idnits 2.17.1 draft-ietf-hip-rfc4423-bis-00.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == Line 1055 has weird spacing: '...ncement to th...' == The document seems to contain a disclaimer for pre-RFC5378 work, but was first submitted on or after 10 November 2008. The disclaimer is usually necessary only for documents that revise or obsolete older RFCs, and that take significant amounts of text from those RFCs. If you can contact all authors of the source material and they are willing to grant the BCP78 rights to the IETF Trust, you can and should remove the disclaimer. Otherwise, the disclaimer is needed and you can ignore this comment. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (August 17, 2010) is 5002 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Unused Reference: 'RFC4025' is defined on line 1038, but no explicit reference was found in the text ** Obsolete normative reference: RFC 5202 (Obsoleted by RFC 7402) ** Obsolete normative reference: RFC 5204 (Obsoleted by RFC 8004) ** Obsolete normative reference: RFC 5205 (Obsoleted by RFC 8005) -- Obsolete informational reference (is this intentional?): RFC 2535 (Obsoleted by RFC 4033, RFC 4034, RFC 4035) -- Obsolete informational reference (is this intentional?): RFC 2766 (Obsoleted by RFC 4966) -- Obsolete informational reference (is this intentional?): RFC 4306 (Obsoleted by RFC 5996) Summary: 3 errors (**), 0 flaws (~~), 4 warnings (==), 4 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group R. Moskowitz 3 Internet-Draft ICSA labs 4 Obsoletes: 4423 (if approved) August 17, 2010 5 Intended status: Standards Track 6 Expires: February 18, 2011 8 Host Identity Protocol Architecture 9 draft-ietf-hip-rfc4423-bis-00 11 Abstract 13 This memo describes a new namespace, the Host Identity namespace, and 14 a new protocol layer, the Host Identity Protocol, between the 15 internetworking and transport layers. Herein are presented the 16 basics of the current namespaces, their strengths and weaknesses, and 17 how a new namespace will add completeness to them. The roles of this 18 new namespace in the protocols are defined. 20 This document obsoletes RFC 4423 and addresses the concerns raised by 21 the IESG, particularly that of crypto agility. It also incorporates 22 lessons learned from the implementations of RFC 5201. 24 Status of this Memo 26 This Internet-Draft is submitted in full conformance with the 27 provisions of BCP 78 and BCP 79. 29 Internet-Drafts are working documents of the Internet Engineering 30 Task Force (IETF). Note that other groups may also distribute 31 working documents as Internet-Drafts. The list of current Internet- 32 Drafts is at http://datatracker.ietf.org/drafts/current/. 34 Internet-Drafts are draft documents valid for a maximum of six months 35 and may be updated, replaced, or obsoleted by other documents at any 36 time. It is inappropriate to use Internet-Drafts as reference 37 material or to cite them other than as "work in progress." 39 This Internet-Draft will expire on February 18, 2011. 41 Copyright Notice 43 Copyright (c) 2010 IETF Trust and the persons identified as the 44 document authors. All rights reserved. 46 This document is subject to BCP 78 and the IETF Trust's Legal 47 Provisions Relating to IETF Documents 48 (http://trustee.ietf.org/license-info) in effect on the date of 49 publication of this document. Please review these documents 50 carefully, as they describe your rights and restrictions with respect 51 to this document. Code Components extracted from this document must 52 include Simplified BSD License text as described in Section 4.e of 53 the Trust Legal Provisions and are provided without warranty as 54 described in the Simplified BSD License. 56 This document may contain material from IETF Documents or IETF 57 Contributions published or made publicly available before November 58 10, 2008. The person(s) controlling the copyright in some of this 59 material may not have granted the IETF Trust the right to allow 60 modifications of such material outside the IETF Standards Process. 61 Without obtaining an adequate license from the person(s) controlling 62 the copyright in such materials, this document may not be modified 63 outside the IETF Standards Process, and derivative works of it may 64 not be created outside the IETF Standards Process, except to format 65 it for publication as an RFC or to translate it into languages other 66 than English. 68 Table of Contents 70 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 4 71 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . 4 72 2.1. Terms common to other documents . . . . . . . . . . . . . . 5 73 2.2. Terms specific to this and other HIP documents . . . . . . . 5 74 3. Background . . . . . . . . . . . . . . . . . . . . . . . . . 7 75 3.1. A desire for a namespace for computing platforms . . . . . . 7 76 4. Host Identity namespace . . . . . . . . . . . . . . . . . . 9 77 4.1. Host Identifiers . . . . . . . . . . . . . . . . . . . . . . 10 78 4.2. Storing Host Identifiers in DNS . . . . . . . . . . . . . . 10 79 4.3. Host Identity Tag (HIT) . . . . . . . . . . . . . . . . . . 11 80 4.4. Local Scope Identifier (LSI) . . . . . . . . . . . . . . . . 11 81 5. New stack architecture . . . . . . . . . . . . . . . . . . . 11 82 5.1. Transport associations and end-points . . . . . . . . . . . 12 83 6. End-host mobility and multi-homing . . . . . . . . . . . . . 13 84 6.1. Rendezvous mechanism . . . . . . . . . . . . . . . . . . . . 13 85 6.2. Protection against flooding attacks . . . . . . . . . . . . 14 86 7. HIP and IPsec . . . . . . . . . . . . . . . . . . . . . . . 14 87 8. HIP and NATs . . . . . . . . . . . . . . . . . . . . . . . . 15 88 8.1. HIP and TCP checksums . . . . . . . . . . . . . . . . . . . 16 89 9. Multicast . . . . . . . . . . . . . . . . . . . . . . . . . 16 90 10. HIP policies . . . . . . . . . . . . . . . . . . . . . . . . 16 91 11. Benefits of HIP . . . . . . . . . . . . . . . . . . . . . . 17 92 11.1. HIP's answers to NSRG questions . . . . . . . . . . . . . . 18 93 12. Security considerations . . . . . . . . . . . . . . . . . . 20 94 12.1. HITs used in ACLs . . . . . . . . . . . . . . . . . . . . . 21 95 12.2. Non-security considerations . . . . . . . . . . . . . . . . 22 96 13. IANA considerations . . . . . . . . . . . . . . . . . . . . 22 97 14. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . 23 98 15. References . . . . . . . . . . . . . . . . . . . . . . . . . 23 99 15.1. Normative References . . . . . . . . . . . . . . . . . . . . 23 100 15.2. Informative references . . . . . . . . . . . . . . . . . . . 23 101 Author's Address . . . . . . . . . . . . . . . . . . . . . . 24 103 1. Introduction 105 The Internet has two important global namespaces: Internet Protocol 106 (IP) addresses and Domain Name Service (DNS) names. These two 107 namespaces have a set of features and abstractions that have powered 108 the Internet to what it is today. They also have a number of 109 weaknesses. Basically, since they are all we have, we try and do too 110 much with them. Semantic overloading and functionality extensions 111 have greatly complicated these namespaces. 113 The proposed Host Identity namespace fills an important gap between 114 the IP and DNS namespaces. The Host Identity namespace consists of 115 Host Identifiers (HI). A Host Identifier is cryptographic in its 116 nature; it is the public key of an asymmetric key-pair. Each host 117 will have at least one Host Identity, but it will typically have more 118 than one. Each Host Identity uniquely identifies a single host, 119 i.e., no two hosts have the same Host Identity. The Host Identity, 120 and the corresponding Host Identifier, can either be public (e.g. 121 published in the DNS), or unpublished. Client systems will tend to 122 have both public and unpublished Identities. 124 There is a subtle but important difference between Host Identities 125 and Host Identifiers. An Identity refers to the abstract entity that 126 is identified. An Identifier, on the other hand, refers to the 127 concrete bit pattern that is used in the identification process. 129 Although the Host Identifiers could be used in many authentication 130 systems, such as IKEv2 [RFC4306], the presented architecture 131 introduces a new protocol, called the Host Identity Protocol (HIP), 132 and a cryptographic exchange, called the HIP base exchange; see also 133 Section 7. The HIP protocols under development provide for limited 134 forms of trust between systems, enhance mobility, multi-homing and 135 dynamic IP renumbering, aid in protocol translation / transition, and 136 reduce certain types of denial-of-service (DoS) attacks. 138 When HIP is used, the actual payload traffic between two HIP hosts is 139 typically, but not necessarily, protected with IPsec. The Host 140 Identities are used to create the needed IPsec Security Associations 141 (SAs) and to authenticate the hosts. When IPsec is used, the actual 142 payload IP packets do not differ in any way from standard IPsec 143 protected IP packets. 145 2. Terminology 146 2.1. Terms common to other documents 148 +---------------+---------------------------------------------------+ 149 | Term | Explanation | 150 +---------------+---------------------------------------------------+ 151 | Public key | The public key of an asymmetric cryptographic key | 152 | | pair. Used as a publicly known identifier for | 153 | | cryptographic identity authentication. | 154 | | | 155 | Private key | The private or secret key of an asymmetric | 156 | | cryptographic key pair. Assumed to be known only | 157 | | to the party identified by the corresponding | 158 | | public key. Used by the identified party to | 159 | | authenticate its identity to other parties. | 160 | | | 161 | Public key | An asymmetric cryptographic key pair consisting | 162 | pair | of public and private keys. For example, | 163 | | Rivest-Shamir-Adelman (RSA) and Digital Signature | 164 | | Algorithm (DSA) key pairs are such key pairs. | 165 | | | 166 | End-point | A communicating entity. For historical reasons, | 167 | | the term 'computing platform' is used in this | 168 | | document as a (rough) synonym for end-point. | 169 +---------------+---------------------------------------------------+ 171 2.2. Terms specific to this and other HIP documents 173 It should be noted that many of the terms defined herein are 174 tautologous, self-referential or defined through circular reference 175 to other terms. This is due to the succinct nature of the 176 definitions. See the text elsewhere in this document for more 177 elaborate explanations. 179 +---------------+---------------------------------------------------+ 180 | Term | Explanation | 181 +---------------+---------------------------------------------------+ 182 | Computing | An entity capable of communicating and computing, | 183 | platform | for example, a computer. See the definition of | 184 | | 'End-point', above. | 185 | | | 186 | HIP base | A cryptographic protocol; see also Section 7. | 187 | exchange | | 188 | | | 189 | HIP packet | An IP packet that carries a 'Host Identity | 190 | | Protocol' message. | 191 | | | 192 | Host Identity | An abstract concept assigned to a 'computing | 193 | | platform'. See 'Host Identifier', below. | 194 | | | 195 | Host Identity | A name space formed by all possible Host | 196 | namespace | Identifiers. | 197 | | | 198 | Host Identity | A protocol used to carry and authenticate Host | 199 | Protocol | Identifiers and other information. | 200 | | | 201 | Host Identity | A 128-bit datum created by taking a cryptographic | 202 | Tag | hash over a Host Identifier. | 203 | | | 204 | Host | A public key used as a name for a Host Identity. | 205 | Identifier | | 206 | | | 207 | Local Scope | A 32-bit datum denoting a Host Identity. | 208 | Identifier | | 209 | | | 210 | Public Host | A published or publicly known Host Identfier used | 211 | Identifier | as a public name for a Host Identity, and the | 212 | and Identity | corresponding Identity. | 213 | | | 214 | Unpublished | A Host Identifier that is not placed in any | 215 | Host | public directory, and the corresponding Host | 216 | Identifier | Identity. Unpublished Host Identities are | 217 | and Identity | typically short lived in nature, being often | 218 | | replaced and possibly used just once. | 219 | | | 220 | Rendezvous | A mechanism used to locate mobile hosts based on | 221 | Mechanism | their HIT. | 222 +---------------+---------------------------------------------------+ 224 3. Background 226 The Internet is built from three principal components: computing 227 platforms (end-points), packet transport (i.e., internetworking) 228 infrastructure, and services (applications). The Internet exists to 229 service two principal components: people and robotic services 230 (silicon based people, if you will). All these components need to be 231 named in order to interact in a scalable manner. Here we concentrate 232 on naming computing platforms and packet transport elements. 234 There are two principal namespaces in use in the Internet for these 235 components: IP numbers, and Domain Names. Domain Names provide 236 hierarchically assigned names for some computing platforms and some 237 services. Each hierarchy is delegated from the level above; there is 238 no anonymity in Domain Names. Email, HTTP, and SIP addresses all 239 reference Domain Names. 241 IP numbers are a confounding of two namespaces, the names of a host's 242 networking interfaces and the names of the locations ('confounding' 243 is a term used in statistics to discuss metrics that are merged into 244 one with a gain in indexing, but a loss in informational value). The 245 names of locations should be understood as denoting routing direction 246 vectors, i.e., information that is used to deliver packets to their 247 destinations. 249 IP numbers name networking interfaces, and typically only when the 250 interface is connected to the network. Originally, IP numbers had 251 long-term significance. Today, the vast number of interfaces use 252 ephemeral and/or non-unique IP numbers. That is, every time an 253 interface is connected to the network, it is assigned an IP number. 255 In the current Internet, the transport layers are coupled to the IP 256 addresses. Neither can evolve separately from the other. IPng 257 deliberations were strongly shaped by the decision that a 258 corresponding TCPng would not be created. 260 There are three critical deficiencies with the current namespaces. 261 Firstly, dynamic readdressing cannot be directly managed. Secondly, 262 anonymity is not provided in a consistent, trustable manner. 263 Finally, authentication for systems and datagrams is not provided. 264 All of these deficiencies arise because computing platforms are not 265 well named with the current namespaces. 267 3.1. A desire for a namespace for computing platforms 269 An independent namespace for computing platforms could be used in 270 end-to-end operations independent of the evolution of the 271 internetworking layer and across the many internetworking layers. 273 This could support rapid readdressing of the internetworking layer 274 because of mobility, rehoming, or renumbering. 276 If the namespace for computing platforms is based on public-key 277 cryptography, it can also provide authentication services. If this 278 namespace is locally created without requiring registration, it can 279 provide anonymity. 281 Such a namespace (for computing platforms) and the names in it should 282 have the following characteristics: 284 o The namespace should be applied to the IP 'kernel'. The IP kernel 285 is the 'component' between applications and the packet transport 286 infrastructure. 288 o The namespace should fully decouple the internetworking layer from 289 the higher layers. The names should replace all occurrences of IP 290 addresses within applications (like in the Transport Control 291 Block, TCB). This may require changes to the current APIs. In 292 the long run, it is probable that some new APIs are needed. 294 o The introduction of the namespace should not mandate any 295 administrative infrastructure. Deployment must come from the 296 bottom up, in a pairwise deployment. 298 o The names should have a fixed length representation, for easy 299 inclusion in datagram headers and existing programming interfaces 300 (e.g the TCB). 302 o Using the namespace should be affordable when used in protocols. 303 This is primarily a packet size issue. There is also a 304 computational concern in affordability. 306 o Name collisions should be avoided as much as possible. The 307 mathematics of the birthday paradox can be used to estimate the 308 chance of a collision in a given population and hash space. In 309 general, for a random hash space of size n bits, we would expect 310 to obtain a collision after approximately 1.2*sqrt(2**n) hashes 311 were obtained. For 64 bits, this number is roughly 4 billion. A 312 hash size of 64 bits may be too small to avoid collisions in a 313 large population; for example, there is a 1% chance of collision 314 in a population of 640M. For 100 bits (or more), we would not 315 expect a collision until approximately 2**50 (1 quadrillion) 316 hashes were generated. 318 o The names should have a localized abstraction so that it can be 319 used in existing protocols and APIs. 321 o It must be possible to create names locally. This can provide 322 anonymity at the cost of making resolvability very difficult. 324 * Sometimes the names may contain a delegation component. This 325 is the cost of resolvability. 327 o The namespace should provide authentication services. 329 o The names should be long lived, but replaceable at any time. This 330 impacts access control lists; short lifetimes will tend to result 331 in tedious list maintenance or require a namespace infrastructure 332 for central control of access lists. 334 In this document, a new namespace approaching these ideas is called 335 the Host Identity namespace. Using Host Identities requires its own 336 protocol layer, the Host Identity Protocol, between the 337 internetworking and transport layers. The names are based on public- 338 key cryptography to supply authentication services. Properly 339 designed, it can deliver all of the above stated requirements. 341 4. Host Identity namespace 343 A name in the Host Identity namespace, a Host Identifier (HI), 344 represents a statistically globally unique name for naming any system 345 with an IP stack. This identity is normally associated with, but not 346 limited to, an IP stack. A system can have multiple identities, some 347 'well known', some unpublished or 'anonymous'. A system may self- 348 assert its own identity, or may use a third-party authenticator like 349 DNSSEC [RFC2535], PGP, or X.509 to 'notarize' the identity assertion. 350 It is expected that the Host Identifiers will initially be 351 authenticated with DNSSEC and that all implementations will support 352 DNSSEC as a minimal baseline. 354 In theory, any name that can claim to be 'statistically globally 355 unique' may serve as a Host Identifier. However, in the authors' 356 opinion, a public key of a 'public key pair' makes the best Host 357 Identifier. As will be specified in the Host Identity Protocol 358 specification, a public-key-based HI can authenticate the HIP packets 359 and protect them for man-in-the-middle attacks. Since authenticated 360 datagrams are mandatory to provide much of HIP's denial-of-service 361 protection, the Diffie-Hellman exchange in HIP has to be 362 authenticated. Thus, only public-key HI and authenticated HIP 363 messages are supported in practice. In this document, the non- 364 cryptographic forms of HI and HIP are presented to complete the 365 theory of HI, but they should not be implemented as they could 366 produce worse denial-of-service attacks than the Internet has without 367 Host Identity. 369 4.1. Host Identifiers 371 Host Identity adds two main features to Internet protocols. The 372 first is a decoupling of the internetworking and transport layers; 373 see Section 5. This decoupling will allow for independent evolution 374 of the two layers. Additionally, it can provide end-to-end services 375 over multiple internetworking realms. The second feature is host 376 authentication. Because the Host Identifier is a public key, this 377 key can be used for authentication in security protocols like IPsec. 379 The only completely defined structure of the Host Identity is that of 380 a public/private key pair. In this case, the Host Identity is 381 referred to by its public component, the public key. Thus, the name 382 representing a Host Identity in the Host Identity namespace, i.e., 383 the Host Identifier, is the public key. In a way, the possession of 384 the private key defines the Identity itself. If the private key is 385 possessed by more than one node, the Identity can be considered to be 386 a distributed one. 388 Architecturally, any other Internet naming convention might form a 389 usable base for Host Identifiers. However, non-cryptographic names 390 should only be used in situations of high trust - low risk. That is 391 any place where host authentication is not needed (no risk of host 392 spoofing) and no use of IPsec. However, at least for interconnected 393 networks spanning several operational domains, the set of 394 environments where the risk of host spoofing allowed by non- 395 cryptographic Host Identifiers is acceptable is the null set. Hence, 396 the current HIP documents do not specify how to use any other types 397 of Host Identifiers but public keys. 399 The actual Host Identities are never directly used in any Internet 400 protocols. The corresponding Host Identifiers (public keys) may be 401 stored in various DNS or LDAP directories as identified elsewhere in 402 this document, and they are passed in the HIP base exchange. A Host 403 Identity Tag (HIT) is used in other protocols to represent the Host 404 Identity. Another representation of the Host Identities, the Local 405 Scope Identifier (LSI), can also be used in protocols and APIs. 407 4.2. Storing Host Identifiers in DNS 409 The public Host Identifiers should be stored in DNS; the unpublished 410 Host Identifiers should not be stored anywhere (besides the 411 communicating hosts themselves). The (public) HI is stored in a new 412 RR type. This RR type is defined in HIP DNS Extension [RFC5205]. 414 Alternatively, or in addition to storing Host Identifiers in the DNS, 415 they may be stored in various kinds of Public Key Infrastructure 416 (PKI). Such a practice may allow them to be used for purposes other 417 than pure host identification. 419 4.3. Host Identity Tag (HIT) 421 A Host Identity Tag is a 128-bit representation for a Host Identity. 422 It is created by taking a cryptographic hash over the corresponding 423 Host Identifier. There are two advantages of using a hash over using 424 the Host Identifier in protocols. Firstly, its fixed length makes 425 for easier protocol coding and also better manages the packet size 426 cost of this technology. Secondly, it presents the identity in a 427 consistent format to the protocol independent of the cryptographic 428 algorithms used. 430 In the HIP packets, the HITs identify the sender and recipient of a 431 packet. Consequently, a HIT should be unique in the whole IP 432 universe as long as it is being used. In the extremely rare case of 433 a single HIT mapping to more than one Host Identity, the Host 434 Identifiers (public keys) will make the final difference. If there 435 is more than one public key for a given node, the HIT acts as a hint 436 for the correct public key to use. 438 4.4. Local Scope Identifier (LSI) 440 An LSI is a 32-bit localized representation for a Host Identity. The 441 purpose of an LSI is to facilitate using Host Identities in existing 442 protocols and APIs. LSI's advantage over HIT is its size; its 443 disadvantage is its local scope. 445 Examples of how LSIs can be used include: as the address in an FTP 446 command and as the address in a socket call. Thus, LSIs act as a 447 bridge for Host Identities into IPv4-based protocols and APIs. LSIs 448 also make it possible for some IPv4 applications to run over an IPv6 449 network. 451 5. New stack architecture 453 One way to characterize Host Identity is to compare the proposed new 454 architecture with the current one. As discussed above, the IP 455 addresses can be seen to be a confounding of routing direction 456 vectors and interface names. Using the terminology from the IRTF 457 Name Space Research Group Report [I-D.irtf-nsrg-report] and, e.g., 458 the unpublished Internet-Draft Endpoints and Endpoint Names 459 [chiappa-endpoints], the IP addresses currently embody the dual role 460 of locators and end-point identifiers. That is, each IP address 461 names a topological location in the Internet, thereby acting as a 462 routing direction vector, or locator. At the same time, the IP 463 address names the physical network interface currently located at the 464 point-of-attachment, thereby acting as a end-point name. 466 In the HIP architecture, the end-point names and locators are 467 separated from each other. IP addresses continue to act as locators. 468 The Host Identifiers take the role of end-point identifiers. It is 469 important to understand that the end-point names based on Host 470 Identities are slightly different from interface names; a Host 471 Identity can be simultaneously reachable through several interfaces. 473 The difference between the bindings of the logical entities are 474 illustrated in Figure 1. 476 Service ------ Socket Service ------ Socket 477 | | 478 | | 479 | | 480 | | 481 End-point | End-point --- Host Identity 482 \ | | 483 \ | | 484 \ | | 485 \ | | 486 Location --- IP address Location --- IP address 488 Figure 1 490 5.1. Transport associations and end-points 492 Architecturally, HIP provides for a different binding of transport- 493 layer protocols. That is, the transport-layer associations, i.e., 494 TCP connections and UDP associations, are no longer bound to IP 495 addresses but to Host Identities. 497 It is possible that a single physical computer hosts several logical 498 end-points. With HIP, each of these end-points would have a distinct 499 Host Identity. Furthermore, since the transport associations are 500 bound to Host Identities, HIP provides for process migration and 501 clustered servers. That is, if a Host Identity is moved from one 502 physical computer to another, it is also possible to simultaneously 503 move all the transport associations without breaking them. 504 Similarly, if it is possible to distribute the processing of a single 505 Host Identity over several physical computers, HIP provides for 506 cluster based services without any changes at the client end-point. 508 6. End-host mobility and multi-homing 510 HIP decouples the transport from the internetworking layer, and binds 511 the transport associations to the Host Identities (through actually 512 either the HIT or LSI). Consequently, HIP can provide for a degree 513 of internetworking mobility and multi-homing at a low infrastructure 514 cost. HIP mobility includes IP address changes (via any method) to 515 either party. Thus, a system is considered mobile if its IP address 516 can change dynamically for any reason like PPP, DHCP, IPv6 prefix 517 reassignments, or a NAT device remapping its translation. Likewise, 518 a system is considered multi-homed if it has more than one globally 519 routable IP address at the same time. HIP links IP addresses 520 together, when multiple IP addresses correspond to the same Host 521 Identity, and if one address becomes unusable, or a more preferred 522 address becomes available, existing transport associations can easily 523 be moved to another address. 525 When a node moves while communication is already on-going, address 526 changes are rather straightforward. The peer of the mobile node can 527 just accept a HIP or an integrity protected IPsec packet from any 528 address and ignore the source address. However, as discussed in 529 Section 6.2 below, a mobile node must send a HIP readdress packet to 530 inform the peer of the new address(es), and the peer must verify that 531 the mobile node is reachable through these addresses. This is 532 especially helpful for those situations where the peer node is 533 sending data periodically to the mobile node (that is re-starting a 534 connection after the initial connection). 536 6.1. Rendezvous mechanism 538 Making a contact to a mobile node is slightly more involved. In 539 order to start the HIP exchange, the initiator node has to know how 540 to reach the mobile node. Although infrequently moving HIP nodes 541 could use Dynamic DNS [RFC2136] to update their reachability 542 information in the DNS, an alternative to using DNS in this fashion 543 is to use a piece of new static infrastructure to facilitate 544 rendezvous between HIP nodes. 546 The mobile node keeps the rendezvous infrastructure continuously 547 updated with its current IP address(es). The mobile nodes must trust 548 the rendezvous mechanism to properly maintain their HIT and IP 549 address mappings. 551 The rendezvous mechanism is also needed if both of the nodes happen 552 to change their address at the same time, either because they are 553 mobile and happen to move at the same time, because one of them is 554 off-line for a while, or because of some other reason. In such a 555 case, the HIP readdress packets will cross each other in the network 556 and never reach the peer node. 558 The HIP rendezvous mechanism is defined in HIP Rendezvous [RFC5204]. 560 6.2. Protection against flooding attacks 562 Although the idea of informing about address changes by simply 563 sending packets with a new source address appears appealing, it is 564 not secure enough. That is, even if HIP does not rely on the source 565 address for anything (once the base exchange has been completed), it 566 appears to be necessary to check a mobile node's reachability at the 567 new address before actually sending any larger amounts of traffic to 568 the new address. 570 Blindly accepting new addresses would potentially lead to flooding 571 Denial-of-Service attacks against third parties [RFC4225]. In a 572 distributed flooding attack an attacker opens high volume HIP 573 connections with a large number of hosts (using unpublished HIs), and 574 then claims to all of these hosts that it has moved to a target 575 node's IP address. If the peer hosts were to simply accept the move, 576 the result would be a packet flood to the target node's address. To 577 close this attack, HIP includes an address check mechanism where the 578 reachability of a node is separately checked at each address before 579 using the address for larger amounts of traffic. 581 Whenever HIP is used between two hosts that fully trust each other, 582 the hosts may optionally decide to skip the address tests. However, 583 such performance optimization must be restricted to peers that are 584 known to be trustworthy and capable of protecting themselves from 585 malicious software. 587 7. HIP and IPsec 589 The preferred way of implementing HIP is to use IPsec to carry the 590 actual data traffic. As of today, the only completely defined method 591 is to use IPsec Encapsulated Security Payload (ESP) to carry the data 592 packets [RFC5202]. In the future, other ways of transporting payload 593 data may be developed, including ones that do not use cryptographic 594 protection. 596 In practice, the HIP base exchange uses the cryptographic Host 597 Identifiers to set up a pair of ESP Security Associations (SAs) to 598 enable ESP in an end-to-end manner. This is implemented in a way 599 that can span addressing realms. 601 While it would be possible, at least in theory, to use some existing 602 cryptographic protocol, such as IKEv2 together with Host Identifiers, 603 to establish the needed SAs, HIP defines a new protocol. There are a 604 number of historical reasons for this, and there are also a few 605 architectural reasons. First, IKE (and IKEv2) were not designed with 606 middle boxes in mind. As adding a new naming layer allows one to 607 potentially add a new forwarding layer (see Section 8, below), it is 608 very important that the HIP protocols are friendly towards any middle 609 boxes. 611 Second, from a conceptual point of view, the IPsec Security Parameter 612 Index (SPI) in ESP provides a simple compression of the HITs. This 613 does require per-HIT-pair SAs (and SPIs), and a decrease of policy 614 granularity over other Key Management Protocols, such as IKE and 615 IKEv2. In particular, the current thinking is limited to a situation 616 where, conceptually, there is only one pair of SAs between any given 617 pair of HITs. In other words, from an architectural point of view, 618 HIP only supports host-to-host (or endpoint-to-endpoint) Security 619 Associations. If two hosts need more pairs of parallel SAs, they 620 should use separate HITs for that. However, future HIP extensions 621 may provide for more granularity and creation of several ESP SAs 622 between a pair of HITs. 624 Since HIP is designed for host usage, not for gateways or so called 625 Bump-in-the-Wire (BITW) implementations, only ESP transport mode is 626 supported. An ESP SA pair is indexed by the SPIs and the two HITs 627 (both HITs since a system can have more than one HIT). The SAs need 628 not to be bound to IP addresses; all internal control of the SA is by 629 the HITs. Thus, a host can easily change its address using Mobile 630 IP, DHCP, PPP, or IPv6 readdressing and still maintain the SAs. 631 Since the transports are bound to the SA (via an LSI or a HIT), any 632 active transport is also maintained. Thus, real-world conditions 633 like loss of a PPP connection and its re-establishment or a mobile 634 handover will not require a HIP negotiation or disruption of 635 transport services [Bel1998]. 637 Since HIP does not negotiate any SA lifetimes, all lifetimes are 638 local policy. The only lifetimes a HIP implementation must support 639 are sequence number rollover (for replay protection), and SA timeout. 640 An SA times out if no packets are received using that SA. 641 Implementations may support lifetimes for the various ESP transforms. 643 8. HIP and NATs 645 Passing packets between different IP addressing realms requires 646 changing IP addresses in the packet header. This may happen, for 647 example, when a packet is passed between the public Internet and a 648 private address space, or between IPv4 and IPv6 networks. The 649 address translation is usually implemented as Network Address 650 Translation (NAT) [RFC3022] or NAT Protocol translation (NAT-PT) 651 [RFC2766]. 653 In a network environment where identification is based on the IP 654 addresses, identifying the communicating nodes is difficult when NAT 655 is used. With HIP, the transport-layer end-points are bound to the 656 Host Identities. Thus, a connection between two hosts can traverse 657 many addressing realm boundaries. The IP addresses are used only for 658 routing purposes; they may be changed freely during packet traversal. 660 For a HIP-based flow, a HIP-aware NAT or NAT-PT system tracks the 661 mapping of HITs, and the corresponding IPsec SPIs, to an IP address. 662 The NAT system has to learn mappings both from HITs and from SPIs to 663 IP addresses. Many HITs (and SPIs) can map to a single IP address on 664 a NAT, simplifying connections on address poor NAT interfaces. The 665 NAT can gain much of its knowledge from the HIP packets themselves; 666 however, some NAT configuration may be necessary. 668 NAT systems cannot touch the datagrams within the IPsec envelope, 669 thus application-specific address translation must be done in the end 670 systems. HIP provides for 'Distributed NAT', and uses the HIT or the 671 LSI as a placeholder for embedded IP addresses. 673 8.1. HIP and TCP checksums 675 There is no way for a host to know if any of the IP addresses in an 676 IP header are the addresses used to calculate the TCP checksum. That 677 is, it is not feasible to calculate the TCP checksum using the actual 678 IP addresses in the pseudo header; the addresses received in the 679 incoming packet are not necessarily the same as they were on the 680 sending host. Furthermore, it is not possible to recompute the 681 upper-layer checksums in the NAT/NAT-PT system, since the traffic is 682 IPsec protected. Consequently, the TCP and UDP checksums are 683 calculated using the HITs in the place of the IP addresses in the 684 pseudo header. Furthermore, only the IPv6 pseudo header format is 685 used. This provides for IPv4 / IPv6 protocol translation. 687 9. Multicast 689 There was little if any concrete thoughts about how HIP might affect 690 IP-layer or application-layer multicast. 692 10. HIP policies 694 There are a number of variables that will influence the HIP exchanges 695 that each host must support. All HIP implementations should support 696 at least 2 HIs, one to publish in DNS and an unpublished one for 697 anonymous usage. Although unpublished HIs will be rarely used as 698 responder HIs, they are likely be common for initiators. Support for 699 multiple HIs is recommended. 701 Many initiators would want to use a different HI for different 702 responders. The implementations should provide for a policy of 703 initiator HIT to responder HIT. This policy should also include 704 preferred transforms and local lifetimes. 706 Responders would need a similar policy, describing the hosts allowed 707 to participate in HIP exchanges, and the preferred transforms and 708 local lifetimes. 710 11. Benefits of HIP 712 In the beginning, the network layer protocol (i.e., IP) had the 713 following four "classic" invariants: 715 o Non-mutable: The address sent is the address received. 717 o Non-mobile: The address doesn't change during the course of an 718 "association". 720 o Reversible: A return header can always be formed by reversing the 721 source and destination addresses. 723 o Omniscient: Each host knows what address a partner host can use to 724 send packets to it. 726 Actually, the fourth can be inferred from 1 and 3, but it is worth 727 mentioning for reasons that will be obvious soon if not already. 729 In the current "post-classic" world, we are intentionally trying to 730 get rid of the second invariant (both for mobility and for multi- 731 homing), and we have been forced to give up the first and the fourth. 732 Realm Specific IP [RFC3102] is an attempt to reinstate the fourth 733 invariant without the first invariant. IPv6 is an attempt to 734 reinstate the first invariant. 736 Few systems on the Internet have DNS names that are meaningful. That 737 is, if they have a Fully Qualified Domain Name (FQDN), that name 738 typically belongs to a NAT device or a dial-up server, and does not 739 really identify the system itself but its current connectivity. 740 FQDNs (and their extensions as email names) are application-layer 741 names; more frequently naming services than a particular system. 742 This is why many systems on the Internet are not registered in the 743 DNS; they do not have services of interest to other Internet hosts. 745 DNS names are references to IP addresses. This only demonstrates the 746 interrelationship of the networking and application layers. DNS, as 747 the Internet's only deployed, distributed database is also the 748 repository of other namespaces, due in part to DNSSEC and application 749 specific key records. Although each namespace can be stretched (IP 750 with v6, DNS with KEY records), neither can adequately provide for 751 host authentication or act as a separation between internetworking 752 and transport layers. 754 The Host Identity (HI) namespace fills an important gap between the 755 IP and DNS namespaces. An interesting thing about the HI is that it 756 actually allows one to give up all but the 3rd network-layer 757 invariant. That is to say, as long as the source and destination 758 addresses in the network-layer protocol are reversible, then things 759 work ok because HIP takes care of host identification, and 760 reversibility allows one to get a packet back to one's partner host. 761 You do not care if the network-layer address changes in transit 762 (mutable) and you don't care what network-layer address the partner 763 is using (non-omniscient). 765 11.1. HIP's answers to NSRG questions 767 The IRTF Name Space Research Group has posed a number of evaluating 768 questions in their report [I-D.irtf-nsrg-report]. In this section, 769 we provide answers to these questions. 771 1. How would a stack name improve the overall functionality of the 772 Internet? 774 HIP decouples the internetworking layer from the transport 775 layer, allowing each to evolve separately. The decoupling 776 makes end-host mobility and multi-homing easier, also across 777 IPv4 and IPv6 networks. HIs make network renumbering easier, 778 and they also make process migration and clustered servers 779 easier to implement. Furthermore, being cryptographic in 780 nature, they provide the basis for solving the security 781 problems related to end-host mobility and multi-homing. 783 2. What does a stack name look like? 785 A HI is a cryptographic public key. However, instead of using 786 the keys directly, most protocols use a fixed size hash of the 787 public key. 789 3. What is its lifetime? 791 HIP provides both stable and temporary Host Identifiers. 792 Stable HIs are typically long lived, with a lifetime of years 793 or more. The lifetime of temporary HIs depends on how long 794 the upper-layer connections and applications need them, and 795 can range from a few seconds to years. 797 4. Where does it live in the stack? 799 The HIs live between the transport and internetworking layers. 801 5. How is it used on the end points 803 The Host Identifiers may be used directly or indirectly (in 804 the form of HITs or LSIs) by applications when they access 805 network services. Additionally, the Host Identifiers, as 806 public keys, are used in the built in key agreement protocol, 807 called the HIP base exchange, to authenticate the hosts to 808 each other. 810 6. What administrative infrastructure is needed to support it? 812 In some environments, it is possible to use HIP 813 opportunistically, without any infrastructure. However, to 814 gain full benefit from HIP, the HIs must be stored in the DNS 815 or a PKI, and a new rendezvous mechanism is needed[RFC5205]. 817 7. If we add an additional layer would it make the address list in 818 SCTP unnecessary? 820 Yes 822 8. What additional security benefits would a new naming scheme 823 offer? 825 HIP reduces dependency on IP addresses, making the so called 826 address ownership [Nik2001] problems easier to solve. In 827 practice, HIP provides security for end-host mobility and 828 multi-homing. Furthermore, since HIP Host Identifiers are 829 public keys, standard public key certificate infrastructures 830 can be applied on the top of HIP. 832 9. What would the resolution mechanisms be, or what characteristics 833 of a resolution mechanisms would be required? 835 For most purposes, an approach where DNS names are resolved 836 simultaneously to HIs and IP addresses is sufficient. 838 However, if it becomes necessary to resolve HIs into IP 839 addresses or back to DNS names, a flat resolution 840 infrastructure is needed. Such an infrastructure could be 841 based on the ideas of Distributed Hash Tables, but would 842 require significant new development and deployment. 844 12. Security considerations 846 HIP takes advantage of the new Host Identity paradigm to provide 847 secure authentication of hosts and to provide a fast key exchange for 848 IPsec. HIP also attempts to limit the exposure of the host to 849 various denial-of-service (DoS) and man-in-the-middle (MitM) attacks. 850 In so doing, HIP itself is subject to its own DoS and MitM attacks 851 that potentially could be more damaging to a host's ability to 852 conduct business as usual. 854 Resource exhausting denial-of-service attacks take advantage of the 855 cost of setting up a state for a protocol on the responder compared 856 to the 'cheapness' on the initiator. HIP allows a responder to 857 increase the cost of the start of state on the initiator and makes an 858 effort to reduce the cost to the responder. This is done by having 859 the responder start the authenticated Diffie-Hellman exchange instead 860 of the initiator, making the HIP base exchange 4 packets long. There 861 are more details on this process in the Host Identity Protocol under 862 development. 864 HIP optionally supports opportunistic negotiation. That is, if a 865 host receives a start of transport without a HIP negotiation, it can 866 attempt to force a HIP exchange before accepting the connection. 867 This has the potential for DoS attacks against both hosts. If the 868 method to force the start of HIP is expensive on either host, the 869 attacker need only spoof a TCP SYN. This would put both systems into 870 the expensive operations. HIP avoids this attack by having the 871 responder send a simple HIP packet that it can pre-build. Since this 872 packet is fixed and easily replayed, the initiator only reacts to it 873 if it has just started a connection to the responder. 875 Man-in-the-middle attacks are difficult to defend against, without 876 third-party authentication. A skillful MitM could easily handle all 877 parts of the HIP base exchange, but HIP indirectly provides the 878 following protection from a MitM attack. If the responder's HI is 879 retrieved from a signed DNS zone or secured by some other means, the 880 initiator can use this to authenticate the signed HIP packets. 881 Likewise, if the initiator's HI is in a secure DNS zone, the 882 responder can retrieve it and validate the signed HIP packets. 883 However, since an initiator may choose to use an unpublished HI, it 884 knowingly risks a MitM attack. The responder may choose not to 885 accept a HIP exchange with an initiator using an unknown HI. 887 In HIP, the Security Association for IPsec is indexed by the SPI; the 888 source address is always ignored, and the destination address may be 889 ignored as well. Therefore, HIP-enabled IPsec Encapsulated Security 890 Payload (ESP) is IP address independent. This might seem to make it 891 easier for an attacker, but ESP with replay protection is already as 892 well protected as possible, and the removal of the IP address as a 893 check should not increase the exposure of IPsec ESP to DoS attacks. 895 Since not all hosts will ever support HIP, ICMPv4 'Destination 896 Unreachable, Protocol Unreachable' and ICMPv6 'Parameter Problem, 897 Unrecognized Next Header' messages are to be expected and present a 898 DoS attack. Against an initiator, the attack would look like the 899 responder does not support HIP, but shortly after receiving the ICMP 900 message, the initiator would receive a valid HIP packet. Thus, to 901 protect against this attack, an initiator should not react to an ICMP 902 message until a reasonable time has passed, allowing it to get the 903 real responder's HIP packet. A similar attack against the responder 904 is more involved. 906 Another MitM attack is simulating a responder's administrative 907 rejection of a HIP initiation. This is a simple ICMP 'Destination 908 Unreachable, Administratively Prohibited' message. A HIP packet is 909 not used because it would either have to have unique content, and 910 thus difficult to generate, resulting in yet another DoS attack, or 911 just as spoofable as the ICMP message. Like in the previous case, 912 the defense against this attack is for the initiator to wait a 913 reasonable time period to get a valid HIP packet. If one does not 914 come, then the initiator has to assume that the ICMP message is 915 valid. Since this is the only point in the HIP base exchange where 916 this ICMP message is appropriate, it can be ignored at any other 917 point in the exchange. 919 12.1. HITs used in ACLs 921 It is expected that HITs will be used in ACLs. Future firewalls can 922 use HITs to control egress and ingress to networks, with an assurance 923 level difficult to achieve today. As discussed above in Section 7, 924 once a HIP session has been established, the SPI value in an IPsec 925 packet may be used as an index, indicating the HITs. In practice, 926 firewalls can inspect HIP packets to learn of the bindings between 927 HITs, SPI values, and IP addresses. They can even explicitly control 928 IPsec usage, dynamically opening IPsec ESP only for specific SPI 929 values and IP addresses. The signatures in HIP packets allow a 930 capable firewall to ensure that the HIP exchange is indeed happening 931 between two known hosts. This may increase firewall security. 933 There has been considerable bad experience with distributed ACLs that 934 contain public key related material, for example, with SSH. If the 935 owner of a key needs to revoke it for any reason, the task of finding 936 all locations where the key is held in an ACL may be impossible. If 937 the reason for the revocation is due to private key theft, this could 938 be a serious issue. 940 A host can keep track of all of its partners that might use its HIT 941 in an ACL by logging all remote HITs. It should only be necessary to 942 log responder hosts. With this information, the host can notify the 943 various hosts about the change to the HIT. There has been no attempt 944 to develop a secure method to issue the HIT revocation notice. 946 HIP-aware NATs, however, are transparent to the HIP aware systems by 947 design. Thus, the host may find it difficult to notify any NAT that 948 is using a HIT in an ACL. Since most systems will know of the NATs 949 for their network, there should be a process by which they can notify 950 these NATs of the change of the HIT. This is mandatory for systems 951 that function as responders behind a NAT. In a similar vein, if a 952 host is notified of a change in a HIT of an initiator, it should 953 notify its NAT of the change. In this manner, NATs will get updated 954 with the HIT change. 956 12.2. Non-security considerations 958 The definition of the Host Identifier states that the HI need not be 959 a public key. It implies that the HI could be any value; for example 960 a FQDN. This document does not describe how to support such a non- 961 cryptographic HI. A non-cryptographic HI would still offer the 962 services of the HIT or LSI for NAT traversal. It would be possible 963 to carry HITs in HIP packets that had neither privacy nor 964 authentication. Since such a mode would offer so little additional 965 functionality for so much addition to the IP kernel, it has not been 966 defined. Given how little public key cryptography HIP requires, HIP 967 should only be implemented using public key Host Identities. 969 If it is desirable to use HIP in a low security situation where 970 public key computations are considered expensive, HIP can be used 971 with very short Diffie-Hellman and Host Identity keys. Such use 972 makes the participating hosts vulnerable to MitM and connection 973 hijacking attacks. However, it does not cause flooding dangers, 974 since the address check mechanism relies on the routing system and 975 not on cryptographic strength. 977 13. IANA considerations 979 This document has no actions for IANA. 981 14. Acknowledgments 983 For the people historically involved in the early stages of HIP, see 984 the Acknowledgements section in the Host Identity Protocol 985 specification. 987 During the later stages of this document, when the editing baton was 988 transfered to Pekka Nikander, the comments from the early 989 implementors and others, including Jari Arkko, Tom Henderson, Petri 990 Jokela, Miika Komu, Mika Kousa, Andrew McGregor, Jan Melen, Tim 991 Shepard, Jukka Ylitalo, and Jorma Wall, were invaluable. Finally, 992 Lars Eggert, Spencer Dawkins and Dave Crocker provided valuable input 993 during the final stages of publication, most of which was 994 incorporated but some of which the authors decided to ignore in order 995 to get this document published in the first place. 997 The authors want to express their special thanks to Tom Henderson, 998 who took the burden of editing the document in response to IESG 999 comments at the time when both of the authors were busy doing other 1000 things. Without his perseverance this document might have never made 1001 it into an RFC form. 1003 15. References 1005 15.1. Normative References 1007 [RFC5202] Jokela, P., Moskowitz, R., and P. Nikander, "Using the 1008 Encapsulating Security Payload (ESP) Transport Format with 1009 the Host Identity Protocol (HIP)", RFC 5202, April 2008. 1011 [RFC5204] Laganier, J. and L. Eggert, "Host Identity Protocol (HIP) 1012 Rendezvous Extension", RFC 5204, April 2008. 1014 [RFC5205] Nikander, P. and J. Laganier, "Host Identity Protocol 1015 (HIP) Domain Name System (DNS) Extensions", RFC 5205, 1016 April 2008. 1018 15.2. Informative references 1020 [RFC2136] Vixie, P., Thomson, S., Rekhter, Y., and J. Bound, 1021 "Dynamic Updates in the Domain Name System (DNS UPDATE)", 1022 RFC 2136, April 1997. 1024 [RFC2535] Eastlake, D., "Domain Name System Security Extensions", 1025 RFC 2535, March 1999. 1027 [RFC2766] Tsirtsis, G. and P. Srisuresh, "Network Address 1028 Translation - Protocol Translation (NAT-PT)", RFC 2766, 1029 February 2000. 1031 [RFC3022] Srisuresh, P. and K. Egevang, "Traditional IP Network 1032 Address Translator (Traditional NAT)", RFC 3022, 1033 January 2001. 1035 [RFC3102] Borella, M., Lo, J., Grabelsky, D., and G. Montenegro, 1036 "Realm Specific IP: Framework", RFC 3102, October 2001. 1038 [RFC4025] Richardson, M., "A Method for Storing IPsec Keying 1039 Material in DNS", RFC 4025, March 2005. 1041 [RFC4225] Nikander, P., Arkko, J., Aura, T., Montenegro, G., and E. 1042 Nordmark, "Mobile IP Version 6 Route Optimization Security 1043 Design Background", RFC 4225, December 2005. 1045 [RFC4306] Kaufman, C., "Internet Key Exchange (IKEv2) Protocol", 1046 RFC 4306, December 2005. 1048 [I-D.irtf-nsrg-report] 1049 Lear, E. and R. Droms, "What's In A Name:Thoughts from the 1050 NSRG", draft-irtf-nsrg-report-10 (work in progress), 1051 September 2003. 1053 [chiappa-endpoints] 1054 Chiappa, J., "Endpoints and Endpoint Names: A Proposed 1055 Enhancement to the Internet Architecture", 1056 URL http://www.chiappa.net/~jnc/tech/endpoints.txt, 1999. 1058 [Nik2001] Nikander, P., "Denial-of-Service, Address Ownership, and 1059 Early Authentication in the IPv6 World", in Proceesings 1060 of Security Protocols, 9th International Workshop, 1061 Cambridge, UK, April 25-27 2001, LNCS 2467, pp. 12-26, 1062 Springer, 2002. 1064 [Bel1998] Bellovin, S., "EIDs, IPsec, and HostNAT", in Proceedings 1065 of 41th IETF, Los Angeles, CA, 1066 URL http://www1.cs.columbia.edu/~smb/talks/hostnat.pdf, 1067 March 1998. 1069 Author's Address 1071 Robert Moskowitz 1072 ICSA labs, An Independent Division of Verizon Business 1073 1000 Bent Creek Blvd, Suite 200 1074 Mechanicsburg, PA 1075 USA 1077 Email: robert.moskowitz@icsalabs.com