idnits 2.17.1 draft-ietf-sidr-rpki-rtr-25.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The abstract seems to contain references ([I-D.ietf-sidr-arch]), which it shouldn't. Please replace those with straight textual mentions of the documents in question. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (January 28, 2012) is 4465 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(See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Outdated reference: A later version (-10) exists of draft-ietf-sidr-pfx-validate-03 ** Obsolete normative reference: RFC 2385 (Obsoleted by RFC 5925) ** Obsolete normative reference: RFC 5226 (Obsoleted by RFC 8126) ** Obsolete normative reference: RFC 5246 (Obsoleted by RFC 8446) Summary: 4 errors (**), 0 flaws (~~), 2 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group R. Bush 3 Internet-Draft Internet Initiative Japan 4 Intended status: Standards Track R. Austein 5 Expires: July 31, 2012 Dragon Research Labs 6 January 28, 2012 8 The RPKI/Router Protocol 9 draft-ietf-sidr-rpki-rtr-25 11 Abstract 13 In order to formally validate the origin ASs of BGP announcements, 14 routers need a simple but reliable mechanism to receive RPKI 15 [I-D.ietf-sidr-arch] prefix origin data from a trusted cache. This 16 document describes a protocol to deliver validated prefix origin data 17 to routers. 19 Requirements Language 21 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 22 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 23 document are to be interpreted as described in [RFC2119]. 25 Status of this Memo 27 This Internet-Draft is submitted in full conformance with the 28 provisions of BCP 78 and BCP 79. 30 Internet-Drafts are working documents of the Internet Engineering 31 Task Force (IETF). Note that other groups may also distribute 32 working documents as Internet-Drafts. The list of current Internet- 33 Drafts is at http://datatracker.ietf.org/drafts/current/. 35 Internet-Drafts are draft documents valid for a maximum of six months 36 and may be updated, replaced, or obsoleted by other documents at any 37 time. It is inappropriate to use Internet-Drafts as reference 38 material or to cite them other than as "work in progress." 40 This Internet-Draft will expire on July 31, 2012. 42 Copyright Notice 44 Copyright (c) 2012 IETF Trust and the persons identified as the 45 document authors. All rights reserved. 47 This document is subject to BCP 78 and the IETF Trust's Legal 48 Provisions Relating to IETF Documents 49 (http://trustee.ietf.org/license-info) in effect on the date of 50 publication of this document. Please review these documents 51 carefully, as they describe your rights and restrictions with respect 52 to this document. Code Components extracted from this document must 53 include Simplified BSD License text as described in Section 4.e of 54 the Trust Legal Provisions and are provided without warranty as 55 described in the Simplified BSD License. 57 Table of Contents 59 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 60 2. Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 61 3. Deployment Structure . . . . . . . . . . . . . . . . . . . . . 4 62 4. Operational Overview . . . . . . . . . . . . . . . . . . . . . 4 63 5. Protocol Data Units (PDUs) . . . . . . . . . . . . . . . . . . 5 64 5.1. Serial Notify . . . . . . . . . . . . . . . . . . . . . . 6 65 5.2. Serial Query . . . . . . . . . . . . . . . . . . . . . . . 6 66 5.3. Reset Query . . . . . . . . . . . . . . . . . . . . . . . 7 67 5.4. Cache Response . . . . . . . . . . . . . . . . . . . . . . 7 68 5.5. IPv4 Prefix . . . . . . . . . . . . . . . . . . . . . . . 8 69 5.6. IPv6 Prefix . . . . . . . . . . . . . . . . . . . . . . . 9 70 5.7. End of Data . . . . . . . . . . . . . . . . . . . . . . . 9 71 5.8. Cache Reset . . . . . . . . . . . . . . . . . . . . . . . 10 72 5.9. Error Report . . . . . . . . . . . . . . . . . . . . . . . 10 73 5.10. Fields of a PDU . . . . . . . . . . . . . . . . . . . . . 11 74 6. Protocol Sequences . . . . . . . . . . . . . . . . . . . . . . 13 75 6.1. Start or Restart . . . . . . . . . . . . . . . . . . . . . 13 76 6.2. Typical Exchange . . . . . . . . . . . . . . . . . . . . . 14 77 6.3. No Incremental Update Available . . . . . . . . . . . . . 15 78 6.4. Cache has No Data Available . . . . . . . . . . . . . . . 15 79 7. Transport . . . . . . . . . . . . . . . . . . . . . . . . . . 16 80 7.1. SSH Transport . . . . . . . . . . . . . . . . . . . . . . 17 81 7.2. TLS Transport . . . . . . . . . . . . . . . . . . . . . . 18 82 7.3. TCP MD5 Transport . . . . . . . . . . . . . . . . . . . . 18 83 7.4. TCP-AO Transport . . . . . . . . . . . . . . . . . . . . . 18 84 8. Router-Cache Set-Up . . . . . . . . . . . . . . . . . . . . . 18 85 9. Deployment Scenarios . . . . . . . . . . . . . . . . . . . . . 19 86 10. Error Codes . . . . . . . . . . . . . . . . . . . . . . . . . 20 87 11. Security Considerations . . . . . . . . . . . . . . . . . . . 21 88 12. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 22 89 13. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 23 90 14. References . . . . . . . . . . . . . . . . . . . . . . . . . . 24 91 14.1. Normative References . . . . . . . . . . . . . . . . . . . 24 92 14.2. Informative References . . . . . . . . . . . . . . . . . . 25 93 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 25 95 1. Introduction 97 In order to formally validate the origin ASs of BGP announcements, 98 routers need a simple but reliable mechanism to receive RPKI 99 [I-D.ietf-sidr-arch] formally validated prefix origin data from a 100 trusted cache. This document describes a protocol to deliver 101 validated prefix origin data to routers. The design is intentionally 102 constrained to be usable on much of the current generation of ISP 103 router platforms. 105 Section 3 describes the deployment structure and Section 4 then 106 presents an operational overview. The binary payloads of the 107 protocol are formally described in Section 5, and the expected PDU 108 sequences are described in Section 6. The transport protocol options 109 are described in Section 7. Section 8 details how routers and caches 110 are configured to connect and authenticate. Section 9 describes 111 likely deployment scenarios. The traditional security and IANA 112 considerations end the document. 114 The protocol is extensible to support new PDUs with new semantics 115 when and as needed, as indicated by deployment experience. PDUs are 116 versioned should deployment experience call for change. 118 For an implementation (not inter-op) report, see 119 [I-D.ymbk-rpki-rtr-impl] 121 2. Glossary 123 The following terms are used with special meaning: 125 Global RPKI: The authoritative data of the RPKI are published in a 126 distributed set of servers at the IANA, RIRs, NIRs, and ISPs, see 127 [I-D.ietf-sidr-repos-struct]. 129 Cache: A coalesced copy of the RPKI which is periodically fetched/ 130 refreshed directly or indirectly from the global RPKI using the 131 [RFC5781] protocol/tools. Relying party software is used to 132 gather and validate the distributed data of the RPKI into a cache. 133 Trusting this cache further is a matter between the provider of 134 the cache and a relying party. 136 Serial Number: A 32-bit monotonically increasing unsigned integer 137 which wraps from 2^32-1 to 0. It denotes the logical version of a 138 cache. A cache increments the value by one when it successfully 139 updates its data from a parent cache or from primary RPKI data. 140 As a cache is receiving, new incoming data and implicit deletes 141 are marked with the new serial but MUST NOT be sent until the 142 fetch is complete. A serial number is not commensurate between 143 caches, nor need it be maintained across resets of the cache 144 server. See [RFC1982] on DNS Serial Number Arithmetic for too 145 much detail on serial number arithmetic. 147 Session ID: When a cache server is started, it generates a session 148 identifier to uniquely identify the instance of the cache and to 149 bind it to the sequence of Serial Numbers that cache instance will 150 generate. This allows the router to restart a failed session 151 knowing that the Serial Number it is using is commensurate with 152 that of the cache. 154 3. Deployment Structure 156 Deployment of the RPKI to reach routers has a three level structure 157 as follows: 159 Global RPKI: The authoritative data of the RPKI are published in a 160 distributed set of servers, RPKI publication repositories, e.g. 161 the IANA, RIRs, NIRs, and ISPs, see [I-D.ietf-sidr-repos-struct]. 163 Local Caches: A local set of one or more collected and verified 164 caches. A relying party, e.g. router or other client, MUST have a 165 trust relationship with, and a trusted transport channel to, any 166 authoritative cache(s) it uses. 168 Routers: A router fetches data from a local cache using the protocol 169 described in this document. It is said to be a client of the 170 cache. There MAY be mechanisms for the router to assure itself of 171 the authenticity of the cache and to authenticate itself to the 172 cache. 174 4. Operational Overview 176 A router establishes and keeps open a connection to one or more 177 caches with which it has client/server relationships. It is 178 configured with a semi-ordered list of caches, and establishes a 179 connection to the most preferred cache, or set of caches, which 180 accept the connections. 182 The router MUST choose the most preferred, by configuration, cache or 183 set of caches so that the operator may control load on their caches 184 and the Global RPKI. 186 Periodically, the router sends to the cache the serial number of the 187 highest numbered data it has received from that cache, i.e. the 188 router's current serial number. When a router establishes a new 189 connection to a cache, or wishes to reset a current relationship, it 190 sends a Reset Query. 192 The Cache responds with all data records which have serial numbers 193 greater than that in the router's query. This may be the null set, 194 in which case the End of Data PDU is still sent. Note that 'greater' 195 must take wrap-around into account, see [RFC1982]. 197 When the router has received all data records from the cache, it sets 198 its current serial number to that of the serial number in the End of 199 Data PDU. 201 When the cache updates its database, it sends a Notify message to 202 every currently connected router. This is a hint that now would be a 203 good time for the router to poll for an update, but is only a hint. 204 The protocol requires the router to poll for updates periodically in 205 any case. 207 Strictly speaking, a router could track a cache simply by asking for 208 a complete data set every time it updates, but this would be very 209 inefficient. The serial number based incremental update mechanism 210 allows an efficient transfer of just the data records which have 211 changed since last update. As with any update protocol based on 212 incremental transfers, the router must be prepared to fall back to a 213 full transfer if for any reason the cache is unable to provide the 214 necessary incremental data. Unlike some incremental transfer 215 protocols, this protocol requires the router to make an explicit 216 request to start the fallback process; this is deliberate, as the 217 cache has no way of knowing whether the router has also established 218 sessions with other caches that may be able to provide better 219 service. 221 As a cache server must evaluate certificates and ROAs (Route Origin 222 Attestations, see [I-D.ietf-sidr-arch]) which are time dependent, 223 servers' clocks MUST be correct to a tolerance of approximately an 224 hour. 226 5. Protocol Data Units (PDUs) 228 The exchanges between the cache and the router are sequences of 229 exchanges of the following PDUs according to the rules described in 230 Section 6. 232 Fields with unspecified content MUST be zero on transmission and MAY 233 be ignored on receipt. 235 5.1. Serial Notify 237 The cache notifies the router that the cache has new data. 239 The Session ID reassures the router that the serial numbers are 240 commensurate, i.e. the cache session has not been changed. 242 Serial Notify is only message that the cache can send that is not in 243 response to a message from the router. 245 0 8 16 24 31 246 .-------------------------------------------. 247 | Protocol | PDU | | 248 | Version | Type | Session ID | 249 | 0 | 0 | | 250 +-------------------------------------------+ 251 | | 252 | Length=12 | 253 | | 254 +-------------------------------------------+ 255 | | 256 | Serial Number | 257 | | 258 `-------------------------------------------' 260 5.2. Serial Query 262 Serial Query: The router sends Serial Query to ask the cache for all 263 payload PDUs which have serial numbers higher than the serial number 264 in the Serial Query. 266 The cache replies to this query with a Cache Response PDU 267 (Section 5.4) if the cache has a, possibly null, record of the 268 changes since the serial number specified by the router. If there 269 have been no changes since the router last queried, the cache then 270 sends an End Of Data PDU. 272 If the cache does not have the data needed to update the router, 273 perhaps because its records do not go back to the Serial Number in 274 the Serial Query, then it responds with a Cache Reset PDU 275 (Section 5.8). 277 The Session ID tells the cache what instance the router expects to 278 ensure that the serial numbers are commensurate, i.e. the cache 279 session has not been changed. 281 0 8 16 24 31 282 .-------------------------------------------. 283 | Protocol | PDU | | 284 | Version | Type | Session ID | 285 | 0 | 1 | | 286 +-------------------------------------------+ 287 | | 288 | Length=12 | 289 | | 290 +-------------------------------------------+ 291 | | 292 | Serial Number | 293 | | 294 `-------------------------------------------' 296 5.3. Reset Query 298 Reset Query: The router tells the cache that it wants to receive the 299 total active, current, non-withdrawn, database. The cache responds 300 with a Cache Response PDU (Section 5.4). 302 0 8 16 24 31 303 .-------------------------------------------. 304 | Protocol | PDU | | 305 | Version | Type | reserved = zero | 306 | 0 | 2 | | 307 +-------------------------------------------+ 308 | | 309 | Length=8 | 310 | | 311 `-------------------------------------------' 313 5.4. Cache Response 315 Cache Response: The cache responds with zero or more payload PDUs. 316 When replying to a Serial Query request (Section 5.2), the cache 317 sends the set of all data records it has with serial numbers greater 318 than that sent by the client router. When replying to a Reset Query, 319 the cache sends the set of all data records it has; in this case the 320 withdraw/announce field in the payload PDUs MUST have the value 1 321 (announce). 323 In response to a Reset Query, the new value of the Session ID tells 324 the router the instance of the cache session for future confirmation. 325 In response to a Serial Query, the Session ID being the same 326 reassures the router that the serial numbers are commensurate, i.e. 327 the cache session has not changed. 329 0 8 16 24 31 330 .-------------------------------------------. 331 | Protocol | PDU | | 332 | Version | Type | Session ID | 333 | 0 | 3 | | 334 +-------------------------------------------+ 335 | | 336 | Length=8 | 337 | | 338 `-------------------------------------------' 340 5.5. IPv4 Prefix 342 0 8 16 24 31 343 .-------------------------------------------. 344 | Protocol | PDU | | 345 | Version | Type | reserved = zero | 346 | 0 | 4 | | 347 +-------------------------------------------+ 348 | | 349 | Length=20 | 350 | | 351 +-------------------------------------------+ 352 | | Prefix | Max | | 353 | Flags | Length | Length | zero | 354 | | 0..32 | 0..32 | | 355 +-------------------------------------------+ 356 | | 357 | IPv4 Prefix | 358 | | 359 +-------------------------------------------+ 360 | | 361 | Autonomous System Number | 362 | | 363 `-------------------------------------------' 365 The lowest order bit of the Flags field is 1 for an announcement and 366 0 for a withdrawal. 368 In the RPKI, nothing prevents a signing certificate from issuing two 369 identical ROAs. In this case there would be no semantic difference 370 between the objects, merely a process redundancy. 372 In the RPKI, there is also an actual need for what might appear to a 373 router as identical IPvX (IPv4 or IPv6) PDUs. This can occur when an 374 upstream certificate is being reissued or there is an address 375 ownership transfer up the validation chain. The ROA would be 376 identical in the router sense, i.e. have the same {prefix, len, max- 377 len, asn}, but a different validation path in the RPKI. This is 378 important to the RPKI, but not to the router. 380 The cache server MUST ensure that it has told the router client to 381 have one and only one IPvX PDU for a unique {prefix, len, max-len, 382 asn} at any one point in time. Should the router client receive an 383 IPvX PDU with a {prefix, len, max-len, asn} identical to one it 384 already has active, it SHOULD raise a Duplicate Announcement Received 385 error. 387 5.6. IPv6 Prefix 389 0 8 16 24 31 390 .-------------------------------------------. 391 | Protocol | PDU | | 392 | Version | Type | reserved = zero | 393 | 0 | 6 | | 394 +-------------------------------------------+ 395 | | 396 | Length=32 | 397 | | 398 +-------------------------------------------+ 399 | | Prefix | Max | | 400 | Flags | Length | Length | zero | 401 | | 0..128 | 0..128 | | 402 +-------------------------------------------+ 403 | | 404 +--- ---+ 405 | | 406 +--- IPv6 Prefix ---+ 407 | | 408 +--- ---+ 409 | | 410 +-------------------------------------------+ 411 | | 412 | Autonomous System Number | 413 | | 414 `-------------------------------------------' 416 Analogous to the IPv4 Prefix PDU, 96 more bits no magic. 418 5.7. End of Data 420 End of Data: Cache tells router it has no more data for the request. 422 The Session ID MUST be the same as that of the corresponding Cache 423 Response which began the, possibly null, sequence of data PDUs. 425 0 8 16 24 31 426 .-------------------------------------------. 427 | Protocol | PDU | | 428 | Version | Type | Session ID | 429 | 0 | 7 | | 430 +-------------------------------------------+ 431 | | 432 | Length=12 | 433 | | 434 +-------------------------------------------+ 435 | | 436 | Serial Number | 437 | | 438 `-------------------------------------------' 440 5.8. Cache Reset 442 The cache may respond to a Serial Query informing the router that the 443 cache cannot provide an incremental update starting from the serial 444 number specified by the router. The router must decide whether to 445 issue a Reset Query or switch to a different cache. 447 0 8 16 24 31 448 .-------------------------------------------. 449 | Protocol | PDU | | 450 | Version | Type | reserved = zero | 451 | 0 | 8 | | 452 +-------------------------------------------+ 453 | | 454 | Length=8 | 455 | | 456 `-------------------------------------------' 458 5.9. Error Report 460 This PDU is used by either party to report an error to the other. 462 Error reports are only sent as responses to other PDUs. 464 The Error Code is described in Section 10. 466 If the error is not associated with any particular PDU, the Erroneous 467 PDU field MUST be empty and the Length of Encapsulated PDU field MUST 468 be zero. 470 An Error Report PDU MUST NOT be sent for an Error Report PDU. If an 471 erroneous Error Report PDU is received, the session SHOULD be 472 dropped. 474 If the error is associated with a PDU of excessive (too long to be 475 any legal PDU other than another Error Report), or possibly corrupt, 476 length, the Erroneous PDU field MAY be truncated. 478 The diagnostic text is optional, if not present the Length of Error 479 Text field SHOULD be zero. If error text is present, it SHOULD be a 480 string in US-ASCII, for maximum portability; if non-US-ASCII 481 characters are absolutely required, the error text MUST use UTF-8 482 encoding. 484 0 8 16 24 31 485 .-------------------------------------------. 486 | Protocol | PDU | | 487 | Version | Type | Error Code | 488 | 0 | 10 | | 489 +-------------------------------------------+ 490 | | 491 | Length | 492 | | 493 +-------------------------------------------+ 494 | | 495 | Length of Encapsulated PDU | 496 | | 497 +-------------------------------------------+ 498 | | 499 ~ Copy of Erroneous PDU ~ 500 | | 501 +-------------------------------------------+ 502 | | 503 | Length of Error Text | 504 | | 505 +-------------------------------------------+ 506 | | 507 | Arbitrary Text | 508 | of | 509 ~ Error Diagnostic Message ~ 510 | | 511 `-------------------------------------------' 513 5.10. Fields of a PDU 515 PDUs contain the following data elements: 517 Protocol Version: An eight-bit unsigned integer, currently 0, 518 denoting the version of this protocol. 520 PDU Type: An eight-bit unsigned integer, denoting the type of the 521 PDU, e.g. IPv4 Prefix, etc. 523 Serial Number: The serial number of the RPKI Cache when this ROA was 524 received from the cache's up-stream cache server or gathered from 525 the global RPKI. A cache increments its serial number when 526 completing an rigorously validated update from a parent cache, for 527 example via rcynic. See [RFC1982] on DNS Serial Number Arithmetic 528 for too much detail on serial number arithmetic. 530 Session ID: When a cache server is started, it generates a Session 531 ID to identify the instance of the cache and to bind it to the 532 sequence of Serial Numbers that cache instance will generate. 533 This allows the router to restart a failed session knowing that 534 the Serial Number it is using is commensurate with that of the 535 cache. If, at any time, either the router or the cache finds the 536 value of the session identifiers they hold disagree, they MUST 537 completely drop the session and the router MUST flush all data 538 learned from that cache. 540 Should a cache erroneously reuse a Session ID so that a router 541 does not realize that the session has changed (old session ID and 542 new session ID have same numeric value), the router may become 543 confused as to the content of the cache. The time it takes the 544 router to discover it is confused will depend on whether the 545 serial numbers are also reused. If the serial numbers in the old 546 and new sessions are different enough, the cache will respond to 547 the router's Serial Query with a Cache Reset, which will solve the 548 problem. If, however, the serial numbers are close, the cache may 549 respond with a Cache Response, which may not be enough to bring 550 the router into sync. In such cases, it's likely but not certain 551 that the router will detect some discrepancy between the state 552 that the cache expects and its own state. For example, the Cache 553 Response may tell the router to drop a record which the router 554 does not hold, or may tell the router to add a record which the 555 router already has. In such cases, a router will detect the error 556 and reset the session. The one case in which the router may stay 557 out of sync is when nothing in the Cache Response contradicts any 558 data currently held by the router. 560 Using persistent storage for the session identifier or a clock- 561 based scheme for generating session identifiers should avoid the 562 risk of session identifier collisions. 564 The Session ID might be a pseudo-random, a monotonically 565 increasing value if the cache has reliable storage, etc. 567 Length: A 32-bit unsigned integer which has as its value the count 568 of the bytes in the entire PDU, including the eight bytes of 569 header which end with the length field. 571 Flags: The lowest order bit of the Flags field is 1 for an 572 announcement and 0 for a withdrawal, whether this PDU announces a 573 new right to announce the prefix or withdraws a previously 574 announced right. A withdraw effectively deletes one previously 575 announced IPvX Prefix PDU with the exact same Prefix, Length, Max- 576 Len, and ASN. 578 Prefix Length: An eight-bit unsigned integer denoting the shortest 579 prefix allowed for the prefix. 581 Max Length: An eight-bit unsigned integer denoting the longest 582 prefix allowed by the prefix. This MUST NOT be less than the 583 Prefix Length element. 585 Prefix: The IPv4 or IPv6 prefix of the ROA. 587 Autonomous System Number: ASN allowed to announce this prefix, a 32- 588 bit unsigned integer. 590 Zero: Fields shown as zero or reserved MUST be zero. The value of 591 such a field MUST be ignored on receipt. 593 6. Protocol Sequences 595 The sequences of PDU transmissions fall into three conversations as 596 follows: 598 6.1. Start or Restart 600 Cache Router 601 ~ ~ 602 | <----- Reset Query -------- | R requests data (or Serial Query) 603 | | 604 | ----- Cache Response -----> | C confirms request 605 | ------- IPvX Prefix ------> | C sends zero or more 606 | ------- IPvX Prefix ------> | IPv4 and IPv6 Prefix 607 | ------- IPvX Prefix ------> | Payload PDUs 608 | ------ End of Data ------> | C sends End of Data 609 | | and sends new serial 610 ~ ~ 612 When a transport session is first established, the router MAY send a 613 Reset Query and the cache responds with a data sequence of all data 614 it contains. 616 Alternatively, if the router has significant unexpired data from a 617 broken session with the same cache, it MAY start with a Serial Query 618 containing the Session ID from the previous session to ensure the 619 serial numbers are commensurate. 621 This Reset Query sequence is also used when the router receives a 622 Cache Reset, chooses a new cache, or fears that it has otherwise lost 623 its way. 625 To limit the length of time a cache must keep the data necessary to 626 generate incremental updates, a router MUST send either a Serial 627 Query or a Reset Query no less frequently than once an hour. This 628 also acts as a keep alive at the application layer. 630 As the cache MAY not keep updates for little more than one hour, the 631 router MUST have a polling interval of no greater than once an hour. 633 6.2. Typical Exchange 635 Cache Router 636 ~ ~ 637 | -------- Notify ----------> | (optional) 638 | | 639 | <----- Serial Query ------- | R requests data 640 | | 641 | ----- Cache Response -----> | C confirms request 642 | ------- IPvX Prefix ------> | C sends zero or more 643 | ------- IPvX Prefix ------> | IPv4 and IPv6 Prefix 644 | ------- IPvX Prefix ------> | Payload PDUs 645 | ------ End of Data ------> | C sends End of Data 646 | | and sends new serial 647 ~ ~ 649 The cache server SHOULD send a notify PDU with its current serial 650 number when the cache's serial changes, with the expectation that the 651 router MAY then issue a serial query earlier than it otherwise might. 652 This is analogous to DNS NOTIFY in [RFC1996]. The cache MUST rate 653 limit Serial Notifies to no more frequently than one per minute. 655 When the transport layer is up and either a timer has gone off in the 656 router, or the cache has sent a Notify, the router queries for new 657 data by sending a Serial Query, and the cache sends all data newer 658 than the serial in the Serial Query. 660 To limit the length of time a cache must keep old withdraws, a router 661 MUST send either a Serial Query or a Reset Query no less frequently 662 than once an hour. 664 6.3. No Incremental Update Available 666 Cache Router 667 ~ ~ 668 | <----- Serial Query ------ | R requests data 669 | ------- Cache Reset ------> | C cannot supply update 670 | | from specified serial 671 | <------ Reset Query ------- | R requests new data 672 | ----- Cache Response -----> | C confirms request 673 | ------- IPvX Prefix ------> | C sends zero or more 674 | ------- IPvX Prefix ------> | IPv4 and IPv6 Prefix 675 | ------- IPvX Prefix ------> | Payload PDUs 676 | ------ End of Data ------> | C sends End of Data 677 | | and sends new serial 678 ~ ~ 680 The cache may respond to a Serial Query with a Cache Reset, informing 681 the router that the cache cannot supply an incremental update from 682 the serial number specified by the router. This might be because the 683 cache has lost state, or because the router has waited too long 684 between polls and the cache has cleaned up old data that it no longer 685 believes it needs, or because the cache has run out of storage space 686 and had to expire some old data early. Regardless of how this state 687 arose, the cache replies with a Cache Reset to tell the router that 688 it cannot honor the request. When a router receives this, the router 689 SHOULD attempt to connect to any more preferred caches in its cache 690 list. If there are no more preferred caches it MUST issue a Reset 691 Query and get an entire new load from the cache. 693 6.4. Cache has No Data Available 695 Cache Router 696 ~ ~ 697 | <----- Serial Query ------ | R requests data 698 | ---- Error Report PDU ----> | C No Data Available 699 ~ ~ 701 Cache Router 702 ~ ~ 703 | <----- Reset Query ------- | R requests data 704 | ---- Error Report PDU ----> | C No Data Available 705 ~ ~ 707 The cache may respond to either a Serial Query or a Reset Query 708 informing the router that the cache cannot supply any update at all. 709 The most likely cause is that the cache has lost state, perhaps due 710 to a restart, and has not yet recovered. While it is possible that a 711 cache might go into such a state without dropping any of its active 712 sessions, a router is more likely to see this behavior when it 713 initially connects and issues a Reset Query while the cache is still 714 rebuilding its database. 716 When a router receives this kind of error, the router SHOULD attempt 717 to connect to any other caches in its cache list, in preference 718 order. If no other caches are available, the router MUST issue 719 periodic Reset Queries until it gets a new usable load from the 720 cache. 722 7. Transport 724 The transport layer session between a router and a cache carries the 725 binary Protocol Data Units (PDUs) in a persistent session. 727 To prevent cache spoofing and DoS attacks by illegitimate routers, it 728 is highly desirable that the router and the cache are authenticated 729 to each other. Integrity protection for payloads is also desirable 730 to protect against monkey in the middle (MITM) attacks. 731 Unfortunately, there is no protocol to do so on all currently used 732 platforms. Therefore, as of this document, there is no mandatory to 733 implement transport which provides authentication and integrity 734 protection. 736 To reduce exposure to dropped but non-terminated sessions, both 737 caches and routers SHOULD enable keep alives when available in the 738 chosen transport protocol. 740 It is expected that, when TCP-AO [RFC5925] is available on all 741 platforms deployed by operators, it will become the mandatory to 742 implement transport. 744 Caches and routers MUST implement unprotected transport over TCP 745 using a port, rpki-rtr, to be assigned, see Section 12. Operators 746 SHOULD use procedural means, e.g. access control lists (ACLs), ... to 747 reduce the exposure to authentication issues. 749 Caches and routers SHOULD use TCP-AO, SSHv2, TCP MD5, or IPsec 750 transport. 752 If unprotected TCP is the transport, the cache and routers MUST be on 753 the same trusted and controlled network. 755 If available to the operator, caches and routers MUST use one of the 756 following more protected protocols. 758 Caches and routers SHOULD use TCP-AO transport [RFC5925] over the 759 rpki-rtr port. 761 Caches and routers MAY use SSHv2 transport [RFC4252] using a the 762 normal SSH port. For an example, see Section 7.1. 764 Caches and routers MAY use TCP MD5 transport [RFC2385] using the 765 rpki-rtr port. Note that TCP MD5 has been obsoleted by TCP-AO 766 [RFC5925]. 768 Caches and routers MAY use IPsec transport [RFC4301] using the rpki- 769 rtr port. 771 Caches and routers MAY use TLS transport [RFC5246] using using a 772 port, rpki-rtr-tls, to be assigned, see Section 12. 774 7.1. SSH Transport 776 To run over SSH, the client router first establishes an SSH transport 777 connection using the SSHv2 transport protocol, and the client and 778 server exchange keys for message integrity and encryption. The 779 client then invokes the "ssh-userauth" service to authenticate the 780 application, as described in the SSH authentication protocol RFC 4252 781 [RFC4252]. Once the application has been successfully authenticated, 782 the client invokes the "ssh-connection" service, also known as the 783 SSH connection protocol. 785 After the ssh-connection service is established, the client opens a 786 channel of type "session", which results in an SSH session. 788 Once the SSH session has been established, the application invokes 789 the application transport as an SSH subsystem called "rpki-rtr". 790 Subsystem support is a feature of SSH version 2 (SSHv2) and is not 791 included in SSHv1. Running this protocol as an SSH subsystem avoids 792 the need for the application to recognize shell prompts or skip over 793 extraneous information, such as a system message that is sent at 794 shell start-up. 796 It is assumed that the router and cache have exchanged keys out of 797 band by some reasonably secured means. 799 Cache servers supporting SSH transport MUST accept RSA and DSA 800 authentication, and SHOULD accept ECDSA authentication. User 801 authentication MUST be supported; host authentication MAY be 802 supported. Implementations MAY support password authentication. 803 Client routers SHOULD verify the public key of the cache, to avoid 804 monkey in the middle attacks. 806 7.2. TLS Transport 808 Client routers using TLS transport MUST use client-side certificates 809 for authentication. While in principle any type of certificate and 810 certificate authority may be used, in general cache operators will 811 generally wish to create their own small-scale CA and issue 812 certificates to each authorized router. This simplifies credential 813 roll-over; any unrevoked, unexpired certificate from the proper CA 814 may be used. If such certificates are used, the CN field [RFC5280] 815 MUST be used to denote the router's identity. 817 Clients SHOULD verify the cache's certificate as well, to avoid 818 monkey in the middle attacks. 820 7.3. TCP MD5 Transport 822 If TCP-MD5 is used, implementations MUST support key lengths of at 823 least 80 printable ASCII bytes, per section 4.5 of [RFC2385]. 824 Implementations MUST also support hexadecimal sequences of at least 825 32 characters, i.e., 128 bits. 827 Key rollover with TCP-MD5 is problematic. Cache servers SHOULD 828 support [RFC4808]. 830 7.4. TCP-AO Transport 832 Implementations MUST support key lengths of at least 80 printable 833 ASCII bytes. Implementations MUST also support hexadecimal sequences 834 of at least 32 characters, i.e., 128 bits. MAC lengths of at least 835 96 bits MUST be supported, per section 5.3 of [RFC2385]. 837 The cryptographic algorithms and associcated parameters described in 838 [RFC5926] MUST be supported. 840 8. Router-Cache Set-Up 842 A cache has the public authentication data for each router it is 843 configured to support. 845 A router may be configured to peer with a selection of caches, and a 846 cache may be configured to support a selection of routers. Each must 847 have the name of, and authentication data for, each peer. In 848 addition, in a router, this list has a non-unique preference value 849 for each server in order of preference. This preference merely 850 denotes proximity, not trust, preferred belief, etc. The client 851 router attempts to establish a session with each potential serving 852 cache in preference order, and then starts to load data from the most 853 preferred cache to which it can connect and authenticate. The 854 router's list of caches has the following elements: 856 Preference: An unsigned integer denoting the router's preference to 857 connect to that cache, the lower the value the more preferred. 859 Name: The IP Address or fully qualified domain name of the cache. 861 Key: Any needed public key of the cache. 863 MyKey: Any needed private key or certificate of this client. 865 Due to the distributed nature of the RPKI, caches simply can not be 866 rigorously synchronous. A client may hold data from multiple caches, 867 but MUST keep the data marked as to source, as later updates MUST 868 affect the correct data. 870 Just as there may be more than one covering ROA from a single cache, 871 there may be multiple covering ROAs from multiple caches. The 872 results are as described in [I-D.ietf-sidr-pfx-validate]. 874 If data from multiple caches are held, implementations MUST NOT 875 distinguish between data sources when performing validation. 877 When a more preferred cache becomes available, if resources allow, it 878 would be prudent for the client to start fetching from that cache. 880 The client SHOULD attempt to maintain at least one set of data, 881 regardless of whether it has chosen a different cache or established 882 a new connection to the previous cache. 884 A client MAY drop the data from a particular cache when it is fully 885 in synch with one or more other caches. 887 A client SHOULD delete the data from a cache when it has been unable 888 to refresh from that cache for a configurable timer value. The 889 default for that value is twice the polling period for that cache. 891 If a client loses connectivity to a cache it is using, or otherwise 892 decides to switch to a new cache, it SHOULD retain the data from the 893 previous cache until it has a full set of data from one or more other 894 caches. Note that this may already be true at the point of 895 connection loss if the client has connections to more than one cache. 897 9. Deployment Scenarios 899 For illustration, we present three likely deployment scenarios. 901 Small End Site: The small multi-homed end site may wish to outsource 902 the RPKI cache to one or more of their upstream ISPs. They would 903 exchange authentication material with the ISP using some out of 904 band mechanism, and their router(s) would connect to one or more 905 up-streams' caches. The ISPs would likely deploy caches intended 906 for customer use separately from the caches with which their own 907 BGP speakers peer. 909 Large End Site: A larger multi-homed end site might run one or more 910 caches, arranging them in a hierarchy of client caches, each 911 fetching from a serving cache which is closer to the global RPKI. 912 They might configure fall-back peerings to up-stream ISP caches. 914 ISP Backbone: A large ISP would likely have one or more redundant 915 caches in each major PoP, and these caches would fetch from each 916 other in an ISP-dependent topology so as not to place undue load 917 on the global RPKI publication infrastructure. 919 Experience with large DNS cache deployments has shown that complex 920 topologies are ill-advised as it is easy to make errors in the graph, 921 e.g. not maintaining a loop-free condition. 923 Of course, these are illustrations and there are other possible 924 deployment strategies. It is expected that minimizing load on the 925 global RPKI servers will be a major consideration. 927 To keep load on global RPKI services from unnecessary peaks, it is 928 recommended that primary caches which load from the distributed 929 global RPKI not do so all at the same times, e.g. on the hour. 930 Choose a random time, perhaps the ISP's AS number modulo 60 and 931 jitter the inter-fetch timing. 933 10. Error Codes 935 This section contains a preliminary list of error codes. The authors 936 expect additions to this section during development of the initial 937 implementations. Errors which are considered fatal SHOULD cause the 938 session to be dropped. 940 0: Corrupt Data (fatal): The receiver believes the received PDU to 941 be corrupt in a manner not specified by other error codes. 943 1: Internal Error (fatal): The party reporting the error experienced 944 some kind of internal error unrelated to protocol operation (ran 945 out of memory, a coding assertion failed, et cetera). 947 2: No Data Available: The cache believes itself to be in good 948 working order, but is unable to answer either a Serial Query or a 949 Reset Query because it has no useful data available at this time. 950 This is likely to be a temporary error, and most likely indicates 951 that the cache has not yet completed pulling down an initial 952 current data set from the global RPKI system after some kind of 953 event that invalidated whatever data it might have previously held 954 (reboot, network partition, et cetera). 956 3: Invalid Request (fatal): The cache server believes the client's 957 request to be invalid. 959 4: Unsupported Protocol Version (fatal): The Protocol Version is not 960 known by the receiver of the PDU. 962 5: Unsupported PDU Type (fatal): The PDU Type is not known by the 963 receiver of the PDU. 965 6: Withdrawal of Unknown Record (fatal): The received PDU has Flag=0 966 but a record for the Prefix/PrefixLength/MaxLength triple does not 967 exist in the receiver's database. 969 7: Duplicate Announcement Received (fatal): The received PDU has an 970 identical {prefix, len, max-len, asn} tuple as a PDU which is 971 still active in the router. 973 11. Security Considerations 975 As this document describes a security protocol, many aspects of 976 security interest are described in the relevant sections. This 977 section points out issues which may not be obvious in other sections. 979 Cache Validation: In order for a collection of caches as described 980 in Section 9 to guarantee a consistent view, they need to be given 981 consistent trust anchors to use in their internal validation 982 process. Distribution of a consistent trust anchor is assumed to 983 be out of band. 985 Cache Peer Identification: The router initiates a transport session 986 to a cache, which it identifies by either IP address or fully 987 qualified domain name. Be aware that a DNS or address spoofing 988 attack could make the correct cache unreachable. No session would 989 be established, as the authorization keys would not match. 991 Transport Security: The RPKI relies on object, not server or 992 transport, trust. I.e. the IANA root trust anchor is distributed 993 to all caches through some out of band means, and can then be used 994 by each cache to validate certificates and ROAs all the way down 995 the tree. The inter-cache relationships are based on this object 996 security model, hence the inter-cache transport can be lightly 997 protected. 999 But this protocol document assumes that the routers can not do the 1000 validation cryptography. Hence the last link, from cache to 1001 router, is secured by server authentication and transport level 1002 security. This is dangerous, as server authentication and 1003 transport have very different threat models than object security. 1005 So the strength of the trust relationship and the transport 1006 between the router(s) and the cache(s) are critical. You're 1007 betting your routing on this. 1009 While we can not say the cache must be on the same LAN, if only 1010 due to the issue of an enterprise wanting to off-load the cache 1011 task to their upstream ISP(s), locality, trust, and control are 1012 very critical issues here. The cache(s) really SHOULD be as 1013 close, in the sense of controlled and protected (against DDoS, 1014 MITM) transport, to the router(s) as possible. It also SHOULD be 1015 topologically close so that a minimum of validated routing data 1016 are needed to bootstrap a router's access to a cache. 1018 The identity of the cache server SHOULD be verified and 1019 authenticated by the router client, and vice versa, before any 1020 data are exchanged. 1022 Transports which can not provide the necessary authentication and 1023 integrity (see Section 7) must rely on network design and 1024 operational controls to provide protection against spoofing/ 1025 corruption attacks. As pointed out in Section 7, TCP-AO is the 1026 long term plan. Protocols which provide integrity and 1027 authenticity SHOULD be used, and if they can not, i.e. TCP is 1028 used as the transport, the router and cache MUST be on the same 1029 trusted, controlled network. 1031 12. IANA Considerations 1033 This document requests the IANA to assign 'well known' TCP Port 1034 Numbers to the RPKI-Router Protocol for the following, see Section 7: 1036 rpki-rtr 1037 rpki-rtr-tls 1039 This document requests the IANA to create a registry for tuples of 1040 Protocol Version / PDU Type, each of which may range from 0 to 255. 1041 The name of the registry should be rpki-rtr-pdu. The policy for 1042 adding to the registry is RFC Required per [RFC5226], either 1043 standards track or experimental. The initial entries should be as 1044 follows: 1046 Protocol 1047 Version PDU Type 1048 -------- ------------------- 1049 0 0 - Serial Notify 1050 0 1 - Serial Query 1051 0 2 - Reset Query 1052 0 3 - Cache Response 1053 0 4 - IPv4 Prefix 1054 0 6 - IPv6 Prefix 1055 0 7 - End of Data 1056 0 8 - Cache Reset 1057 0 10 - Error Report 1058 0 255 - Reserved 1060 This document requests the IANA to create a registry for Error Codes 1061 0 to 255. The name of the registry should be rpki-rtr-error. The 1062 policy for adding to the registry is Expert Review per [RFC5226], 1063 where the responsible IESG area director should appoint the Expert 1064 Reviewer. The initial entries should be as follows: 1066 0 - Corrupt Data 1067 1 - Internal Error 1068 2 - No Data Available 1069 3 - Invalid Request 1070 4 - Unsupported Protocol Version 1071 5 - Unsupported PDU Type 1072 6 - Withdrawal of Unknown Record 1073 7 - Duplicate Announcement Received 1074 255 - Reserved 1076 This document requests the IANA to add an SSH Connection Protocol 1077 Subsystem Name, as defined in [RFC4250], of 'rpki-rtr'. 1079 13. Acknowledgments 1081 The authors wish to thank Steve Bellovin, Rex Fernando, Paul Hoffman, 1082 Russ Housley, Pradosh Mohapatra, Keyur Patel, Sandy Murphy, Robert 1083 Raszuk, John Scudder, Ruediger Volk, and David Ward. Particular 1084 thanks go to Hannes Gredler for showing us the dangers of unnecessary 1085 fields. 1087 14. References 1089 14.1. Normative References 1091 [I-D.ietf-sidr-pfx-validate] 1092 Mohapatra, P., Scudder, J., Ward, D., Bush, R., and R. 1093 Austein, "BGP Prefix Origin Validation", 1094 draft-ietf-sidr-pfx-validate-03 (work in progress), 1095 October 2011. 1097 [RFC1982] Elz, R. and R. Bush, "Serial Number Arithmetic", RFC 1982, 1098 August 1996. 1100 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1101 Requirement Levels", BCP 14, RFC 2119, March 1997. 1103 [RFC2385] Heffernan, A., "Protection of BGP Sessions via the TCP MD5 1104 Signature Option", RFC 2385, August 1998. 1106 [RFC4250] Lehtinen, S. and C. Lonvick, "The Secure Shell (SSH) 1107 Protocol Assigned Numbers", RFC 4250, January 2006. 1109 [RFC4252] Ylonen, T. and C. Lonvick, "The Secure Shell (SSH) 1110 Authentication Protocol", RFC 4252, January 2006. 1112 [RFC4301] Kent, S. and K. Seo, "Security Architecture for the 1113 Internet Protocol", RFC 4301, December 2005. 1115 [RFC5226] Narten, T. and H. Alvestrand, "Guidelines for Writing an 1116 IANA Considerations Section in RFCs", BCP 26, RFC 5226, 1117 May 2008. 1119 [RFC5246] Dierks, T. and E. Rescorla, "The Transport Layer Security 1120 (TLS) Protocol Version 1.2", RFC 5246, August 2008. 1122 [RFC5280] Cooper, D., Santesson, S., Farrell, S., Boeyen, S., 1123 Housley, R., and W. Polk, "Internet X.509 Public Key 1124 Infrastructure Certificate and Certificate Revocation List 1125 (CRL) Profile", RFC 5280, May 2008. 1127 [RFC5925] Touch, J., Mankin, A., and R. Bonica, "The TCP 1128 Authentication Option", RFC 5925, June 2010. 1130 [RFC5926] Lebovitz, G. and E. Rescorla, "Cryptographic Algorithms 1131 for the TCP Authentication Option (TCP-AO)", RFC 5926, 1132 June 2010. 1134 14.2. Informative References 1136 [I-D.ietf-sidr-arch] 1137 Lepinski, M. and S. Kent, "An Infrastructure to Support 1138 Secure Internet Routing", draft-ietf-sidr-arch-13 (work in 1139 progress), May 2011. 1141 [I-D.ietf-sidr-repos-struct] 1142 Huston, G., Loomans, R., and G. Michaelson, "A Profile for 1143 Resource Certificate Repository Structure", 1144 draft-ietf-sidr-repos-struct-09 (work in progress), 1145 July 2011. 1147 [I-D.ymbk-rpki-rtr-impl] 1148 Bush, R., Austein, R., Patel, K., Gredler, H., and M. 1149 Waehlisch, "RPKI Router Implementation Report", 1150 draft-ymbk-rpki-rtr-impl-01 (work in progress), 1151 January 2012. 1153 [RFC1996] Vixie, P., "A Mechanism for Prompt Notification of Zone 1154 Changes (DNS NOTIFY)", RFC 1996, August 1996. 1156 [RFC4808] Bellovin, S., "Key Change Strategies for TCP-MD5", 1157 RFC 4808, March 2007. 1159 [RFC5781] Weiler, S., Ward, D., and R. Housley, "The rsync URI 1160 Scheme", RFC 5781, February 2010. 1162 Authors' Addresses 1164 Randy Bush 1165 Internet Initiative Japan 1166 5147 Crystal Springs 1167 Bainbridge Island, Washington 98110 1168 US 1170 Phone: +1 206 780 0431 x1 1171 Email: randy@psg.com 1173 Rob Austein 1174 Dragon Research Labs 1176 Email: sra@hactrn.net