idnits 2.17.1 draft-ietf-sidr-rpki-rtr-05.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** 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 == Using lowercase 'not' together with uppercase 'MUST', 'SHALL', 'SHOULD', or 'RECOMMENDED' is not an accepted usage according to RFC 2119. Please use uppercase 'NOT' together with RFC 2119 keywords (if that is what you mean). Found 'MUST not' in this paragraph: Serial Number: A 32-bit monotonically increasing ordinal which wraps from 2^32-1 to 0. It denotes the logical version of a cache. A cache increments the value by one when it successfully updates its data from a parent cache or from primary RPKI data. As a cache is receiving, new incoming data, and implicit deletes, are marked with the new serial but MUST not be sent until the fetch is complete. A serial number is not commensurate between caches, nor need it be maintained across resets of the cache server. See [RFC1982] on DNS Serial Number Arithmetic for too much detail on serial number arithmetic. -- The document date (December 16, 2010) is 4877 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) ** Downref: Normative reference to an Informational draft: draft-ietf-sidr-roa-validation (ref. 'I-D.ietf-sidr-roa-validation') ** Downref: Normative reference to an Informational RFC: RFC 5781 == Outdated reference: A later version (-13) exists of draft-ietf-sidr-arch-11 == Outdated reference: A later version (-09) exists of draft-ietf-sidr-repos-struct-06 Summary: 3 errors (**), 0 flaws (~~), 4 warnings (==), 1 comment (--).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 IIJ 4 Intended status: Standards Track R. Austein 5 Expires: June 19, 2011 ISC 6 December 16, 2010 8 The RPKI/Router Protocol 9 draft-ietf-sidr-rpki-rtr-05 11 Abstract 13 In order to formally validate the origin ASes of BGP announcements, 14 routers need a simple but reliable mechanism to receive RPKI 15 [I-D.ietf-sidr-arch] or analogous prefix origin data from a trusted 16 cache. This document describes a protocol to deliver validated 17 prefix origin data to routers over ssh. 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 RFC 2119 [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 June 19, 2011. 42 Copyright Notice 44 Copyright (c) 2010 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. Deployment Structure . . . . . . . . . . . . . . . . . . . . . 3 61 3. Operational Overview . . . . . . . . . . . . . . . . . . . . . 3 62 4. Protocol Data Units (PDUs) . . . . . . . . . . . . . . . . . . 4 63 4.1. Serial Notify . . . . . . . . . . . . . . . . . . . . . . 5 64 4.2. Serial Query . . . . . . . . . . . . . . . . . . . . . . . 5 65 4.3. Reset Query . . . . . . . . . . . . . . . . . . . . . . . 6 66 4.4. Cache Response . . . . . . . . . . . . . . . . . . . . . . 6 67 4.5. IPv4 Prefix . . . . . . . . . . . . . . . . . . . . . . . 7 68 4.6. IPv6 Prefix . . . . . . . . . . . . . . . . . . . . . . . 8 69 4.7. End of Data . . . . . . . . . . . . . . . . . . . . . . . 8 70 4.8. Cache Reset . . . . . . . . . . . . . . . . . . . . . . . 9 71 4.9. Error Report . . . . . . . . . . . . . . . . . . . . . . . 9 72 4.10. Fields of a PDU . . . . . . . . . . . . . . . . . . . . . 10 73 5. Protocol Sequences . . . . . . . . . . . . . . . . . . . . . . 11 74 5.1. Start or Restart . . . . . . . . . . . . . . . . . . . . . 11 75 5.2. Typical Exchange . . . . . . . . . . . . . . . . . . . . . 12 76 5.3. No Incremental Update Available . . . . . . . . . . . . . 13 77 5.4. Cache has No Data Available . . . . . . . . . . . . . . . 13 78 6. SSH Transport . . . . . . . . . . . . . . . . . . . . . . . . 14 79 7. Router-Cache Set-Up . . . . . . . . . . . . . . . . . . . . . 14 80 8. Deployment Scenarios . . . . . . . . . . . . . . . . . . . . . 16 81 9. Error Codes . . . . . . . . . . . . . . . . . . . . . . . . . 16 82 10. Security Considerations . . . . . . . . . . . . . . . . . . . 17 83 11. Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 84 12. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 19 85 13. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 19 86 14. References . . . . . . . . . . . . . . . . . . . . . . . . . . 19 87 14.1. Normative References . . . . . . . . . . . . . . . . . . . 19 88 14.2. Informative References . . . . . . . . . . . . . . . . . . 19 89 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 20 91 1. Introduction 93 In order to formally validate the origin ASes of BGP announcements, 94 routers need a simple but reliable mechanism to receive RPKI 95 [I-D.ietf-sidr-arch] or analogous formally validated prefix origin 96 data from a trusted cache. This document describes a protocol to 97 deliver validated prefix origin data to routers over ssh. 99 Section 2 describes the deployment structure and Section 3 then 100 presents an operational overview. The binary payloads of the 101 protocol are formally described in Section 4, and the expected PDU 102 sequences are described in Section 5. The transport protocol is 103 described in Section 6. Section 7 details how routers and caches are 104 configured to connect and authenticate. Section 8 describes likely 105 deployment scenarios. The traditional security and IANA 106 considerations end the document. 108 The protocol is extensible to support new PDUs with new semantics 109 when and as needed, as indicated by deployment experience. PDUs are 110 versioned should deployment experience call for change. 112 2. Deployment Structure 114 Deployment of the RPKI to reach routers has a three level structure 115 as follows: 117 Global RPKI: The authoritative data of the RPKI are published in a 118 distributed set of servers, RPKI publication repositories, e.g. 119 the IANA, RIRs, NIRs, and ISPs, see [I-D.ietf-sidr-repos-struct]. 121 Local Caches: A local set of one or more collected and verified non- 122 authoritative caches. A relying party, e.g. router or other 123 client, MUST have a formally authenticated trust relationship 124 with, and a secure transport channel to, any non-authoritative 125 cache(s) it uses. 127 Routers: A router fetches data from a local cache using the protocol 128 described in this document. It is said to be a client of the 129 cache. There are mechanisms for the router to assure itself of 130 the authenticity of the cache and to authenticate itself to the 131 cache. 133 3. Operational Overview 135 A router establishes and keeps open an authenticated connection to a 136 cache with which it has a client/server relationship. It is 137 configured with a semi-ordered list of caches, and establishes a 138 connection to the most preferred cache, or set of caches, which 139 accepts one. 141 Periodically, the router sends to the cache the serial number of the 142 highest numbered data record it has received from that cache, i.e. 143 the router's current serial number. When a router establishes a new 144 connection to a cache, or wishes to reset a current relationship, it 145 sends a Reset Query. 147 The Cache responds with all data records which have serial numbers 148 greater than that in the router's query. This may be the null set, 149 in which case the End of Data PDU is still sent. Note that 'greater' 150 must take wrap-around into account, see [RFC1982]. 152 When the router has received all data records from the cache, it sets 153 its current serial number to that of the serial number in the End of 154 Data PDU. 156 When the cache updates its database, it sends a Notify message to 157 every currently connected router. This is a hint that now would be a 158 good time for the router to poll for an update, but is only a hint. 159 The protocol requires the router to poll for updates periodically in 160 any case. 162 Strictly speaking, a router could track a cache simply by asking for 163 a complete data set every time it updates, but this would be very 164 inefficient. The serial number based incremental update mechanism 165 allows an efficient transfer of just the data records which have 166 changed since last update. As with any update protocol based on 167 incremental transfers, the router must be prepared to fall back to a 168 full transfer if for any reason the cache is unable to provide the 169 necessary incremental data. Unlike some incremental transfer 170 protocols, this protocol requires the router to make an explicit 171 request to start the fallback process; this is deliberate, as the 172 cache has no way of knowing whether the router has also established 173 sessions with other caches that may be able to provide better 174 service. 176 4. Protocol Data Units (PDUs) 178 The exchanges between the cache and the router are sequences of 179 exchanges of the following PDUs according to the rules described in 180 Section 5. 182 4.1. Serial Notify 184 The cache notifies the router that the cache has new data. 186 The Cache Nonce reassures the router that the serial numbers are 187 comensurate, i.e. the cache session has not been changed. 189 0 8 16 24 31 190 .-------------------------------------------. 191 | Protocol | PDU | | 192 | Version | Type | Cache Nonce | 193 | 0 | 0 | | 194 +-------------------------------------------+ 195 | | 196 | Length=12 | 197 | | 198 +-------------------------------------------+ 199 | | 200 | Serial Number | 201 | | 202 `-------------------------------------------' 204 4.2. Serial Query 206 Serial Query: The router sends Serial Query to ask the cache for all 207 payload PDUs which have serial numbers higher than the serial number 208 in the Serial Query. 210 The cache replies to this query with a Cache Response PDU 211 (Section 4.4) if the cache has a record of the changes since the 212 serial number specified by the router. If there have been no changes 213 since the router last queried, the cache responds with an End Of Data 214 PDU. If the cache does not have the data needed to update the 215 router, perhaps because its records do not go back to the Serial 216 Number in the Serial Query, then it responds with a Cache Reset PDU 217 (Section 4.8). 219 The Cache Nonce tells the cache what instance the router expects to 220 ensure that the serial numbers are comensurate, i.e. the cache 221 session has not been changed. 223 0 8 16 24 31 224 .-------------------------------------------. 225 | Protocol | PDU | | 226 | Version | Type | Cache Nonce | 227 | 0 | 1 | | 228 +-------------------------------------------+ 229 | | 230 | Length=12 | 231 | | 232 +-------------------------------------------+ 233 | | 234 | Serial Number | 235 | | 236 `-------------------------------------------' 238 4.3. Reset Query 240 Reset Query: The router tells the cache that it wants to receive the 241 total active, current, non-withdrawn, database. The cache responds 242 with a Cache Response PDU (Section 4.4). 244 0 8 16 24 31 245 .-------------------------------------------. 246 | Protocol | PDU | | 247 | Version | Type | reserved = zero | 248 | 0 | 2 | | 249 +-------------------------------------------+ 250 | | 251 | Length=8 | 252 | | 253 `-------------------------------------------' 255 4.4. Cache Response 257 Cache Response: The cache responds with zero or more payload PDUs. 258 When replying to a Serial Query request (Section 4.2), the cache 259 sends the set of all data records it has with serial numbers greater 260 than that sent by the client router. When replying to a Reset Query, 261 the cache sends the set of all data records it has; in this case the 262 withdraw/announce field in the payload PDUs MUST have the value 1 263 (announce). 265 In response to a Reset Query, the Cache Nonce tells the router the 266 instance of the cache session for future confirmation. In response 267 to a Serial Query, the Cache Nonce reassures the router that the 268 serial numbers are comensurate, i.e. the cache session has not been 269 changed. 271 0 8 16 24 31 272 .-------------------------------------------. 273 | Protocol | PDU | | 274 | Version | Type | Cache Nonce | 275 | 0 | 3 | | 276 +-------------------------------------------+ 277 | | 278 | Length=8 | 279 | | 280 `-------------------------------------------' 282 4.5. IPv4 Prefix 284 0 8 16 24 31 285 .-------------------------------------------. 286 | Protocol | PDU | | 287 | Version | Type | reserved = zero | 288 | 0 | 4 | | 289 +-------------------------------------------+ 290 | | 291 | Length=20 | 292 | | 293 +-------------------------------------------+ 294 | | Prefix | Max | | 295 | Flags | Length | Length | zero | 296 | | 0..32 | 0..32 | | 297 +-------------------------------------------+ 298 | | 299 | IPv4 prefix | 300 | | 301 +-------------------------------------------+ 302 | | 303 | Autonomous System Number | 304 | | 305 `-------------------------------------------' 307 Due to the nature of the RPKI and the IRR, there can be multiple 308 identical IPvX PDUs. A router MUST be prepared to receive multiple 309 identical record announcements and MUST NOT consider a record to have 310 been deleted until it has received a corresponding number of 311 withdrawals or a reset is performed Hence the router will likely keep 312 an internal reference count on each IPvX PDU. 314 In the RPKI, nothing prevents a signing certificate from issuing two 315 identical ROAs, and nothing prohibits the existence of two identical 316 route: or route6: objects in the IRR. In this case there would be no 317 semantic difference between the objects, merely a process redundancy. 319 In the RPKI, there is also an actual need for what will appear to the 320 router as identical IPvX PDUs. This occurs when an upstream 321 certificate is being reissued or a site is changing providers, often 322 a 'make and break' situation. The ROA is identical in the router 323 sense, i.e. has the same {prefix, len, max-len, asn}, but has a 324 different validation path in the RPKI. This is important to the 325 RPKI, but not to the router. 327 The lowest order bit of the Flags field is 1 for an announcement and 328 0 for a withdrawal. 330 4.6. IPv6 Prefix 332 0 8 16 24 31 333 .-------------------------------------------. 334 | Protocol | PDU | | 335 | Version | Type | reserved = zero | 336 | 0 | 6 | | 337 +-------------------------------------------+ 338 | | 339 | Length=32 | 340 | | 341 +-------------------------------------------+ 342 | | Prefix | Max | | 343 | Flags | Length | Length | zero | 344 | | 0..32 | 0..128 | | 345 +-------------------------------------------+ 346 | | 347 +--- ---+ 348 | | 349 +--- IPv6 prefix ---+ 350 | | 351 +--- ---+ 352 | | 353 +-------------------------------------------+ 354 | | 355 | Autonomous System Number | 356 | | 357 `-------------------------------------------' 359 4.7. End of Data 361 End of Data: Cache tells router it has no more data for the request. 363 0 8 16 24 31 364 .-------------------------------------------. 365 | Protocol | PDU | | 366 | Version | Type | reserved = zero | 367 | 0 | 7 | | 368 +-------------------------------------------+ 369 | | 370 | Length=12 | 371 | | 372 +-------------------------------------------+ 373 | | 374 | Serial Number | 375 | | 376 `-------------------------------------------' 378 4.8. Cache Reset 380 The cache may respond to a Serial Query informing the router that the 381 cache cannot provide an incremental update starting from the serial 382 number specified by the router. The router must decide whether to 383 issue a Reset Query or switch to a different cache. 385 0 8 16 24 31 386 .-------------------------------------------. 387 | Protocol | PDU | | 388 | Version | Type | reserved = zero | 389 | 0 | 8 | | 390 +-------------------------------------------+ 391 | | 392 | Length=8 | 393 | | 394 `-------------------------------------------' 396 4.9. Error Report 398 This PDU is used by either party to report an error to the other. 400 The Error Number is described in Section 9. 402 If the error is not associated with any particular PDU, the Erroneous 403 PDU field should be empty and the Length of Encapsulated PDU field 404 should be zero. 406 The diagnostic text is optional, if not present the Length of Error 407 Text field should be zero. If error text is present, it SHOULD be a 408 string in US-ASCII, for maximum portability; if non-US-ASCII 409 characters are absolutely required, the error text MUST use UTF-8 410 encoding. 412 0 8 16 24 31 413 .-------------------------------------------. 414 | Protocol | PDU | | 415 | Version | Type | Error Number | 416 | 0 | 10 | | 417 +-------------------------------------------+ 418 | | 419 | Length | 420 | | 421 +-------------------------------------------+ 422 | | 423 | Length of Encapsulated PDU | 424 | | 425 +-------------------------------------------+ 426 | | 427 ~ Copy of Erroneous PDU ~ 428 | | 429 +-------------------------------------------+ 430 | | 431 | Length of Error Text | 432 | | 433 +-------------------------------------------+ 434 | | 435 | Arbitrary Text | 436 | of | 437 ~ Error Diagnostic Message ~ 438 | | 439 `-------------------------------------------' 441 4.10. Fields of a PDU 443 PDUs contain the following data elements: 445 Protocol Version: An ordinal, currently 0, denoting the version of 446 this protocol. 448 Serial Number: The serial number of the RPKI Cache when this ROA was 449 received from the cache's up-stream cache server or gathered from 450 the global RPKI. A cache increments its serial number when 451 completing an rigorously validated update from a parent cache, for 452 example via rcynic. See [RFC1982] on DNS Serial Number Arithmetic 453 for too much detail on serial number arithmetic. 455 Cache Nonce: When a cache server is started, it generates a nonce to 456 identify the instance of the cache and to bind it to the sequence 457 of Serial Numbers that cache instance will generate. This allows 458 the router to restart a failed session knowing that the Serial 459 Number it is using is comensurate with that of the cache. If, at 460 any time, either the router or the cache finds the value of the 461 nonces they hold disagree, they MUST completely drop the session 462 and the router MUST flush all data learned from that cache. 464 Length: A 32 bit ordinal which has as its value the count of the 465 bytes in the entire PDU, including the eight bytes of header which 466 end with the length field. 468 Flags: The lowest order bit of the Flags field is 1 for an 469 announcement and 0 for a withdrawal, whether this PDU announces a 470 new right to announce the prefix or withdraws a previously 471 announced right. A withdraw effectively deletes one previously 472 announced IPvX Prefix PDU with the exact same Prefix, Length, Max- 473 Len, and ASN. 475 Prefix Length: An ordinal denoting the shortest prefix allowed for 476 the prefix. 478 Max Length: An ordinal denoting the longest prefix allowed by the 479 prefix. This MUST NOT be less than the Prefix Length element. 481 Prefix: The IPv4 or IPv6 prefix of the ROA. 483 Autonomous System Number: ASN allowed to announce this prefix, a 32 484 bit ordinal. 486 Zero: Fields shown as zero or reserved MUST be zero. The value of 487 such a field MUST be ignored on receipt. 489 5. Protocol Sequences 491 The sequences of PDU transmissions fall into three conversations as 492 follows: 494 5.1. Start or Restart 496 Cache Router 497 ~ ~ 498 | <----- Reset Query -------- | R requests data (or Serial Query) 499 | | 500 | ----- Cache Response -----> | C confirms request 501 | ------- IPvX Prefix ------> | C sends zero or more 502 | ------- IPvX Prefix ------> | IPv4 and IPv6 Prefix 503 | ------- IPvX Prefix ------> | Payload PDUs 504 | ------ End of Data ------> | C sends End of Data 505 | | and sends new serial 506 ~ ~ 508 When a transport session is first established, the router MAY send a 509 Reset Query and the cache responds with a data sequence of all data 510 it contains. 512 Alternatively, if the router has significant unexpired data from a 513 broken session with the same cache, it MAY start with a Serial Query 514 containing the Cache Nonce from the previous session to ensure the 515 serial numbers are comensurate. 517 This Reset Query sequence is also used when the router receives a 518 Cache Reset, chooses a new cache, or fears that it has otherwise lost 519 its way. 521 To limit the length of time a cache must keep the data necessary to 522 generate incremental updates, a router MUST send either a Serial 523 Query or a Reset Query no less frequently than once an hour. This 524 also acts as a keep alive at the application layer. 526 As the cache MAY not keep updates for more than one hour, the router 527 MUST have a polling interval of no greater than half an hour 529 5.2. Typical Exchange 531 Cache Router 532 ~ ~ 533 | -------- Notify ----------> | (optional) 534 | | 535 | <----- Serial Query ------- | R requests data 536 | | 537 | ----- Cache Response -----> | C confirms request 538 | ------- IPvX Prefix ------> | C sends zero or more 539 | ------- IPvX Prefix ------> | IPv4 and IPv6 Prefix 540 | ------- IPvX Prefix ------> | Payload PDUs 541 | ------ End of Data ------> | C sends End of Data 542 | | and sends new serial 543 ~ ~ 545 The cache server SHOULD send a notify PDU with its current serial 546 number when the cache's serial changes, with the expectation that the 547 router MAY then issue a serial query earlier than it otherwise might. 548 This is analogous to DNS NOTIFY in [RFC1996]. The cache SHOULD rate 549 limit Serial Notifies to no more frequently than one per minute. 551 When the transport layer is up and either a timer has gone off in the 552 router, or the cache has sent a Notify, the router queries for new 553 data by sending a Serial Query, and the cache sends all data newer 554 than the serial in the Serial Query. 556 To limit the length of time a cache must keep old withdraws, a router 557 MUST send either a Serial Query or a Reset Query no less frequently 558 than once an hour. 560 5.3. No Incremental Update Available 562 Cache Router 563 ~ ~ 564 | <----- Serial Query ------ | R requests data 565 | ------- Cache Reset ------> | C cannot supply update 566 | | from specified serial 567 | <------ Reset Query ------- | R requests new data 568 | ----- Cache Response -----> | C confirms request 569 | ------- IPvX Prefix ------> | C sends zero or more 570 | ------- IPvX Prefix ------> | IPv4 and IPv6 Prefix 571 | ------- IPvX Prefix ------> | Payload PDUs 572 | ------ End of Data ------> | C sends End of Data 573 | | and sends new serial 574 ~ ~ 576 The cache may respond to a Serial Query with a Cache Reset, informing 577 the router that the cache cannot supply an incremental update from 578 the serial number specified by the router. This might be because the 579 cache has lost state, or because the router has waited too long 580 between polls and the cache has cleaned up old data that it no longer 581 believes it needs, or because the cache has run out of storage space 582 and had to expire some old data early. Regardless of how this state 583 arose, the cache replies with a Cache Reset to tell the router that 584 it cannot honor the request. When a router receives this, the router 585 SHOULD attempt to connect to any more preferred caches in its cache 586 list. If there are no more preferred caches it MUST issue a Reset 587 Query and get an entire new load from the cache. 589 5.4. Cache has No Data Available 591 Cache Router 592 ~ ~ 593 | <----- Serial Query ------ | R requests data 594 | ---- Error Report PDU ----> | C cannot supply update 595 ~ ~ 597 Cache Router 598 ~ ~ 599 | <----- Reset Query ------- | R requests data 600 | ---- Error Report PDU ----> | C cannot supply update 601 ~ ~ 603 The cache may respond to either a Serial Query or a Reset Query 604 informing the router that the cache cannot supply any update at all. 605 The most likely cause is that the cache has lost state, perhaps due 606 to a restart, and has not yet recovered. While it is possible that a 607 cache might go into such a state without dropping any of its active 608 sessions, a router is more likely to see this behavior when it 609 initially connects and issues a Reset Query while the cache is still 610 rebuilding its database. 612 When a router receives this kind of error, the router SHOULD attempt 613 to connect to any other caches in its cache list, in preference 614 order. If no other caches are available, the router MUST issue 615 periodic Reset Queries until it gets a new usable load from the 616 cache. 618 6. SSH Transport 620 The transport layer session between a router and a cache carries the 621 binary Protocol Data Units (PDUs) in a persistent SSH session. 623 To run over SSH, the client router first establishes an SSH transport 624 connection using the SSH transport protocol, and the client and 625 server exchange keys for message integrity and encryption. The 626 client then invokes the "ssh-userauth" service to authenticate the 627 application, as described in the SSH authentication protocol RFC 4252 628 [RFC4252]. Once the application has been successfully authenticated, 629 the client invokes the "ssh-connection" service, also known as the 630 SSH connection protocol. 632 After the ssh-connection service is established, the client opens a 633 channel of type "session", which results in an SSH session. 635 Once the SSH session has been established, the application invokes 636 the application transport as an SSH subsystem called "rpki-rtr". 637 Subsystem support is a feature of SSH version 2 (SSHv2) and is not 638 included in SSHv1. Running this protocol as an SSH subsystem avoids 639 the need for the application to recognize shell prompts or skip over 640 extraneous information, such as a system message that is sent at 641 shell start-up. 643 It is assumed that the router and cache have exchanged keys out of 644 band by some reasonably secured means. 646 7. Router-Cache Set-Up 648 A cache has the public authentication data for each router it is 649 configured to support. 651 A router may be configured to peer with a selection of caches, and a 652 cache may be configured to support a selection of routers. Each must 653 have the name of, and authentication data for, each peer. In 654 addition, in a router, this list has a non-unique preference value 655 for each server in order of preference. This preference merely 656 denotes proximity, not trust, preferred belief, etc. The client 657 router attempts to establish a session with each potential serving 658 cache in preference order, and then starts to load data from the most 659 preferred cache to which it can connect and authenticate. The 660 router's list of caches has the following elements: 662 Preference: An ordinal denoting the router's preference to connect 663 to that cache, the lower the value the more preferred. 665 Name: The IP Address or fully qualified domain name of the cache. 667 Key: The public ssh key of the cache. 669 MyKey: The private ssh key of this client. 671 Due to the distributed nature of the RPKI, caches simply can not be 672 rigorously synchronous. A client may hold data from multiple caches, 673 but MUST keep the data marked as to source, as later updates MUST 674 affect the correct data. 676 Just as there may be more than one covering ROA from a single cache, 677 there may be multiple covering ROAs from multiple caches. The 678 results are as described in [I-D.ietf-sidr-roa-validation]. 680 If data from multiple caches are held, implementations MUST NOT 681 distinguish between data sources when performing validation. 683 When a more preferred cache becomes available, if resources allow, it 684 would be prudent for the client to start fetching from that cache. 686 The client SHOULD attempt to maintain at least one set of data, 687 regardless of whether it has chosen a different cache or established 688 a new connection to the previous cache. 690 A client MAY drop the data from a particular cache when it is fully 691 in synch with one or more other caches. 693 A client SHOULD delete the data from a cache when it has been unable 694 to refresh from that cache for a configurable timer value. The 695 default for that value is twice the polling period for that cache. 697 If a client loses connectivity to a cache it is using, or otherwise 698 decides to switch to a new cache, it SHOULD retain the data from the 699 previous cache until it has a full set of data from one or more other 700 caches. Note that this may already be true at the point of 701 connection loss if the client has connections to more than one cache. 703 8. Deployment Scenarios 705 For illustration, we present three likely deployment scenarios. 707 Small End Site: The small multi-homed end site may wish to outsource 708 the RPKI cache to one or more of their upstream ISPs. They would 709 exchange authentication material with the ISP using some out of 710 band mechanism, and their router(s) would connect to one or more 711 up-streams' caches. The ISPs would likely deploy caches intended 712 for customer use separately from the caches with which their own 713 BGP speakers peer. 715 Large End Site: A larger multi-homed end site might run one or more 716 caches, arranging them in a hierarchy of client caches, each 717 fetching from a serving cache which is closer to the global RPKI. 718 They might configure fall-back peerings to up-stream ISP caches. 720 ISP Backbone: A large ISP would likely have one or more redundant 721 caches in each major PoP, and these caches would fetch from each 722 other in an ISP-dependent topology so as not to place undue load 723 on the global RPKI publication infrastructure. 725 Experience with large DNS cache deployments has shown that complex 726 topologies are ill-advised as it is easy to make errors in the graph, 727 e.g. not maintaining a loop-free condition. 729 Of course, these are illustrations and there are other possible 730 deployment strategies. It is expected that minimizing load on the 731 global RPKI servers will be a major consideration. 733 To keep load on global RPKI services from unnecessary peaks, it is 734 recommended that primary caches which load from the distributed 735 global RPKI not do so all at the same times, e.g. on the hour. 736 Choose a random time, perhaps the ISP's AS number modulo 60 and 737 jitter the inter-fetch timing. 739 9. Error Codes 741 This section contains a preliminary list of error codes. The authors 742 expect additions to this section during development of the initial 743 implementations. Eventually, these error codes will probably need to 744 reside in an IANA registry. 746 0: Reserved. 748 1: Internal Error: The party reporting the error experienced some 749 kind of internal error unrelated to protocol operation (ran out of 750 memory, a coding assertion failed, et cetera). 752 2: No Data Available: The cache believes itself to be in good 753 working order, but is unable to answer either a Serial Query or a 754 Reset Query because it has no useful data available at this time. 755 This is likely to be a temporary error, and most likely indicates 756 that the cache has not yet completed pulling down an initial 757 current data set from the global RPKI system after some kind of 758 event that invalidated whatever data it might have previously held 759 (reboot, network partition, etcetera). 761 3: Invalid Request: The cache server believes the client's request 762 to be invalid. 764 10. Security Considerations 766 As this document describes a security protocol, many aspects of 767 security interest are described in the relevant sections. This 768 section points out issues which may not be obvious in other sections. 770 Cache Validation: In order for a collection of caches as described 771 in Section 8 to guarantee a consistent view, they need to be given 772 consistent trust anchors to use in their internal validation 773 process. Distribution of a consistent trust anchor is assumed to 774 be out of band. 776 Cache Peer Identification: The router initiates an ssh transport 777 session to a cache, which it identifies by either IP address or 778 fully qualified domain name. Be aware that a DNS or address 779 spoofing attack could make the correct cache unreachable. No 780 session would be established, as the authorization keys would not 781 match. 783 Transport Security: The RPKI relies on object, not server or 784 transport, trust. I.e. the IANA root trust anchor is distributed 785 to all caches through some out of band means, and can then be used 786 by each cache to validate certificates and ROAs all the way down 787 the tree. The inter-cache relationships are based on this object 788 security model, hence the inter-cache transport can be lightly 789 protected. 791 But this protocol document assumes that the routers can not do the 792 validation cryptography. Hence the last link, from cache to 793 router, is secured by server authentication and transport level 794 security. This is dangerous, as server authentication and 795 transport have very different threat models than object security. 797 So the strength of the trust relationship and the transport 798 between the router(s) and the cache(s) are critical. You're 799 betting your routing on this. 801 While we can not say the cache must be on the same LAN, if only 802 due to the issue of an enterprise wanting to off-load the cache 803 task to their upstream ISP(s), locality, trust, and control are 804 very critical issues here. The cache(s) really SHOULD be as 805 close, in the sense of controlled and protected (against DDoS, 806 MITM) transport, to the router(s) as possible. It also SHOULD be 807 topologically close so that a minimum of validated routing data 808 are needed to bootstrap a router's access to a cache. 810 11. Glossary 812 The following terms are used with special meaning: 814 Global RPKI: The authoritative data of the RPKI are published in a 815 distributed set of servers at the IANA, RIRs, NIRs, and ISPs, see 816 [I-D.ietf-sidr-repos-struct]. 818 Non-authorative Cache: A coalesced copy of the RPKI which is 819 periodically fetched/refreshed directly or indirectly from the 820 global RPKI using the [RFC5781] protocol/tools 822 Cache: Relying party update sofcware such as rcynic is used to 823 gather and validate the distributed data of the RPKI into a cache. 824 Trusting this cache further is a matter between the provider of 825 the cache and a relying party. 827 Serial Number: A 32-bit monotonically increasing ordinal which wraps 828 from 2^32-1 to 0. It denotes the logical version of a cache. A 829 cache increments the value by one when it successfully updates its 830 data from a parent cache or from primary RPKI data. As a cache is 831 receiving, new incoming data, and implicit deletes, are marked 832 with the new serial but MUST not be sent until the fetch is 833 complete. A serial number is not commensurate between caches, nor 834 need it be maintained across resets of the cache server. See 835 [RFC1982] on DNS Serial Number Arithmetic for too much detail on 836 serial number arithmetic. 838 12. IANA Considerations 840 This document requests the IANA to create a registry for PDU types. 842 This document requests the IANA to create a registry for Error Codes. 844 In addition, a registry for Version Numbers would be needed if new 845 Version Number is defined in a new RFC. 847 Note to RFC Editor: this section may be replaced on publication as an 848 RFC. 850 13. Acknowledgments 852 The authors wish to thank Steve Bellovin, Rex Fernando, Russ Housley, 853 Pradosh Mohapatra, Keyur Patel, Sandy Murphy, Robert Raszuk, John 854 Scudder, Ruediger Volk, and David Ward. Particular thanks go to 855 Hannes Gredler for showing us the dangers of unnecessary fields. 857 14. References 859 14.1. Normative References 861 [I-D.ietf-sidr-roa-validation] 862 Huston, G. and G. Michaelson, "Validation of Route 863 Origination using the Resource Certificate PKI and ROAs", 864 draft-ietf-sidr-roa-validation-10 (work in progress), 865 November 2010. 867 [RFC1982] Elz, R. and R. Bush, "Serial Number Arithmetic", RFC 1982, 868 August 1996. 870 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 871 Requirement Levels", BCP 14, RFC 2119, March 1997. 873 [RFC4252] Ylonen, T. and C. Lonvick, "The Secure Shell (SSH) 874 Authentication Protocol", RFC 4252, January 2006. 876 [RFC5781] Weiler, S., Ward, D., and R. Housley, "The rsync URI 877 Scheme", RFC 5781, February 2010. 879 14.2. Informative References 881 [I-D.ietf-sidr-arch] 882 Lepinski, M. and S. Kent, "An Infrastructure to Support 883 Secure Internet Routing", draft-ietf-sidr-arch-11 (work in 884 progress), September 2010. 886 [I-D.ietf-sidr-repos-struct] 887 Huston, G., Loomans, R., and G. Michaelson, "A Profile for 888 Resource Certificate Repository Structure", 889 draft-ietf-sidr-repos-struct-06 (work in progress), 890 November 2010. 892 [RFC1996] Vixie, P., "A Mechanism for Prompt Notification of Zone 893 Changes (DNS NOTIFY)", RFC 1996, August 1996. 895 Authors' Addresses 897 Randy Bush 898 Internet Initiative Japan, Inc. 899 5147 Crystal Springs 900 Bainbridge Island, Washington 98110 901 US 903 Phone: +1 206 780 0431 x1 904 Email: randy@psg.com 906 Rob Austein 907 Internet Systems Consortium 908 950 Charter Street 909 Redwood City, CA 94063 910 USA 912 Email: sra@isc.org