idnits 2.17.1 draft-ietf-dprive-dnsoquic-06.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == The document seems to lack the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords -- however, there's a paragraph with a matching beginning. Boilerplate error? (The document does seem to have the reference to RFC 2119 which the ID-Checklist requires). -- The document date (20 October 2021) is 918 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(See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Outdated reference: A later version (-10) exists of draft-ietf-dnsop-rfc8499bis-03 ** Downref: Normative reference to an Experimental RFC: RFC 8094 ** Downref: Normative reference to an Experimental RFC: RFC 8467 Summary: 2 errors (**), 0 flaws (~~), 3 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group C. Huitema 3 Internet-Draft Private Octopus Inc. 4 Intended status: Standards Track S. Dickinson 5 Expires: 23 April 2022 Sinodun IT 6 A. Mankin 7 Salesforce 8 20 October 2021 10 DNS over Dedicated QUIC Connections 11 draft-ietf-dprive-dnsoquic-06 13 Abstract 15 This document describes the use of QUIC to provide transport privacy 16 for DNS. The encryption provided by QUIC has similar properties to 17 that provided by TLS, while QUIC transport eliminates the head-of- 18 line blocking issues inherent with TCP and provides more efficient 19 packet loss recovery than UDP. DNS over QUIC (DoQ) has privacy 20 properties similar to DNS over TLS (DoT) specified in RFC7858, and 21 latency characteristics similar to classic DNS over UDP. 23 Status of This Memo 25 This Internet-Draft is submitted in full conformance with the 26 provisions of BCP 78 and BCP 79. 28 Internet-Drafts are working documents of the Internet Engineering 29 Task Force (IETF). Note that other groups may also distribute 30 working documents as Internet-Drafts. The list of current Internet- 31 Drafts is at https://datatracker.ietf.org/drafts/current/. 33 Internet-Drafts are draft documents valid for a maximum of six months 34 and may be updated, replaced, or obsoleted by other documents at any 35 time. It is inappropriate to use Internet-Drafts as reference 36 material or to cite them other than as "work in progress." 38 This Internet-Draft will expire on 23 April 2022. 40 Copyright Notice 42 Copyright (c) 2021 IETF Trust and the persons identified as the 43 document authors. All rights reserved. 45 This document is subject to BCP 78 and the IETF Trust's Legal 46 Provisions Relating to IETF Documents (https://trustee.ietf.org/ 47 license-info) in effect on the date of publication of this document. 48 Please review these documents carefully, as they describe your rights 49 and restrictions with respect to this document. Code Components 50 extracted from this document must include Simplified BSD License text 51 as described in Section 4.e of the Trust Legal Provisions and are 52 provided without warranty as described in the Simplified BSD License. 54 Table of Contents 56 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 57 2. Key Words . . . . . . . . . . . . . . . . . . . . . . . . . . 4 58 3. Document work via GitHub . . . . . . . . . . . . . . . . . . 4 59 4. Design Considerations . . . . . . . . . . . . . . . . . . . . 4 60 4.1. Provide DNS Privacy . . . . . . . . . . . . . . . . . . . 5 61 4.2. Design for Minimum Latency . . . . . . . . . . . . . . . 5 62 4.3. No Specific Middlebox Bypass Mechanism . . . . . . . . . 6 63 4.4. No Server Initiated Transactions . . . . . . . . . . . . 6 64 5. Specifications . . . . . . . . . . . . . . . . . . . . . . . 6 65 5.1. Connection Establishment . . . . . . . . . . . . . . . . 6 66 5.1.1. Draft Version Identification . . . . . . . . . . . . 6 67 5.1.2. Port Selection . . . . . . . . . . . . . . . . . . . 7 68 5.2. Stream Mapping and Usage . . . . . . . . . . . . . . . . 7 69 5.2.1. DNS Message IDs . . . . . . . . . . . . . . . . . . . 8 70 5.3. DoQ Error Codes . . . . . . . . . . . . . . . . . . . . . 8 71 5.3.1. Transaction Cancellation . . . . . . . . . . . . . . 9 72 5.3.2. Transaction Errors . . . . . . . . . . . . . . . . . 9 73 5.3.3. Protocol Errors . . . . . . . . . . . . . . . . . . . 10 74 5.3.4. Alternative error codes . . . . . . . . . . . . . . . 11 75 5.4. Connection Management . . . . . . . . . . . . . . . . . . 11 76 5.5. Session Resumption and 0-RTT . . . . . . . . . . . . . . 12 77 5.6. Message Sizes . . . . . . . . . . . . . . . . . . . . . . 12 78 6. Implementation Requirements . . . . . . . . . . . . . . . . . 13 79 6.1. Authentication . . . . . . . . . . . . . . . . . . . . . 13 80 6.2. Fallback to Other Protocols on Connection Failure . . . . 13 81 6.3. Address Validation . . . . . . . . . . . . . . . . . . . 13 82 6.4. Padding . . . . . . . . . . . . . . . . . . . . . . . . . 14 83 6.5. Connection Handling . . . . . . . . . . . . . . . . . . . 15 84 6.5.1. Connection Reuse . . . . . . . . . . . . . . . . . . 15 85 6.5.2. Resource Management . . . . . . . . . . . . . . . . . 15 86 6.5.3. Using 0-RTT and Session Resumption . . . . . . . . . 16 87 6.5.4. Controlling Connection Migration For Privacy . . . . 16 88 6.6. Processing Queries in Parallel . . . . . . . . . . . . . 17 89 6.7. Zone transfer . . . . . . . . . . . . . . . . . . . . . . 17 90 6.8. Flow Control Mechanisms . . . . . . . . . . . . . . . . . 17 91 7. Implementation Status . . . . . . . . . . . . . . . . . . . . 18 92 7.1. Performance Measurements . . . . . . . . . . . . . . . . 19 93 8. Security Considerations . . . . . . . . . . . . . . . . . . . 19 94 9. Privacy Considerations . . . . . . . . . . . . . . . . . . . 19 95 9.1. Privacy Issues With 0-RTT data . . . . . . . . . . . . . 20 96 9.2. Privacy Issues With Session Resumption . . . . . . . . . 20 97 9.3. Privacy Issues With Address Validation Tokens . . . . . . 21 98 9.4. Privacy Issues With Long Duration Sessions . . . . . . . 22 99 9.5. Traffic Analysis . . . . . . . . . . . . . . . . . . . . 22 100 10. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 22 101 10.1. Registration of DoQ Identification String . . . . . . . 22 102 10.2. Reservation of Dedicated Port . . . . . . . . . . . . . 23 103 10.2.1. Port number 784 for experimentations . . . . . . . . 23 104 10.3. Reservation of Extended DNS Error Code Too Early . . . . 23 105 10.4. DNS over QUIC Error Codes Registry . . . . . . . . . . . 24 106 11. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 25 107 12. References . . . . . . . . . . . . . . . . . . . . . . . . . 26 108 12.1. Normative References . . . . . . . . . . . . . . . . . . 26 109 12.2. Informative References . . . . . . . . . . . . . . . . . 28 110 Appendix A. The NOTIFY Service . . . . . . . . . . . . . . . . . 29 111 Appendix B. Notable Changes From Previous Versions . . . . . . . 30 112 B.1. Stream Mapping Incompatibility With Draft-02 . . . . . . 30 113 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 30 115 1. Introduction 117 Domain Name System (DNS) concepts are specified in "Domain names - 118 concepts and facilities" [RFC1034]. The transmission of DNS queries 119 and responses over UDP and TCP is specified in "Domain names - 120 implementation and specification" [RFC1035]. 122 This document presents a mapping of the DNS protocol over the QUIC 123 transport [RFC9000] [RFC9001]. DNS over QUIC is referred here as 124 DoQ, in line with "DNS Terminology" [I-D.ietf-dnsop-rfc8499bis]. 126 The goals of the DoQ mapping are: 128 1. Provide the same DNS privacy protection as DNS over TLS (DoT) 129 [RFC7858]. This includes an option for the client to 130 authenticate the server by means of an authentication domain name 131 as specified in "Usage Profiles for DNS over TLS and DNS over 132 DTLS" [RFC8310]. 134 2. Provide an improved level of source address validation for DNS 135 servers compared to classic DNS over UDP. 137 3. Provide a transport that is not constrained by path MTU 138 limitations on the size of DNS responses it can send. 140 In order to achieve these goals, and to support ongoing work on 141 encryption of DNS, the scope of this document includes 143 * the "stub to recursive resolver" scenario 144 * the "recursive resolver to authoritative nameserver" scenario and 146 * the "nameserver to nameserver" scenario (mainly used for zone 147 transfers (XFR) [RFC1995], [RFC5936]). 149 In other words, this document is intended to specify QUIC as a 150 general purpose transport for DNS. 152 The specific non-goals of this document are: 154 1. No attempt is made to evade potential blocking of DNS over QUIC 155 traffic by middleboxes. 157 2. No attempt to support server initiated transactions, which are 158 used only in DNS Stateful Operations (DSO) [RFC8490]. 160 Specifying the transmission of an application over QUIC requires 161 specifying how the application's messages are mapped to QUIC streams, 162 and generally how the application will use QUIC. This is done for 163 HTTP in "Hypertext Transfer Protocol Version 3 164 (HTTP/3)"[I-D.ietf-quic-http]. The purpose of this document is to 165 define the way DNS messages can be transmitted over QUIC. 167 In this document, Section 4 presents the reasoning that guided the 168 proposed design. Section 5 specifies the actual mapping of DoQ. 169 Section 6 presents guidelines on the implementation, usage and 170 deployment of DoQ. 172 2. Key Words 174 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 175 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 176 document are to be interpreted as described in BCP 14 [RFC8174]. 178 3. Document work via GitHub 180 (RFC EDITOR NOTE: THIS SECTION TO BE REMOVED BEFORE PUBLICATION)The 181 Github repository for this document is at https://github.com/huitema/ 182 dnsoquic. Proposed text and editorial changes are very much welcomed 183 there, but any functional changes should always first be discussed on 184 the IETF DPRIVE WG (dns-privacy) mailing list. 186 4. Design Considerations 188 This section and its subsections present the design guidelines that 189 were used for DoQ. This section is informative in nature. 191 4.1. Provide DNS Privacy 193 DoT [RFC7858] defines how to mitigate some of the issues described in 194 "DNS Privacy Considerations" [RFC9076] by specifying how to transmit 195 DNS messages over TLS. The "Usage Profiles for DNS over TLS and DNS 196 over DTLS" [RFC8310] specify Strict and Opportunistic Usage Profiles 197 for DoT including how stub resolvers can authenticate recursive 198 resolvers. 200 QUIC connection setup includes the negotiation of security parameters 201 using TLS, as specified in "Using TLS to Secure QUIC" [RFC9001], 202 enabling encryption of the QUIC transport. Transmitting DNS messages 203 over QUIC will provide essentially the same privacy protections as 204 DoT [RFC7858] including Strict and Opportunistic Usage Profiles 205 [RFC8310]. Further discussion on this is provided in Section 9. 207 4.2. Design for Minimum Latency 209 QUIC is specifically designed to reduce protocol-induced delays, with 210 features such as: 212 1. Support for 0-RTT data during session resumption. 214 2. Support for advanced packet loss recovery procedures as specified 215 in "QUIC Loss Detection and Congestion Control" [RFC9002]. 217 3. Mitigation of head-of-line blocking by allowing parallel delivery 218 of data on multiple streams. 220 This mapping of DNS to QUIC will take advantage of these features in 221 three ways: 223 1. Optional support for sending 0-RTT data during session resumption 224 (the security and privacy implications of this are discussed in 225 later sections). 227 2. Long-lived QUIC connections over which multiple DNS transactions 228 are performed, generating the sustained traffic required to 229 benefit from advanced recovery features. 231 3. Mapping of each DNS Query/Response transaction to a separate 232 stream, to mitigate head-of-line blocking. This enables servers 233 to respond to queries "out of order". It also enables clients to 234 process responses as soon as they arrive, without having to wait 235 for in order delivery of responses previously posted by the 236 server. 238 These considerations are reflected in the mapping of DNS traffic to 239 QUIC streams in Section 5.2. 241 4.3. No Specific Middlebox Bypass Mechanism 243 The mapping of DoQ is defined for minimal overhead and maximum 244 performance. This means a different traffic profile than HTTP3 over 245 QUIC. This difference can be noted by firewalls and middleboxes. 246 There may be environments in which HTTP3 over QUIC will be able to 247 pass through, but DoQ will be blocked by these middle boxes. 249 4.4. No Server Initiated Transactions 251 As stated in Section 1, this document does not specify support for 252 server initiated transactions within established DoQ connections. 253 That is, only the initiator of the DoQ connection may send queries 254 over the connection. 256 DSO does support server-initiated transactions within existing 257 connections. However DoQ as defined here does not meet the criteria 258 for an applicable transport for DSO because it does not guarantee in- 259 order delivery of messages, see Section 4.2 of [RFC8490]. 261 5. Specifications 263 5.1. Connection Establishment 265 DoQ connections are established as described in the QUIC transport 266 specification [RFC9000]. During connection establishment, DoQ 267 support is indicated by selecting the ALPN token "doq" in the crypto 268 handshake. 270 5.1.1. Draft Version Identification 272 (RFC EDITOR NOTE: THIS SECTION TO BE REMOVED BEFORE PUBLICATION) Only 273 implementations of the final, published RFC can identify themselves 274 as "doq". Until such an RFC exists, implementations MUST NOT 275 identify themselves using this string. 277 Implementations of draft versions of the protocol MUST add the string 278 "-" and the corresponding draft number to the identifier. For 279 example, draft-ietf-dprive-dnsoquic-00 is identified using the string 280 "doq-i00". 282 5.1.2. Port Selection 284 By default, a DNS server that supports DoQ MUST listen for and accept 285 QUIC connections on the dedicated UDP port TBD (number to be defined 286 in Section 10), unless there is a mutual agreement to use another 287 port. 289 By default, a DNS client desiring to use DoQ with a particular server 290 MUST establish a QUIC connection to UDP port TBD on the server, 291 unless there is a mutual agreement to use another port. 293 In order to use a port other than TBD, both clients and servers would 294 need a configuration option in their software. 296 DoQ connections MUST NOT use UDP port 53. This recommendation 297 against use of port 53 for DoQ is to avoid confusion between DoQ and 298 the use of DNS over UDP [RFC1035]. 300 In the stub to recursive scenario, the use of port 443 as a mutually 301 agreed alternative port can be operationally beneficial, since port 302 443 is less likely to be blocked than other ports. Several 303 mechanisms for stubs to discover recursives offering encrypted 304 transports, including the use of custom ports, are the subject of 305 ongoing work. 307 5.2. Stream Mapping and Usage 309 The mapping of DNS traffic over QUIC streams takes advantage of the 310 QUIC stream features detailed in Section 2 of [RFC9000], the QUIC 311 transport specification. 313 DNS traffic follows a simple pattern in which the client sends a 314 query, and the server provides one or more responses (multiple 315 responses can occur in zone transfers). 317 The mapping specified here requires that the client selects a 318 separate QUIC stream for each query. The server then uses the same 319 stream to provide all the response messages for that query. In order 320 that multiple responses can be parsed, a 2-octet length field is used 321 in exactly the same way as the 2-octet length field defined for DNS 322 over TCP [RFC1035]. The practical result of this is that the content 323 of each QUIC stream is exactly the same as the content of a TCP 324 connection that would manage exactly one query. 326 All DNS messages (queries and responses) sent over DoQ connections 327 MUST be encoded as a 2-octet length field followed by the message 328 content as specified in [RFC1035]. 330 The client MUST select the next available client-initiated 331 bidirectional stream for each subsequent query on a QUIC connection, 332 in conformance with the QUIC transport specification [RFC9000]. 334 The client MUST send the DNS query over the selected stream, and MUST 335 indicate through the STREAM FIN mechanism that no further data will 336 be sent on that stream. 338 The server MUST send the response(s) on the same stream and MUST 339 indicate, after the last response, through the STREAM FIN mechanism 340 that no further data will be sent on that stream. 342 Therefore, a single client initiated DNS transaction consumes a 343 single stream. This means that the client's first query occurs on 344 QUIC stream 0, the second on 4, and so on. 346 Servers MAY defer processing of a query until the STREAM FIN has been 347 indicated on the stream selected by the client. Servers and clients 348 MAY monitor the number of "dangling" streams for which the expected 349 queries or responses have been received but not the STREAM FIN. 350 Implementations MAY impose a limit on the number of such dangling 351 streams. If limits are encountered, implementations MAY close the 352 connection. 354 5.2.1. DNS Message IDs 356 When sending queries over a QUIC connection, the DNS Message ID MUST 357 be set to zero. 359 This has implications for proxying DoQ message to and from other 360 transports. For example, proxies may have to manage the fact that 361 DoQ can support a larger number of outstanding queries on a single 362 connection than e.g., DNS over TCP because DoQ is not limited by the 363 Message ID space. 365 When forwarding a DNS message from DoQ over another transport, a DNS 366 Message ID MUST be generated according to the rules of the protocol 367 that is in use. When forwarding a DNS message from another transport 368 over DoQ, the Message ID MUST be set to zero. 370 5.3. DoQ Error Codes 372 The following error codes are defined for use when abruptly 373 terminating streams, aborting reading of streams, or immediately 374 closing connections: 376 DOQ_NO_ERROR (0x0): No error. This is used when the connection or 377 stream needs to be closed, but there is no error to signal. 379 DOQ_INTERNAL_ERROR (0x1): The DoQ implementation encountered an 380 internal error and is incapable of pursuing the transaction or the 381 connection. 383 DOQ_PROTOCOL_ERROR (0x2): The DoQ implementation encountered an 384 protocol error and is forcibly aborting the connection. 386 DOQ_REQUEST_CANCELLED (0x3): A DoQ client uses this to signal that 387 it wants to cancel an outstanding transaction. 389 DOQ_EXCESSIVE_LOAD (0x4): A DoQ implementation uses this to signal 390 when closing a connection due to excessive load. 392 DOQ_ERROR_RESERVED (0xd098ea5e): Alternative error code used for 393 tests. 395 See Section 10.4 for details on registering new error codes. 397 5.3.1. Transaction Cancellation 399 In QUIC, sending STOP_SENDING requests that a peer cease transmission 400 on a stream. If a DoQ client wishes to cancel an outstanding 401 request, it MUST issue a QUIC Stop Sending with error code 402 DOQ_REQUEST_CANCELLED. This may be sent at any time but will be 403 ignored if the server has already sent the response. The 404 corresponding DNS transaction MUST be abandoned. 406 Servers that receive STOP_SENDING act in accordance with Section 3.5 407 of [RFC9000]. Servers MAY impose implementation limits on the total 408 number or rate of request cancellations. If limits are encountered, 409 servers MAY close the connection. In this case, servers wanting to 410 help client debugging MAY use the error code DOQ_EXCESSIVE_LOAD. 411 There is always a trade-off between helping good faith clients debug 412 issues and allowing denial-of-service attackers to test server 413 defenses, so depending on circumstances servers might very well chose 414 to send different error codes. 416 Note that this mechanism provides a way for secondaries to cancel a 417 single zone transfer occurring on a given stream without having to 418 close the QUIC connection. 420 5.3.2. Transaction Errors 422 Servers normally complete transactions by sending a DNS response (or 423 responses) on the transaction's stream, including cases where the DNS 424 response indicates a DNS error. For example, a Server Failure 425 (SERVFAIL, [RFC1035]) SHOULD be notified to the client by sending 426 back a response with the Response Code set to SERVFAIL. 428 If a server is incapable of sending a DNS response due to an internal 429 error, it SHOULD issue a QUIC Stream Reset. The error code SHOULD be 430 set to DOQ_INTERNAL_ERROR. The corresponding DNS transaction MUST be 431 abandoned. Clients MAY limit the number of unsolicited QUIC Stream 432 Resets received on a connection before choosing to close the 433 connection. 435 Note that this mechanism provides a way for primaries to abort a 436 single zone transfer occurring on a given stream without having to 437 close the QUIC connection. 439 5.3.3. Protocol Errors 441 Other error scenarios can occur due to malformed, incomplete or 442 unexpected messages during a transaction. These include (but are not 443 limited to) 445 * a client or server receives a message with a non-zero Message ID 447 * a client or server receives a STREAM FIN before receiving all the 448 bytes for a message indicated in the 2-octet length field 450 * a client receives a STREAM FIN before receiving all the expected 451 responses 453 * a server receives more than one query on a stream 455 * a client receives a different number of responses on a stream than 456 expected (e.g. multiple responses to a query for an A record) 458 * a client receives a STOP_SENDING request 460 * the client or server does not indicate the expected STREAM FIN 461 after sending requests or responses (see Section 5.2). 463 * an implementation receives a message containing the edns-tcp- 464 keepalive EDNS(0) Option [RFC7828] (see Section 6.5.2) 466 * a client or a server attempts to open an unidirectional QUIC 467 stream 469 * a server attempts to open a server-initiated bidirectional QUIC 470 stream 472 If a peer encounters such an error condition it is considered a fatal 473 error. It SHOULD forcibly abort the connection using QUIC's 474 CONNECTION_CLOSE mechanism, and SHOULD use the DoQ error code 475 DOQ_PROTOCOL_ERROR. 477 It is noted that the restrictions on use of the above EDNS(0) options 478 has implications for proxying message from TCP/DoT/DoH over DoQ. 480 5.3.4. Alternative error codes 482 This specification suggests specific error codes Section 5.3.1, 483 Section 5.3.2, and Section 5.3.3. These error codes are meant to 484 facilitates investigation of failures and other incidents. New error 485 codes may be defined in future versions of DoQ, or registered as 486 specified in Section 10.4. 488 Because new error codes can be defined without negotiation, use of an 489 error code in an unexpected context or receipt of an unknown error 490 code MUST be treated as equivalent to DOQ_NO_ERROR. 492 Implementations MAY wish to test the support for the error code 493 extension mechanism by using error codes not listed in this document, 494 or they MAY use DOQ_ERROR_RESERVED. 496 5.4. Connection Management 498 Section 10 of [RFC9000], the QUIC transport specification, specifies 499 that connections can be closed in three ways: 501 * idle timeout 503 * immediate close 505 * stateless reset 507 Clients and servers implementing DoQ SHOULD negotiate use of the idle 508 timeout. Closing on idle timeout is done without any packet 509 exchange, which minimizes protocol overhead. Per Section 10.1 of 510 [RFC9000], the QUIC transport specification, the effective value of 511 the idle timeout is computed as the minimum of the values advertised 512 by the two endpoints. Practical considerations on setting the idle 513 timeout are discussed in Section 6.5.2. 515 Clients SHOULD monitor the idle time incurred on their connection to 516 the server, defined by the time spent since the last packet from the 517 server has been received. When a client prepares to send a new DNS 518 query to the server, it will check whether the idle time is 519 sufficient lower than the idle timer. If it is, the client will send 520 the DNS query over the existing connection. If not, the client will 521 establish a new connection and send the query over that connection. 523 Clients MAY discard their connections to the server before the idle 524 timeout expires. A client that has outstanding queries SHOULD close 525 the connection explicitly using QUIC's CONNECTION_CLOSE mechanism and 526 the DoQ error code DOQ_NO_ERROR. 528 Clients and servers MAY close the connection for a variety of other 529 reasons, indicated using QUIC's CONNECTION_CLOSE. Client and servers 530 that send packets over a connection discarded by their peer MAY 531 receive a stateless reset indication. If a connection fails, all the 532 in progress transaction on that connection MUST be abandoned. 534 5.5. Session Resumption and 0-RTT 536 A client MAY take advantage of the session resumption mechanisms 537 supported by QUIC transport [RFC9000] and QUIC TLS [RFC9001]. 538 Clients SHOULD consider potential privacy issues associated with 539 session resumption before deciding to use this mechanism. These 540 privacy issues are detailed in Section 9.2 and Section 9.1, and the 541 implementation considerations are discussed in Section 6.5.3. 543 The 0-RTT mechanism SHOULD NOT be used to send DNS requests that are 544 not "replayable" transactions. In this specification, only 545 transactions that have an OPCODE of QUERY or NOTIFY are considered 546 replayable and MAY be sent in 0-RTT data. See Appendix A for a 547 detailed discussion of why NOTIFY is included here. 549 Servers MUST NOT execute non replayable transactions received in 550 0-RTT data. Servers MUST adopt one of the following behaviors: 552 * Queue the offending transaction and only execute it after the QUIC 553 handshake has been completed, as defined in Section 4.1.1 of 554 [RFC9001]. 556 * Reply to the offending transaction with a response code REFUSED 557 and an Extended DNS Error Code (EDE) "Too Early", see 558 Section 10.3. 560 * Close the connection with the error code DOQ_PROTOCOL_ERROR. 562 5.6. Message Sizes 564 DoQ Queries and Responses are sent on QUIC streams, which in theory 565 can carry up to 2^62 bytes. However, DNS messages are restricted in 566 practice to a maximum size of 65535 bytes. This maximum size is 567 enforced by the use of a two-octet message length field in DNS over 568 TCP [RFC1035] and DNS over TLS [RFC7858], and by the definition of 569 the "application/dns-message" for DNS over HTTP [RFC8484]. DoQ 570 enforces the same restriction. 572 The Extension Mechanisms for DNS (EDNS) [RFC6891] allow peers to 573 specify the UDP message size. This parameter is ignored by DoQ. DoQ 574 implementations always assume that the maximum message size is 65535 575 bytes. 577 6. Implementation Requirements 579 6.1. Authentication 581 For the stub to recursive resolver scenario, the authentication 582 requirements are the same as described in DoT [RFC7858] and "Usage 583 Profiles for DNS over TLS and DNS over DTLS" [RFC8310]. [RFC8932] 584 states that DNS privacy services SHOULD provide credentials that 585 clients can use to authenticate the server. Given this, and to align 586 with the authentication model for DoH, DoQ stubs SHOULD use a Strict 587 authentication profile. Client authentication for the encrypted stub 588 to recursive scenario is not described in any DNS RFC. 590 For zone transfer, the requirements are the same as described in 591 [RFC9103]. 593 For the recursive resolver to authoritative nameserver scenario, 594 authentication requirements are unspecified at the time of writing 595 and are the subject on ongoing work in the DPRIVE WG. 597 6.2. Fallback to Other Protocols on Connection Failure 599 If the establishment of the DoQ connection fails, clients MAY attempt 600 to fall back to DoT and then potentially clear text, as specified in 601 DoT [RFC7858] and "Usage Profiles for DNS over TLS and DNS over DTLS" 602 [RFC8310], depending on their privacy profile. 604 DNS clients SHOULD remember server IP addresses that don't support 605 DoQ. Timeouts, connection refusals, and QUIC handshake failures are 606 valid indicators that a server does not support DoQ. Clients SHOULD 607 NOT attempt DoQ queries to a server that does not support DoQ for a 608 reasonable period (such as one hour per server). DNS clients 609 following an out-of-band key-pinned privacy profile ([RFC7858]) MAY 610 be more aggressive about retrying DoQ connection failures. 612 6.3. Address Validation 614 Section 8 of [RFC9000], the QUIC transport specification, defines 615 Address Validation procedures to avoid servers being used in address 616 amplification attacks. DoQ implementations MUST conform to this 617 specification, which limits the worst case amplification to a factor 618 3. 620 DoQ implementations SHOULD consider configuring servers to use the 621 Address Validation using Retry Packets procedure defined in 622 Section 8.1.2 of [RFC9000], the QUIC transport specification. This 623 procedure imposes a 1-RTT delay for verifying the return routability 624 of the source address of a client, similar to the DNS Cookies 625 mechanism [RFC7873]. 627 DoQ implementations that configure Address Validation using Retry 628 Packets SHOULD implement the Address Validation for Future 629 Connections procedure defined in Section 8.1.3 of [RFC9000], the QUIC 630 transport specification. This defines how servers can send NEW_TOKEN 631 frames to clients after the client address is validated, in order to 632 avoid the 1-RTT penalty during subsequent connections by the client 633 from the same address. 635 6.4. Padding 637 Implementations SHOULD protect against the traffic analysis attacks 638 described in Section 9.5 by the judicious injection of padding. This 639 could be done either by padding individual DNS messages using the 640 EDNS(0) Padding Option [RFC7830] and by padding QUIC packets (see 641 Section 8.6 of [RFC9000], the QUIC transport specification. 643 In theory, padding at the QUIC level could result in better 644 performance for the equivalent protection, because the amount of 645 padding can take into account non-DNS frames such as acknowledgeemnts 646 or flow control updates, and also because QUIC packets can carry 647 multiple DNS messages. However, applications can only control the 648 amount of padding in QUIC packets if the implementation of QUIC 649 exposes adequate APIs. This leads to the following recommendation: 651 * if the implementation of QUIC exposes APIs to set a padding 652 policy, DNS over QUIC SHOULD use that API to align the packet 653 length to a small set of fixed sizes, aligned with the 654 recommendations of the "Padding Policies for Extension Mechanisms 655 for DNS (EDNS(0))" [RFC8467]. 657 * if padding at the QUIC level is not available or not used, DNS 658 over QUIC MUST ensure that all DNS queries and responses are 659 padded to a small set of fixed sizes, using the EDNS padding 660 extension as specified in "Padding Policies for Extension 661 Mechanisms for DNS (EDNS(0))" [RFC8467]. 663 6.5. Connection Handling 665 "DNS Transport over TCP - Implementation Requirements" [RFC7766] 666 provides updated guidance on DNS over TCP, some of which is 667 applicable to DoQ. This section provides similar advice on 668 connection handling for DoQ. 670 6.5.1. Connection Reuse 672 Historic implementations of DNS clients are known to open and close 673 TCP connections for each DNS query. To amortise connection setup 674 costs, both clients and servers SHOULD support connection reuse by 675 sending multiple queries and responses over a single persistent QUIC 676 connection. 678 In order to achieve performance on par with UDP, DNS clients SHOULD 679 send their queries concurrently over the QUIC streams on a QUIC 680 connection. That is, when a DNS client sends multiple queries to a 681 server over a QUIC connection, it SHOULD NOT wait for an outstanding 682 reply before sending the next query. 684 6.5.2. Resource Management 686 Proper management of established and idle connections is important to 687 the healthy operation of a DNS server. 689 An implementation of DoQ SHOULD follow best practices similar to 690 those specified for DNS over TCP [RFC7766], in particular with regard 691 to: 693 * Concurrent Connections (Section 6.2.2 of [RFC7766], updated by 694 Section 6.4 of [RFC9103]) 696 * Security Considerations (Section 10 of [RFC7766]) 698 Failure to do so may lead to resource exhaustion and denial of 699 service. 701 Clients that want to maintain long duration DoQ connections SHOULD 702 use the idle timeout mechanisms defined in Section 10.1 of [RFC9000], 703 the QUIC transport specification. Clients and servers MUST NOT send 704 the edns-tcp-keepalive EDNS(0) Option [RFC7828] in any messages sent 705 on a DoQ connection (because it is specific to the use of TCP/TLS as 706 a transport). 708 This document does not make specific recommendations for timeout 709 values on idle connections. Clients and servers should reuse and/or 710 close connections depending on the level of available resources. 711 Timeouts may be longer during periods of low activity and shorter 712 during periods of high activity. 714 6.5.3. Using 0-RTT and Session Resumption 716 Using 0-RTT for DNS over QUIC has many compelling advantages. 717 Clients can establish connections and send queries without incurring 718 a connection delay. Servers can thus negotiate low values of the 719 connection timers, which reduces the total number of connections that 720 they need to manage. They can do that because the clients that use 721 0-RTT will not incur latency penalties if new connections are 722 required for a query. 724 Session resumption and 0-RTT data transmission create privacy risks 725 detailed in detailed in Section 9.2 and Section 9.1. The following 726 recommendations are meant to reduce the privacy risks while enjoying 727 the performance benefits of 0-RTT data, with the restriction 728 specified in Section 5.5. 730 Clients SHOULD use resumption tickets only once, as specified in 731 Appendix C.4 to [RFC8446]. By default, clients SHOULD NOT use 732 session resumption if the client's connectivity has changed. 734 Clients could receive address validation tokens from the server using 735 the NEW_TOKEN mechanism; see Section 8 of [RFC9000]. The associated 736 tracking risks are mentioned in Section 9.3. Clients SHOULD only use 737 the address validation tokens when they are also using session 738 resumption, thus avoiding additional tracking risks. 740 Servers SHOULD issue session resumption tickets with a sufficiently 741 long life time (e.g., 6 hours), so that clients are not tempted to 742 either keep connection alive or frequently poll the server to renew 743 session resumption tickets. Servers SHOULD implement the anti-replay 744 mechanisms specified in Section 8 of [RFC8446]. 746 6.5.4. Controlling Connection Migration For Privacy 748 DoQ implementation might consider using the connection migration 749 features defined in Section 9 of [RFC9000]. These features enable 750 connections to continue operating as the client's connectivity 751 changes. As detailed in Section 9.4, these features trade off 752 privacy for latency. By default, clients SHOULD be configured to 753 prioritise privacy and start new sessions if their connectivity 754 changes. 756 6.6. Processing Queries in Parallel 758 As specified in Section 7 of [RFC7766] "DNS Transport over TCP - 759 Implementation Requirements", resolvers are RECOMMENDED to support 760 the preparing of responses in parallel and sending them out of order. 761 In DoQ, they do that by sending responses on their specific stream as 762 soon as possible, without waiting for availability of responses for 763 previously opened streams. 765 6.7. Zone transfer 767 [RFC9103] specifies zone transfer over TLS (XoT) and includes updates 768 to [RFC1995] (IXFR), [RFC5936] (AXFR) and [RFC7766]. Considerations 769 relating to the re-use of XoT connections described there apply 770 analogously to zone transfers performed using DoQ connections. For 771 example: 773 * DoQ servers MUST be able to handle multiple concurrent IXFR 774 requests on a single QUIC connection 776 * DoQ servers MUST be able to handle multiple concurrent AXFR 777 requests on a single QUIC connection 779 * DoQ implementations SHOULD 781 - use the same QUIC connection for both AXFR and IXFR requests to 782 the same primary 784 - pipeline such requests (if they pipeline XFR requests in 785 general) and MAY intermingle them 787 - send the response(s) for each request as soon as they are 788 available i.e. responses MAY be sent intermingled 790 6.8. Flow Control Mechanisms 792 Servers and Clients manage flow control using the mechanisms defined 793 in Section 4 of [RFC9000]. These mechanisms allow clients and 794 servers to specify how many streams can be created, how much data can 795 be sent on a stream, and how much data can be sent on the union of 796 all streams. For DNS over QUIC, controlling how many streams are 797 created allows servers to control how many new requests the client 798 can send on a given connection. 800 Flow control exists to protect endpoint resources. For servers, 801 global and per-stream flow control limits control how much data can 802 be sent by clients. The same mechanisms allow clients to control how 803 much data can be sent by servers. Values that are too small will 804 unnecessarily limit performance. Values that are too large might 805 expose endpoints to overload or memory exhaustion. Implementations 806 or deployments will need to adjust flow control limits to balance 807 these concerns. In particular, zone transfer implementations will 808 need to control these limits carefully to ensure both large and 809 concurrent zone transfers are well managed. 811 Initial values of parameters control how many requests and how much 812 data can be sent by clients and servers at the beginning of the 813 connection. These values are specified in transport parameters 814 exchanged during the connection handshake. The parameter values 815 received in the initial connection also control how many requests and 816 how much data can be sent by clients using 0-RTT data in a resumed 817 connection. Using too small values of these initial parameters would 818 restrict the usefulness of allowing 0-RTT data. 820 7. Implementation Status 822 (RFC EDITOR NOTE: THIS SECTION TO BE REMOVED BEFORE PUBLICATION) This 823 section records the status of known implementations of the protocol 824 defined by this specification at the time of posting of this 825 Internet-Draft, and is based on a proposal described in [RFC7942]. 827 1. AdGuard launched a DoQ recursive resolver service in December 828 2020. They have released a suite of open source tools that 829 support DoQ: 831 1. AdGuard C++ DNS libraries (https://github.com/AdguardTeam/ 832 DnsLibs) A DNS proxy library that supports all existing DNS 833 protocols including DNS-over-TLS, DNS-over-HTTPS, DNSCrypt 834 and DNS-over-QUIC (experimental). 836 2. DNS Proxy (https://github.com/AdguardTeam/dnsproxy) A simple 837 DNS proxy server that supports all existing DNS protocols 838 including DNS-over-TLS, DNS-over-HTTPS, DNSCrypt, and DNS- 839 over-QUIC. Moreover, it can work as a DNS-over-HTTPS, DNS- 840 over-TLS or DNS-over-QUIC server. 842 3. CoreDNS fork for AdGuard DNS (https://github.com/AdguardTeam/ 843 coredns) Includes DNS-over-QUIC server-side support. 845 4. dnslookup (https://github.com/ameshkov/dnslookup) Simple 846 command line utility to make DNS lookups. Supports all known 847 DNS protocols: plain DNS, DoH, DoT, DoQ, DNSCrypt. 849 2. Quicdoq (https://github.com/private-octopus/quicdoq) Quicdoq is a 850 simple open source implementation of DoQ. It is written in C, 851 based on Picoquic (https://github.com/private-octopus/picoquic). 853 3. Flamethrower (https://github.com/DNS-OARC/flamethrower/tree/dns- 854 over-quic) is an open source DNS performance and functional 855 testing utility written in C++ that has an experimental 856 implementation of DoQ. 858 4. aioquic (https://github.com/aiortc/aioquic) is an implementation 859 of QUIC in Python. It includes example client and server for 860 DoQ. 862 7.1. Performance Measurements 864 To our knowledge, no benchmarking studies comparing DoT, DoH and DoQ 865 are published yet. However anecdotal evidence from the AdGuard DoQ 866 recursive resolver deployment (https://adguard.com/en/blog/dns-over- 867 quic.html) indicates that it performs well compared to the other 868 encrypted protocols, particularly in mobile environments. Reasons 869 given for this include that DoQ 871 * Uses less bandwidth due to a more efficient handshake (and due to 872 less per message overhead when compared to DoH). 874 * Performs better in mobile environments due to the increased 875 resilience to packet loss 877 * Can maintain connections as users move between mobile networks via 878 its connection management 880 8. Security Considerations 882 The security considerations of DoQ should be comparable to those of 883 DoT [RFC7858]. 885 9. Privacy Considerations 887 The general considerations of encrypted transports provided in "DNS 888 Privacy Considerations" [RFC9076] apply to DoQ. The specific 889 considerations provided there do not differ between DoT and DoQ, and 890 are not discussed further here. Similarly, "Recommendations for DNS 891 Privacy Service Operators" [RFC8932] (which covers operational, 892 policy, and security considerations for DNS privacy services) is also 893 applicable to DoQ services. 895 QUIC incorporates the mechanisms of TLS 1.3 [RFC8446] and this 896 enables QUIC transmission of "0-RTT" data. This can provide 897 interesting latency gains, but it raises two concerns: 899 1. Adversaries could replay the 0-RTT data and infer its content 900 from the behavior of the receiving server. 902 2. The 0-RTT mechanism relies on TLS session resumption, which can 903 provide linkability between successive client sessions. 905 These issues are developed in Section 9.1 and Section 9.2. 907 9.1. Privacy Issues With 0-RTT data 909 The 0-RTT data can be replayed by adversaries. That data may trigger 910 queries by a recursive resolver to authoritative resolvers. 911 Adversaries may be able to pick a time at which the recursive 912 resolver outgoing traffic is observable, and thus find out what name 913 was queried for in the 0-RTT data. 915 This risk is in fact a subset of the general problem of observing the 916 behavior of the recursive resolver discussed in "DNS Privacy 917 Considerations" [RFC9076]. The attack is partially mitigated by 918 reducing the observability of this traffic. The mandatory replay 919 protection mechanisms in TLS 1.3 [RFC8446] limit but do not eliminate 920 the risk of replay. 0-RTT packets can only be replayed within a 921 narrow window, which is only wide enough to account for variations in 922 clock skew and network transmission. 924 The recommendation for TLS 1.3 [RFC8446] is that the capability to 925 use 0-RTT data should be turned off by default, and only enabled if 926 the user clearly understands the associated risks. In our case, 927 allowing 0-RTT data provides significant performance gains, and we 928 are concerned that a recommendation to not use it would simply be 929 ignored. Instead, we provide a set of practical recommendations in 930 Section 5.5 and Section 6.5.3. 932 The prevention on allowing replayable transactions in 0-RTT data 933 expressed in Section 5.5 blocks the most obvious risks of replay 934 attacks, as it only allows for transactions that will not change the 935 long term state of the server. 937 The attacks described above apply to the stub resolver to recursive 938 resolver scenario, but similar attacks might be envisaged in the 939 recursive resolver to authoritative resolver scenario, and the same 940 mitigations apply. 942 9.2. Privacy Issues With Session Resumption 944 The QUIC session resumption mechanism reduces the cost of re- 945 establishing sessions and enables 0-RTT data. There is a linkability 946 issue associated with session resumption, if the same resumption 947 token is used several times. Attackers on path between client and 948 server could observe repeated usage of the token and use that to 949 track the client over time or over multiple locations. 951 The session resumption mechanism allows servers to correlate the 952 resumed sessions with the initial sessions, and thus to track the 953 client. This creates a virtual long duration session. The series of 954 queries in that session can be used by the server to identify the 955 client. Servers can most probably do that already if the client 956 address remains constant, but session resumption tickets also enable 957 tracking after changes of the client's address. 959 The recommendations in Section 6.5.3 are designed to mitigate these 960 risks. Using session tickets only once mitigates the risk of 961 tracking by third parties. Refusing to resume a session if addresses 962 change mitigates the risk of tracking by the server. 964 The privacy trade-offs here may be context specific. Stub resolvers 965 will have a strong motivation to prefer privacy over latency since 966 they often change location. However, recursive resolvers that use a 967 small set of static IP addresses are more likely to prefer the 968 reduced latency provided by session resumption and may consider this 969 a valid reason to use resumption tickets even if the IP address 970 changed between sessions. 972 Encrypted zone transfer (RFC9103) explicitly does not attempt to hide 973 the identity of the parties involved in the transfer, but at the same 974 time such transfers are not particularly latency sensitive. This 975 means that applications supporting zone transfers may decide to apply 976 the same protections as stub to recursive applications. 978 9.3. Privacy Issues With Address Validation Tokens 980 QUIC specifies address validation mechanisms in Section 8 of 981 [RFC9000]. Use of an address validation token allows QUIC servers to 982 avoid an extra RTT for new connections. Address validation tokens 983 are typically tied to an IP address. QUIC clients normally only use 984 these tokens when setting a new connection from a previously used 985 address. However, due to the prevalence of NAT, clients are not 986 always aware that they are using a new address. There is a 987 linkability risk if clients mistakenly use address validation tokens 988 after unknowingly moving to a new location. 990 The recommendations in Section 6.5.3 mitigates this risk by tying the 991 usage of the NEW_TOKEN to that of session resumption. 993 9.4. Privacy Issues With Long Duration Sessions 995 A potential alternative to session resumption is the use of long 996 duration sessions: if a session remains open for a long time, new 997 queries can be sent without incurring connection establishment 998 delays. It is worth pointing out that the two solutions have similar 999 privacy characteristics. Session resumption may allow servers to 1000 keep track of the IP addresses of clients, but long duration sessions 1001 have the same effect. 1003 In particular, a DoQ implementation might take advantage of the 1004 connection migration features of QUIC to maintain a session even if 1005 the client's connectivity changes, for example if the client migrates 1006 from a Wi-Fi connection to a cellular network connection, and then to 1007 another Wi-Fi connection. The server would be able to track the 1008 client location by monitoring the succession of IP addresses used by 1009 the long duration connection. 1011 The recommendation in Section 6.5.4 mitigates the privacy concerns 1012 related to long duration sessions using multiple client addresses. 1014 9.5. Traffic Analysis 1016 Even though QUIC packets are encrypted, adversaries can gain 1017 information from observing packet lengths, in both queries and 1018 responses, as well as packet timing. Many DNS requests are emitted 1019 by web browsers. Loading a specific web page may require resolving 1020 dozen of DNS names. If an application adopts a simple mapping of one 1021 query or response per packet, or "one QUIC STREAM frame per packet", 1022 then the succession of packet lengths may provide enough information 1023 to identify the requested site. 1025 Implementations SHOULD use the mechanisms defined in Section 6.4 to 1026 mitigate this attack. 1028 10. IANA Considerations 1030 10.1. Registration of DoQ Identification String 1032 This document creates a new registration for the identification of 1033 DoQ in the "Application Layer Protocol Negotiation (ALPN) Protocol 1034 IDs" registry [RFC7301]. 1036 The "doq" string identifies DoQ: 1038 Protocol: DoQ 1040 Identification Sequence: 0x64 0x6F 0x71 ("doq") 1041 Specification: This document 1043 10.2. Reservation of Dedicated Port 1045 Port 853 is currently reserved for 'DNS query-response protocol run 1046 over TLS/DTLS' [RFC7858]. However, the specification for DNS over 1047 DTLS (DoD) [RFC8094] is experimental, limited to stub to resolver, 1048 and no implementations or deployments currently exist to our 1049 knowledge (even though several years have passed since the 1050 specification was published). 1052 This specification proposes to additionally reserve the use of port 1053 853 for DoQ. QUIC was designed to be able to co-exist with other 1054 protocols on the same port, including DTLS , see Section 17.2 of 1055 [RFC9000]. 1057 IANA is requested to add the following value to the "Service Name and 1058 Transport Protocol Port Number Registry" in the System Range. The 1059 registry for that range requires IETF Review or IESG Approval 1060 [RFC6335]. 1062 Service Name: dns-over-quic 1064 Port Number: 853 1066 Transport Protocol(s): UDP 1068 Assignee: IESG 1070 Contact: IETF Chair 1072 Description: DNS query-response protocol run over QUIC 1074 Reference: This document 1076 10.2.1. Port number 784 for experimentations 1078 (RFC EDITOR NOTE: THIS SECTION TO BE REMOVED BEFORE PUBLICATION) 1079 Early experiments MAY use port 784. This port is marked in the IANA 1080 registry as unassigned. 1082 (Note that version in -02 of this draft experiments were directed to 1083 use port 8853.) 1085 10.3. Reservation of Extended DNS Error Code Too Early 1087 IANA is requested to add the following value to the Extended DNS 1088 Error Codes registry [RFC8914]: 1090 INFO-CODE: TBD 1092 Purpose: Too Early 1094 Reference: This document 1096 10.4. DNS over QUIC Error Codes Registry 1098 IANA [SHALL add/has added] a registry for "DNS over QUIC Error Codes" 1099 on the "Domain Name System (DNS) Parameters" web page. 1101 The "DNS over QUIC Error Codes" registry governs a 62-bit space. 1102 This space is split into three regions that are governed by different 1103 policies: 1105 * Permanent registrations for values between 0x00 and 0x3f (in 1106 hexadecimal; inclusive), which are assigned using Standards Action 1107 or IESG Approval as defined in Section 4.9 and Section 4.10 of 1108 [RFC8126] 1110 * Permanent registrations for values larger than 0x3f, which are 1111 assigned using the Specification Required policy ([RFC8126]) 1113 * Provisonal registrations for values larger than 0x3f, which 1114 require Expert Review, as defined in Section 4.5 of [RFC8126]. 1116 Provisional reservations share the range of values larger than 0x3f 1117 with some permanent registrations. This is by design, to enable 1118 conversion of provisional registrations into permanent registrations 1119 without requiring changes in deployed systems. (This design is 1120 aligned with the principles set in Section 22 of [RFC9000].) 1122 Registrations in this registry MUST include the following fields: 1124 Value: The assigned codepoint. 1126 Status: "Permanent" or "Provisional". 1128 Contact: Contact details for the registrant. 1130 Notes: Supplementary notes about the registration. 1132 In addition, permanent registrations MUST include: 1134 Error: A short mnemonic for the parameter. 1136 Specification: A reference to a publicly available specification for 1137 the value (optional for provisional registrations). 1139 Description: A brief description of the error code semantics, which 1140 MAY be a summary if a specification reference is provided. 1142 Provisional registrations of codepoints are intended to allow for 1143 private use and experimentation with extensions to DNS over QUIC. 1144 However, provisional registrations could be reclaimed and reassigned 1145 for another purpose. In addition to the parameters listed above, 1146 provisional registrations MUST include: 1148 Date: The date of last update to the registration. 1150 A request to update the date on any provisional registration can be 1151 made without review from the designated expert(s). 1153 The initial contents of this registry are shown in Table 1. 1155 +==========+=======================+================+===============+ 1156 |Value | Error |Description | Specification | 1157 +==========+=======================+================+===============+ 1158 |0x0 | DOQ_NO_ERROR |No error | Section 5.3 | 1159 +----------+-----------------------+----------------+---------------+ 1160 |0x1 | DOQ_INTERNAL_ERROR |Implementation | Section 5.3 | 1161 | | |error | | 1162 +----------+-----------------------+----------------+---------------+ 1163 |0x2 | DOQ_PROTOCOL_ERROR |Generic protocol| Section 5.3 | 1164 | | |violation | | 1165 +----------+-----------------------+----------------+---------------+ 1166 |0x3 | DOQ_REQUEST_CANCELLED |Request | Section 5.3 | 1167 | | |cancelled by | | 1168 | | |client | | 1169 +----------+-----------------------+----------------+---------------+ 1170 |0x4 | DOQ_EXCESSIVE_LOAD |Closing a | Section 5.3 | 1171 | | |connection for | | 1172 | | |excessive load | | 1173 +----------+-----------------------+----------------+---------------+ 1174 |0xd098ea5e| DOQ_ERROR_RESERVED |Alternative | Section 5.3 | 1175 | | |error code used | | 1176 | | |for tests | | 1177 +----------+-----------------------+----------------+---------------+ 1179 Table 1: Initial DNS over QUIC Error Codes Entries 1181 11. Acknowledgements 1183 This document liberally borrows text from the HTTP-3 specification 1184 [I-D.ietf-quic-http] edited by Mike Bishop, and from the DoT 1185 specification [RFC7858] authored by Zi Hu, Liang Zhu, John Heidemann, 1186 Allison Mankin, Duane Wessels, and Paul Hoffman. 1188 The privacy issue with 0-RTT data and session resumption were 1189 analyzed by Daniel Kahn Gillmor (DKG) in a message to the IETF 1190 "DPRIVE" working group [DNS0RTT]. 1192 Thanks to Tony Finch for an extensive review of the initial version 1193 of this draft, and to Robert Evans for the discussion of 0-RTT 1194 privacy issues. Reviews by Paul Hoffman and Martin Thomson and 1195 interoperability tests conducted by Stephane Bortzmeyer helped 1196 improve the definition of the protocol. 1198 12. References 1200 12.1. Normative References 1202 [I-D.ietf-dnsop-rfc8499bis] 1203 Hoffman, P. and K. Fujiwara, "DNS Terminology", Work in 1204 Progress, Internet-Draft, draft-ietf-dnsop-rfc8499bis-03, 1205 28 September 2021, . 1208 [RFC1034] Mockapetris, P., "Domain names - concepts and facilities", 1209 STD 13, RFC 1034, DOI 10.17487/RFC1034, November 1987, 1210 . 1212 [RFC1035] Mockapetris, P., "Domain names - implementation and 1213 specification", STD 13, RFC 1035, DOI 10.17487/RFC1035, 1214 November 1987, . 1216 [RFC1995] Ohta, M., "Incremental Zone Transfer in DNS", RFC 1995, 1217 DOI 10.17487/RFC1995, August 1996, 1218 . 1220 [RFC5936] Lewis, E. and A. Hoenes, Ed., "DNS Zone Transfer Protocol 1221 (AXFR)", RFC 5936, DOI 10.17487/RFC5936, June 2010, 1222 . 1224 [RFC6891] Damas, J., Graff, M., and P. Vixie, "Extension Mechanisms 1225 for DNS (EDNS(0))", STD 75, RFC 6891, 1226 DOI 10.17487/RFC6891, April 2013, 1227 . 1229 [RFC7301] Friedl, S., Popov, A., Langley, A., and E. Stephan, 1230 "Transport Layer Security (TLS) Application-Layer Protocol 1231 Negotiation Extension", RFC 7301, DOI 10.17487/RFC7301, 1232 July 2014, . 1234 [RFC7766] Dickinson, J., Dickinson, S., Bellis, R., Mankin, A., and 1235 D. Wessels, "DNS Transport over TCP - Implementation 1236 Requirements", RFC 7766, DOI 10.17487/RFC7766, March 2016, 1237 . 1239 [RFC7828] Wouters, P., Abley, J., Dickinson, S., and R. Bellis, "The 1240 edns-tcp-keepalive EDNS0 Option", RFC 7828, 1241 DOI 10.17487/RFC7828, April 2016, 1242 . 1244 [RFC7830] Mayrhofer, A., "The EDNS(0) Padding Option", RFC 7830, 1245 DOI 10.17487/RFC7830, May 2016, 1246 . 1248 [RFC7858] Hu, Z., Zhu, L., Heidemann, J., Mankin, A., Wessels, D., 1249 and P. Hoffman, "Specification for DNS over Transport 1250 Layer Security (TLS)", RFC 7858, DOI 10.17487/RFC7858, May 1251 2016, . 1253 [RFC7873] Eastlake 3rd, D. and M. Andrews, "Domain Name System (DNS) 1254 Cookies", RFC 7873, DOI 10.17487/RFC7873, May 2016, 1255 . 1257 [RFC8094] Reddy, T., Wing, D., and P. Patil, "DNS over Datagram 1258 Transport Layer Security (DTLS)", RFC 8094, 1259 DOI 10.17487/RFC8094, February 2017, 1260 . 1262 [RFC8126] Cotton, M., Leiba, B., and T. Narten, "Guidelines for 1263 Writing an IANA Considerations Section in RFCs", BCP 26, 1264 RFC 8126, DOI 10.17487/RFC8126, June 2017, 1265 . 1267 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 1268 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 1269 May 2017, . 1271 [RFC8310] Dickinson, S., Gillmor, D., and T. Reddy, "Usage Profiles 1272 for DNS over TLS and DNS over DTLS", RFC 8310, 1273 DOI 10.17487/RFC8310, March 2018, 1274 . 1276 [RFC8467] Mayrhofer, A., "Padding Policies for Extension Mechanisms 1277 for DNS (EDNS(0))", RFC 8467, DOI 10.17487/RFC8467, 1278 October 2018, . 1280 [RFC8484] Hoffman, P. and P. McManus, "DNS Queries over HTTPS 1281 (DoH)", RFC 8484, DOI 10.17487/RFC8484, October 2018, 1282 . 1284 [RFC8914] Kumari, W., Hunt, E., Arends, R., Hardaker, W., and D. 1285 Lawrence, "Extended DNS Errors", RFC 8914, 1286 DOI 10.17487/RFC8914, October 2020, 1287 . 1289 [RFC9000] Iyengar, J., Ed. and M. Thomson, Ed., "QUIC: A UDP-Based 1290 Multiplexed and Secure Transport", RFC 9000, 1291 DOI 10.17487/RFC9000, May 2021, 1292 . 1294 [RFC9001] Thomson, M., Ed. and S. Turner, Ed., "Using TLS to Secure 1295 QUIC", RFC 9001, DOI 10.17487/RFC9001, May 2021, 1296 . 1298 [RFC9103] Toorop, W., Dickinson, S., Sahib, S., Aras, P., and A. 1299 Mankin, "DNS Zone Transfer over TLS", RFC 9103, 1300 DOI 10.17487/RFC9103, August 2021, 1301 . 1303 12.2. Informative References 1305 [DNS0RTT] Kahn Gillmor, D., "DNS + 0-RTT", Message to DNS-Privacy WG 1306 mailing list, 6 April 2016, . 1309 [I-D.ietf-quic-http] 1310 Bishop, M., "Hypertext Transfer Protocol Version 3 1311 (HTTP/3)", Work in Progress, Internet-Draft, draft-ietf- 1312 quic-http-34, 2 February 2021, 1313 . 1316 [RFC1996] Vixie, P., "A Mechanism for Prompt Notification of Zone 1317 Changes (DNS NOTIFY)", RFC 1996, DOI 10.17487/RFC1996, 1318 August 1996, . 1320 [RFC6335] Cotton, M., Eggert, L., Touch, J., Westerlund, M., and S. 1321 Cheshire, "Internet Assigned Numbers Authority (IANA) 1322 Procedures for the Management of the Service Name and 1323 Transport Protocol Port Number Registry", BCP 165, 1324 RFC 6335, DOI 10.17487/RFC6335, August 2011, 1325 . 1327 [RFC7942] Sheffer, Y. and A. Farrel, "Improving Awareness of Running 1328 Code: The Implementation Status Section", BCP 205, 1329 RFC 7942, DOI 10.17487/RFC7942, July 2016, 1330 . 1332 [RFC8446] Rescorla, E., "The Transport Layer Security (TLS) Protocol 1333 Version 1.3", RFC 8446, DOI 10.17487/RFC8446, August 2018, 1334 . 1336 [RFC8490] Bellis, R., Cheshire, S., Dickinson, J., Dickinson, S., 1337 Lemon, T., and T. Pusateri, "DNS Stateful Operations", 1338 RFC 8490, DOI 10.17487/RFC8490, March 2019, 1339 . 1341 [RFC8932] Dickinson, S., Overeinder, B., van Rijswijk-Deij, R., and 1342 A. Mankin, "Recommendations for DNS Privacy Service 1343 Operators", BCP 232, RFC 8932, DOI 10.17487/RFC8932, 1344 October 2020, . 1346 [RFC9002] Iyengar, J., Ed. and I. Swett, Ed., "QUIC Loss Detection 1347 and Congestion Control", RFC 9002, DOI 10.17487/RFC9002, 1348 May 2021, . 1350 [RFC9076] Wicinski, T., Ed., "DNS Privacy Considerations", RFC 9076, 1351 DOI 10.17487/RFC9076, July 2021, 1352 . 1354 Appendix A. The NOTIFY Service 1356 This appendix discusses why it is considered acceptable to send 1357 NOTIFY (see [RFC1996]) in 0-RTT data. 1359 Section 5.5 says "The 0-RTT mechanism SHOULD NOT be used to send DNS 1360 requests that are not "replayable" transactions". This specification 1361 supports sending a NOTIFY in 0-RTT data because although a NOTIFY 1362 technically changes the state of the receiving server, the effect of 1363 replaying NOTIFYs has negligible impact in practice. 1365 NOTIFY messages prompt a secondary to either send an SOA query or an 1366 XFR request to the primary on the basis that a newer version of the 1367 zone is available. It has long been recognized that NOTIFYs can be 1368 forged and, in theory, used to cause a secondary to send repeated 1369 unnecessary requests to the primary. For this reason, most 1370 implementations have some form of throttling of the SOA/XFR queries 1371 triggered by the receipt of one or more NOTIFYs. 1373 [RFC9103] describes the privacy risks associated with both NOTIFY and 1374 SOA queries and does not include addressing those risks within the 1375 scope of encrypting zone transfers. Given this, the privacy benefit 1376 of using DoQ for NOTIFY is not clear - but for the same reason, 1377 sending NOTIFY as 0-RTT data has no privacy risk above that of 1378 sending it using cleartext DNS. 1380 Appendix B. Notable Changes From Previous Versions 1382 (RFC EDITOR NOTE: THIS SECTION TO BE REMOVED BEFORE PUBLICATION) 1384 B.1. Stream Mapping Incompatibility With Draft-02 1386 Versions prior to -02 of this specification proposed a simpler 1387 mapping scheme of queries and responses to QUIc stream, which omitted 1388 the 2 byte length field and supported only a single response on a 1389 given stream. The more complex mapping in Section 5.2 was adopted to 1390 specifically cater for XFR support, however it breaks compatibility 1391 with earlier versions. 1393 Authors' Addresses 1395 Christian Huitema 1396 Private Octopus Inc. 1397 427 Golfcourse Rd 1398 Friday Harbor, WA 98250 1399 United States of America 1401 Email: huitema@huitema.net 1403 Sara Dickinson 1404 Sinodun IT 1405 Oxford Science Park 1406 Oxford 1407 OX4 4GA 1408 United Kingdom 1410 Email: sara@sinodun.com 1412 Allison Mankin 1413 Salesforce 1415 Email: allison.mankin@gmail.com