused addresses were to be reassigned, host-based DDNS reverse 910 updates would need policy considerations for DNS record modification, 911 as noted above. On the other hand, if disused address were not to be 912 assigned, host-based DNS reverse updates would have similar 913 considerations as SLAAC in Section 7.3. Server-based updates have 914 similar properties except that the janitorial process could be 915 integrated with DHCP address assignment. 916 917 7.5. DDNS with Dynamic Prefix Delegation 918 919 In cases where a prefix, instead of an address, is being used and 920 updated, one should consider what is the location of the server where 921 DDNS updates are made. That is, where the DNS server is located: 922 923 1. At the same organization as the prefix delegator. 924 925 2. At the site where the prefixes are delegated to. In this case, 926 the authority of the DNS reverse zone corresponding to the 927 delegated prefix is also delegated to the site. 928 929 930 931 932 Durand, et al. Informational [Page 17] 933 RFC 4472 Considerations with IPv6 DNS April 2006 934 935 936 3. Elsewhere; this implies a relationship between the site and where 937 the DNS server is located, and such a relationship should be 938 rather straightforward to secure as well. Like in the previous 939 case, the authority of the DNS reverse zone is also delegated. 940 941 In the first case, managing the reverse DNS (delegation) is simpler 942 as the DNS server and the prefix delegator are in the same 943 administrative domain (as there is no need to delegate anything at 944 all); alternatively, the prefix delegator might forgo DDNS reverse 945 capability altogether, and use, e.g., wildcard records (as described 946 in Section 7.2). In the other cases, it can be slightly more 947 difficult, particularly as the site will have to configure the DNS 948 server to be authoritative for the delegated reverse zone, implying 949 automatic configuration of the DNS server -- as the prefix may be 950 dynamic. 951 952 Managing the DDNS reverse updates is typically simple in the second 953 case, as the updated server is located at the local site, and 954 arguably IP address-based authentication could be sufficient (or if 955 not, setting up security relationships would be simpler). As there 956 is an explicit (security) relationship between the parties in the 957 third case, setting up the security relationships to allow reverse 958 DDNS updates should be rather straightforward as well (but IP 959 address-based authentication might not be acceptable). In the first 960 case, however, setting up and managing such relationships might be a 961 lot more difficult. 962 963 8. Miscellaneous DNS Considerations 964 965 This section describes miscellaneous considerations about DNS that 966 seem related to IPv6, for which no better place has been found in 967 this document. 968 969 8.1. NAT-PT with DNS-ALG 970 971 The DNS-ALG component of NAT-PT [RFC2766] mangles A records to look 972 like AAAA records to the IPv6-only nodes. Numerous problems have 973 been identified with [WIP-AD2005]. This is a strong reason not to 974 use NAT-PT in the first place. 975 976 8.2. Renumbering Procedures and Applications' Use of DNS 977 978 One of the most difficult problems of systematic IP address 979 renumbering procedures [RFC4192] is that an application that looks up 980 a DNS name disregards information such as TTL, and uses the result 981 obtained from DNS as long as it happens to be stored in the memory of 982 the application. For applications that run for a long time, this 983 984 985 986 987 Durand, et al. Informational [Page 18] 988 RFC 4472 Considerations with IPv6 DNS April 2006 989 990 991 could be days, weeks, or even months. Some applications may be 992 clever enough to organize the data structures and functions in such a 993 manner that lookups get refreshed now and then. 994 995 While the issue appears to have a clear solution, "fix the 996 applications", practically, this is not reasonable immediate advice. 997 The TTL information is not typically available in the APIs and 998 libraries (so, the advice becomes "fix the applications, APIs, and 999 libraries"), and a lot more analysis is needed on how to practically 1000 go about to achieve the ultimate goal of avoiding using the names 1001 longer than expected. 1002 1003 9. Acknowledgements 1004 1005 Some recommendations (Section 4.3, Section 5.1) about IPv6 service 1006 provisioning were moved here from [RFC4213] by Erik Nordmark and Bob 1007 Gilligan. Havard Eidnes and Michael Patton provided useful feedback 1008 and improvements. Scott Rose, Rob Austein, Masataka Ohta, and Mark 1009 Andrews helped in clarifying the issues regarding additional data and 1010 the use of TTL. Jefsey Morfin, Ralph Droms, Peter Koch, Jinmei 1011 Tatuya, Iljitsch van Beijnum, Edward Lewis, and Rob Austein provided 1012 useful feedback during the WG last call. Thomas Narten provided 1013 extensive feedback during the IESG evaluation. 1014 1015 10. Security Considerations 1016 1017 This document reviews the operational procedures for IPv6 DNS 1018 operations and does not have security considerations in itself. 1019 1020 However, it is worth noting that in particular with Dynamic DNS 1021 updates, security models based on the source address validation are 1022 very weak and cannot be recommended -- they could only be considered 1023 in the environments where ingress filtering [RFC3704] has been 1024 deployed. On the other hand, it should be noted that setting up an 1025 authorization mechanism (e.g., a shared secret, or public-private 1026 keys) between a node and the DNS server has to be done manually, and 1027 may require quite a bit of time and expertise. 1028 1029 To re-emphasize what was already stated, the reverse+forward DNS 1030 check provides very weak security at best, and the only 1031 (questionable) security-related use for them may be in conjunction 1032 with other mechanisms when authenticating a user. 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 Durand, et al. Informational [Page 19] 1043 RFC 4472 Considerations with IPv6 DNS April 2006 1044 1045 1046 11. References 1047 1048 11.1. Normative References 1049 1050 [RFC1034] Mockapetris, P., "Domain names - concepts and 1051 facilities", STD 13, RFC 1034, November 1987. 1052 1053 [RFC2136] Vixie, P., Thomson, S., Rekhter, Y., and J. Bound, 1054 "Dynamic Updates in the Domain Name System (DNS 1055 UPDATE)", RFC 2136, April 1997. 1056 1057 [RFC2181] Elz, R. and R. Bush, "Clarifications to the DNS 1058 Specification", RFC 2181, July 1997. 1059 1060 [RFC2182] Elz, R., Bush, R., Bradner, S., and M. Patton, 1061 "Selection and Operation of Secondary DNS Servers", 1062 BCP 16, RFC 2182, July 1997. 1063 1064 [RFC2462] Thomson, S. and T. Narten, "IPv6 Stateless Address 1065 Autoconfiguration", RFC 2462, December 1998. 1066 1067 [RFC2671] Vixie, P., "Extension Mechanisms for DNS (EDNS0)", 1068 RFC 2671, August 1999. 1069 1070 [RFC2821] Klensin, J., "Simple Mail Transfer Protocol", RFC 2821, 1071 April 2001. 1072 1073 [RFC3007] Wellington, B., "Secure Domain Name System (DNS) 1074 Dynamic Update", RFC 3007, November 2000. 1075 1076 [RFC3041] Narten, T. and R. Draves, "Privacy Extensions for 1077 Stateless Address Autoconfiguration in IPv6", RFC 3041, 1078 January 2001. 1079 1080 [RFC3056] Carpenter, B. and K. Moore, "Connection of IPv6 Domains 1081 via IPv4 Clouds", RFC 3056, February 2001. 1082 1083 [RFC3152] Bush, R., "Delegation of IP6.ARPA", BCP 49, RFC 3152, 1084 August 2001. 1085 1086 [RFC3315] Droms, R., Bound, J., Volz, B., Lemon, T., Perkins, C., 1087 and M. Carney, "Dynamic Host Configuration Protocol for 1088 IPv6 (DHCPv6)", RFC 3315, July 2003. 1089 1090 [RFC3363] Bush, R., Durand, A., Fink, B., Gudmundsson, O., and T. 1091 Hain, "Representing Internet Protocol version 6 (IPv6) 1092 Addresses in the Domain Name System (DNS)", RFC 3363, 1093 August 2002. 1094 1095 1096 1097 Durand, et al. Informational [Page 20] 1098 RFC 4472 Considerations with IPv6 DNS April 2006 1099 1100 1101 [RFC3364] Austein, R., "Tradeoffs in Domain Name System (DNS) 1102 Support for Internet Protocol version 6 (IPv6)", 1103 RFC 3364, August 2002. 1104 1105 [RFC3596] Thomson, S., Huitema, C., Ksinant, V., and M. Souissi, 1106 "DNS Extensions to Support IP Version 6", RFC 3596, 1107 October 2003. 1108 1109 [RFC3646] Droms, R., "DNS Configuration options for Dynamic Host 1110 Configuration Protocol for IPv6 (DHCPv6)", RFC 3646, 1111 December 2003. 1112 1113 [RFC3736] Droms, R., "Stateless Dynamic Host Configuration 1114 Protocol (DHCP) Service for IPv6", RFC 3736, 1115 April 2004. 1116 1117 [RFC3879] Huitema, C. and B. Carpenter, "Deprecating Site Local 1118 Addresses", RFC 3879, September 2004. 1119 1120 [RFC3901] Durand, A. and J. Ihren, "DNS IPv6 Transport 1121 Operational Guidelines", BCP 91, RFC 3901, 1122 September 2004. 1123 1124 [RFC4038] Shin, M-K., Hong, Y-G., Hagino, J., Savola, P., and E. 1125 Castro, "Application Aspects of IPv6 Transition", 1126 RFC 4038, March 2005. 1127 1128 [RFC4074] Morishita, Y. and T. Jinmei, "Common Misbehavior 1129 Against DNS Queries for IPv6 Addresses", RFC 4074, 1130 May 2005. 1131 1132 [RFC4192] Baker, F., Lear, E., and R. Droms, "Procedures for 1133 Renumbering an IPv6 Network without a Flag Day", 1134 RFC 4192, September 2005. 1135 1136 [RFC4193] Hinden, R. and B. Haberman, "Unique Local IPv6 Unicast 1137 Addresses", RFC 4193, October 2005. 1138 1139 [RFC4291] Hinden, R. and S. Deering, "IP Version 6 Addressing 1140 Architecture", RFC 4291, February 2006. 1141 1142 [RFC4339] Jeong, J., Ed., "IPv6 Host Configuration of DNS Server 1143 Information Approaches", RFC 4339, February 2006. 1144 1145 1146 1147 1148 1149 1150 1151 1152 Durand, et al. Informational [Page 21] 1153 RFC 4472 Considerations with IPv6 DNS April 2006 1154 1155 1156 11.2. Informative References 1157 1158 [RFC2766] Tsirtsis, G. and P. Srisuresh, "Network Address 1159 Translation - Protocol Translation (NAT-PT)", RFC 2766, 1160 February 2000. 1161 1162 [RFC2782] Gulbrandsen, A., Vixie, P., and L. Esibov, "A DNS RR 1163 for specifying the location of services (DNS SRV)", 1164 RFC 2782, February 2000. 1165 1166 [RFC2826] Internet Architecture Board, "IAB Technical Comment on 1167 the Unique DNS Root", RFC 2826, May 2000. 1168 1169 [RFC3704] Baker, F. and P. Savola, "Ingress Filtering for 1170 Multihomed Networks", BCP 84, RFC 3704, March 2004. 1171 1172 [RFC3972] Aura, T., "Cryptographically Generated Addresses 1173 (CGA)", RFC 3972, March 2005. 1174 1175 [RFC4025] Richardson, M., "A Method for Storing IPsec Keying 1176 Material in DNS", RFC 4025, March 2005. 1177 1178 [RFC4213] Nordmark, E. and R. Gilligan, "Basic Transition 1179 Mechanisms for IPv6 Hosts and Routers", RFC 4213, 1180 October 2005. 1181 1182 [RFC4215] Wiljakka, J., "Analysis on IPv6 Transition in Third 1183 Generation Partnership Project (3GPP) Networks", 1184 RFC 4215, October 2005. 1185 1186 [RFC4380] Huitema, C., "Teredo: Tunneling IPv6 over UDP through 1187 Network Address Translations (NATs)", RFC 4380, 1188 February 2006. 1189 1190 [TC-TEST] Jinmei, T., "Thread "RFC2181 section 9.1: TC bit 1191 handling and additional data" on DNSEXT mailing list, 1192 Message- 1193 Id:firstname.lastname@example.org", August 1194 1, 2005, <http://ops.ietf.org/lists/namedroppers/ 1195 namedroppers.2005/msg01102.html>. 1196 1197 [WIP-AD2005] Aoun, C. and E. Davies, "Reasons to Move NAT-PT to 1198 Experimental", Work in Progress, October 2005. 1199 1200 [WIP-DC2005] Durand, A. and T. Chown, "To publish, or not to 1201 publish, that is the question", Work in Progress, 1202 October 2005. 1203 1204 1205 1206 1207 Durand, et al. Informational [Page 22] 1208 RFC 4472 Considerations with IPv6 DNS April 2006 1209 1210 1211 [WIP-H2005] Huston, G., "6to4 Reverse DNS Delegation 1212 Specification", Work in Progress, November 2005. 1213 1214 [WIP-J2006] Jeong, J., "IPv6 Router Advertisement Option for DNS 1215 Configuration", Work in Progress, January 2006. 1216 1217 [WIP-LB2005] Larson, M. and P. Barber, "Observed DNS Resolution 1218 Misbehavior", Work in Progress, February 2006. 1219 1220 [WIP-O2004] Ohta, M., "Preconfigured DNS Server Addresses", Work in 1221 Progress, February 2004. 1222 1223 [WIP-R2006] Roy, S., "IPv6 Neighbor Discovery On-Link Assumption 1224 Considered Harmful", Work in Progress, January 2006. 1225 1226 [WIP-RDP2004] Roy, S., Durand, A., and J. Paugh, "Issues with Dual 1227 Stack IPv6 on by Default", Work in Progress, July 2004. 1228 1229 [WIP-S2005a] Stapp, M., "The DHCP Client FQDN Option", Work in 1230 Progress, March 2006. 1231 1232 [WIP-S2005b] Stapp, M., "A DNS RR for Encoding DHCP Information 1233 (DHCID RR)", Work in Progress, March 2006. 1234 1235 [WIP-S2005c] Senie, D., "Encouraging the use of DNS IN-ADDR 1236 Mapping", Work in Progress, August 2005. 1237 1238 [WIP-SV2005] Stapp, M. and B. Volz, "Resolution of FQDN Conflicts 1239 among DHCP Clients", Work in Progress, March 2006. 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 Durand, et al. Informational [Page 23] 1263 RFC 4472 Considerations with IPv6 DNS April 2006 1264 1265 1266 Appendix A. Unique Local Addressing Considerations for DNS 1267 1268 Unique local addresses [RFC4193] have replaced the now-deprecated 1269 site-local addresses [RFC3879]. From the perspective of the DNS, the 1270 locally generated unique local addresses (LUL) and site-local 1271 addresses have similar properties. 1272 1273 The interactions with DNS come in two flavors: forward and reverse 1274 DNS. 1275 1276 To actually use local addresses within a site, this implies the 1277 deployment of a "split-faced" or a fragmented DNS name space, for the 1278 zones internal to the site, and the outsiders' view to it. The 1279 procedures to achieve this are not elaborated here. The implication 1280 is that local addresses must not be published in the public DNS. 1281 1282 To facilitate reverse DNS (if desired) with local addresses, the stub 1283 resolvers must look for DNS information from the local DNS servers, 1284 not, e.g., starting from the root servers, so that the local 1285 information may be provided locally. Note that the experience of 1286 private addresses in IPv4 has shown that the root servers get loaded 1287 for requests for private address lookups in any case. This 1288 requirement is discussed in [RFC4193]. 1289 1290 Appendix B. Behavior of Additional Data in IPv4/IPv6 Environments 1291 1292 DNS responses do not always fit in a single UDP packet. We'll 1293 examine the cases that happen when this is due to too much data in 1294 the Additional section. 1295 1296 B.1. Description of Additional Data Scenarios 1297 1298 There are two kinds of additional data: 1299 1300 1. "critical" additional data; this must be included in all 1301 scenarios, with all the RRsets, and 1302 1303 2. "courtesy" additional data; this could be sent in full, with only 1304 a few RRsets, or with no RRsets, and can be fetched separately as 1305 well, but at the cost of additional queries. 1306 1307 The responding server can algorithmically determine which type the 1308 additional data is by checking whether it's at or below a zone cut. 1309 1310 Only those additional data records (even if sometimes carelessly 1311 termed "glue") are considered "critical" or real "glue" if and only 1312 if they meet the above-mentioned condition, as specified in Section 1313 4.2.1 of [RFC1034]. 1314 1315 1316 1317 Durand, et al. Informational [Page 24] 1318 RFC 4472 Considerations with IPv6 DNS April 2006 1319 1320 1321 Remember that resource record sets (RRsets) are never "broken up", so 1322 if a name has 4 A records and 5 AAAA records, you can either return 1323 all 9, all 4 A records, all 5 AAAA records, or nothing. In 1324 particular, notice that for the "critical" additional data getting 1325 all the RRsets can be critical. 1326 1327 In particular, [RFC2181] specifies (in Section 9) that: 1328 1329 a. if all the "critical" RRsets do not fit, the sender should set 1330 the TC bit, and the recipient should discard the whole response 1331 and retry using mechanism allowing larger responses such as TCP. 1332 1333 b. "courtesy" additional data should not cause the setting of the TC 1334 bit, but instead all the non-fitting additional data RRsets 1335 should be removed. 1336 1337 An example of the "courtesy" additional data is A/AAAA records in 1338 conjunction with MX records as shown in Section 4.4; an example of 1339 the "critical" additional data is shown below (where getting both the 1340 A and AAAA RRsets is critical with respect to the NS RR): 1341 1342 child.example.com. IN NS ns.child.example.com. 1343 ns.child.example.com. IN A 192.0.2.1 1344 ns.child.example.com. IN AAAA 2001:db8::1 1345 1346 When there is too much "courtesy" additional data, at least the non- 1347 fitting RRsets should be removed [RFC2181]; however, as the 1348 additional data is not critical, even all of it could be safely 1349 removed. 1350 1351 When there is too much "critical" additional data, TC bit will have 1352 to be set, and the recipient should ignore the response and retry 1353 using TCP; if some data were to be left in the UDP response, the 1354 issue is which data could be retained. 1355 1356 However, the practice may differ from the specification. Testing and 1357 code analysis of three recent implementations [TC-TEST] confirm this. 1358 None of the tested implementations have a strict separation of 1359 critical and courtesy additional data, while some forms of additional 1360 data may be treated preferably. All the implementations remove some 1361 (critical or courtesy) additional data RRsets without setting the TC 1362 bit if the response would not otherwise fit. 1363 1364 Failing to discard the response with the TC bit or omitting critical 1365 information but not setting the TC bit lead to an unrecoverable 1366 problem. Omitting only some of the RRsets if all would not fit (but 1367 not setting the TC bit) leads to a performance problem. These are 1368 discussed in the next two subsections. 1369 1370 1371 1372 Durand, et al. Informational [Page 25] 1373 RFC 4472 Considerations with IPv6 DNS April 2006 1374 1375 1376 B.2. Which Additional Data to Keep, If Any? 1377 1378 NOTE: omitting some critical additional data instead of setting the 1379 TC bit violates a 'should' in Section 9 of RFC2181. However, as many 1380 implementations still do that [TC-TEST], operators need to understand 1381 its implications, and we describe that behavior as well. 1382 1383 If the implementation decides to keep as much data (whether 1384 "critical" or "courtesy") as possible in the UDP responses, it might 1385 be tempting to use the transport of the DNS query as a hint in either 1386 of these cases: return the AAAA records if the query was done over 1387 IPv6, or return the A records if the query was done over IPv4. 1388 However, this breaks the model of independence of DNS transport and 1389 resource records, as noted in Section 1.2. 1390 1391 With courtesy additional data, as long as enough RRsets will be 1392 removed so that TC will not be set, it is allowed to send as many 1393 complete RRsets as the implementations prefers. However, the 1394 implementations are also free to omit all such RRsets, even if 1395 complete. Omitting all the RRsets (when removing only some would 1396 suffice) may create a performance penalty, whereby the client may 1397 need to issue one or more additional queries to obtain necessary 1398 and/or consistent information. 1399 1400 With critical additional data, the alternatives are either returning 1401 nothing (and absolutely requiring a retry with TCP) or returning 1402 something (working also in the case if the recipient does not discard 1403 the response and retry using TCP) in addition to setting the TC bit. 1404 If the process for selecting "something" from the critical data would 1405 otherwise be practically "flipping the coin" between A and AAAA 1406 records, it could be argued that if one looked at the transport of 1407 the query, it would have a larger possibility of being right than 1408 just 50/50. In other words, if the returned critical additional data 1409 would have to be selected somehow, using something more sophisticated 1410 than a random process would seem justifiable. 1411 1412 That is, leaving in some intelligently selected critical additional 1413 data is a trade-off between creating an optimization for those 1414 resolvers that ignore the "should discard" recommendation and causing 1415 a protocol problem by propagating inconsistent information about 1416 "critical" records in the caches. 1417 1418 Similarly, leaving in the complete courtesy additional data RRsets 1419 instead of removing all the RRsets is a performance trade-off as 1420 described in the next section. 1421 1422 1423 1424 1425 1426 1427 Durand, et al. Informational [Page 26] 1428 RFC 4472 Considerations with IPv6 DNS April 2006 1429 1430 1431 B.3. Discussion of the Potential Problems 1432 1433 As noted above, the temptation for omitting only some of the 1434 additional data could be problematic. This is discussed more below. 1435 1436 For courtesy additional data, this causes a potential performance 1437 problem as this requires that the clients issue re-queries for the 1438 potentially omitted RRsets. For critical additional data, this 1439 causes a potential unrecoverable problem if the response is not 1440 discarded and the query not re-tried with TCP, as the nameservers 1441 might be reachable only through the omitted RRsets. 1442 1443 If an implementation would look at the transport used for the query, 1444 it is worth remembering that often the host using the records is 1445 different from the node requesting them from the authoritative DNS 1446 server (or even a caching resolver). So, whichever version the 1447 requestor (e.g., a recursive server in the middle) uses makes no 1448 difference to the ultimate user of the records, whose transport 1449 capabilities might differ from those of the requestor. This might 1450 result in, e.g., inappropriately returning A records to an IPv6-only 1451 node, going through a translation, or opening up another IP-level 1452 session (e.g., a Packet Data Protocol (PDP) context [RFC4215]). 1453 Therefore, at least in many scenarios, it would be very useful if the 1454 information returned would be consistent and complete -- or if that 1455 is not feasible, leave it to the client to query again. 1456 1457 The problem of too much additional data seems to be an operational 1458 one: the zone administrator entering too many records that will be 1459 returned truncated (or missing some RRsets, depending on 1460 implementations) to the users. A protocol fix for this is using 1461 Extension Mechanisms for DNS (EDNS0) [RFC2671] to signal the capacity 1462 for larger UDP packet sizes, pushing up the relevant threshold. 1463 Further, DNS server implementations should omit courtesy additional 1464 data completely rather than including only some RRsets [RFC2181]. An 1465 operational fix for this is having the DNS server implementations 1466 return a warning when the administrators create zones that would 1467 result in too much additional data being returned. Further, DNS 1468 server implementations should warn of or disallow such zone 1469 configurations that are recursive or otherwise difficult to manage by 1470 the protocol. 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 Durand, et al. Informational [Page 27] 1483 RFC 4472 Considerations with IPv6 DNS April 2006 1484 1485 1486 Authors' Addresses 1487 1488 Alain Durand 1489 Comcast 1490 1500 Market St. 1491 Philadelphia, PA 19102 1492 USA 1493 1494 EMail: Alain_Durand@cable.comcast.com 1495 1496 1497 Johan Ihren 1498 Autonomica 1499 Bellmansgatan 30 1500 SE-118 47 Stockholm 1501 Sweden 1502 1503 EMail: email@example.com 1504 1505 1506 Pekka Savola 1507 CSC/FUNET 1508 Espoo 1509 Finland 1510 1511 EMail: firstname.lastname@example.org 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 Durand, et al. Informational [Page 28] 1538 RFC 4472 Considerations with IPv6 DNS April 2006 1539 1540 1541 Full Copyright Statement 1542 1543 Copyright (C) The Internet Society (2006). 1544 1545 This document is subject to the rights, licenses and restrictions 1546 contained in BCP 78, and except as set forth therein, the authors 1547 retain all their rights. 1548 1549 This document and the information contained herein are provided on an 1550 "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS 1551 OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET 1552 ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, 1553 INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE 1554 INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED 1555 WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. 1556 1557 Intellectual Property 1558 1559 The IETF takes no position regarding the validity or scope of any 1560 Intellectual Property Rights or other rights that might be claimed to 1561 pertain to the implementation or use of the technology described in 1562 this document or the extent to which any license under such rights 1563 might or might not be available; nor does it represent that it has 1564 made any independent effort to identify any such rights. Information 1565 on the procedures with respect to rights in RFC documents can be 1566 found in BCP 78 and BCP 79. 1567 1568 Copies of IPR disclosures made to the IETF Secretariat and any 1569 assurances of licenses to be made available, or the result of an 1570 attempt made to obtain a general license or permission for the use of 1571 such proprietary rights by implementers or users of this 1572 specification can be obtained from the IETF on-line IPR repository at 1573 http://www.ietf.org/ipr. 1574 1575 The IETF invites any interested party to bring to its attention any 1576 copyrights, patents or patent applications, or other proprietary 1577 rights that may cover technology that may be required to implement 1578 this standard. Please address the information to the IETF at 1579 email@example.com. 1580 1581 Acknowledgement 1582 1583 Funding for the RFC Editor function is provided by the IETF 1584 Administrative Support Activity (IASA). 1585 1586 1587 1588 1589 1590 1591 1592 Durand, et al. Informational [Page 29] 1593
The IETF is responsible for the creation and maintenance of the DNS RFCs. The ICANN DNS RFC annotation project provides a forum for collecting community annotations on these RFCs as an aid to understanding for implementers and any interested parties. The annotations displayed here are not the result of the IETF consensus process.
This RFC is included in the DNS RFCs annotation project whose home page is here.