idnits 2.17.1 draft-ietf-dnsop-session-signal-07.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 'Updates: ' line in the draft header should list only the _numbers_ of the RFCs which will be updated by this document (if approved); it should not include the word 'RFC' in the list. -- The draft header indicates that this document updates RFC1035, but the abstract doesn't seem to mention this, which it should. -- The draft header indicates that this document updates RFC7766, but the abstract doesn't seem to mention this, which it should. 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 exact meaning of the all-uppercase expression 'NOT REQUIRED' is not defined in RFC 2119. If it is intended as a requirements expression, it should be rewritten using one of the combinations defined in RFC 2119; otherwise it should not be all-uppercase. (Using the creation date from RFC1035, updated by this document, for RFC5378 checks: 1987-11-01) -- The document seems to lack a disclaimer for pre-RFC5378 work, but may have content which was first submitted before 10 November 2008. If you have contacted all the original authors and they are all willing to grant the BCP78 rights to the IETF Trust, then this is fine, and you can ignore this comment. If not, you may need to add the pre-RFC5378 disclaimer. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (March 19, 2018) is 2228 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 (-25) exists of draft-ietf-dnssd-push-13 == Outdated reference: A later version (-06) exists of draft-ietf-dprive-padding-policy-04 == Outdated reference: A later version (-28) exists of draft-ietf-tls-tls13-26 == Outdated reference: A later version (-04) exists of draft-sctl-dnssd-mdns-relay-03 Summary: 0 errors (**), 0 flaws (~~), 7 warnings (==), 5 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 DNSOP Working Group R. Bellis 3 Internet-Draft ISC 4 Updates: RFC 1035, RFC 7766 (if S. Cheshire 5 approved) Apple Inc. 6 Intended status: Standards Track J. Dickinson 7 Expires: September 20, 2018 S. Dickinson 8 Sinodun 9 T. Lemon 10 Barefoot Consulting 11 T. Pusateri 12 Unaffiliated 13 March 19, 2018 15 DNS Stateful Operations 16 draft-ietf-dnsop-session-signal-07 18 Abstract 20 This document defines a new DNS OPCODE for DNS Stateful Operations 21 (DSO). DSO messages communicate operations within persistent 22 stateful sessions, using type-length-value (TLV) syntax. Three TLVs 23 are defined that manage session timeouts, termination, and encryption 24 padding, and a framework is defined for extensions to enable new 25 stateful operations. 27 Status of This Memo 29 This Internet-Draft is submitted in full conformance with the 30 provisions of BCP 78 and BCP 79. 32 Internet-Drafts are working documents of the Internet Engineering 33 Task Force (IETF). Note that other groups may also distribute 34 working documents as Internet-Drafts. The list of current Internet- 35 Drafts is at https://datatracker.ietf.org/drafts/current/. 37 Internet-Drafts are draft documents valid for a maximum of six months 38 and may be updated, replaced, or obsoleted by other documents at any 39 time. It is inappropriate to use Internet-Drafts as reference 40 material or to cite them other than as "work in progress." 42 This Internet-Draft will expire on September 20, 2018. 44 Copyright Notice 46 Copyright (c) 2018 IETF Trust and the persons identified as the 47 document authors. All rights reserved. 49 This document is subject to BCP 78 and the IETF Trust's Legal 50 Provisions Relating to IETF Documents 51 (https://trustee.ietf.org/license-info) in effect on the date of 52 publication of this document. Please review these documents 53 carefully, as they describe your rights and restrictions with respect 54 to this document. Code Components extracted from this document must 55 include Simplified BSD License text as described in Section 4.e of 56 the Trust Legal Provisions and are provided without warranty as 57 described in the Simplified BSD License. 59 Table of Contents 61 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 62 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 5 63 3. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 9 64 4. Protocol Details . . . . . . . . . . . . . . . . . . . . . . 10 65 4.1. DSO Session Establishment . . . . . . . . . . . . . . . . 10 66 4.1.1. Connection Sharing . . . . . . . . . . . . . . . . . 12 67 4.1.2. Zero Round-Trip Operation . . . . . . . . . . . . . . 12 68 4.1.3. Middlebox Considerations . . . . . . . . . . . . . . 13 69 4.2. Message Format . . . . . . . . . . . . . . . . . . . . . 14 70 4.2.1. DNS Header Fields in DSO Messages . . . . . . . . . . 15 71 4.2.2. DSO Data . . . . . . . . . . . . . . . . . . . . . . 17 72 4.2.3. EDNS(0) and TSIG . . . . . . . . . . . . . . . . . . 22 73 4.3. Message Handling . . . . . . . . . . . . . . . . . . . . 23 74 4.3.1. Error Responses . . . . . . . . . . . . . . . . . . . 24 75 4.4. DSO Response Generation . . . . . . . . . . . . . . . . . 25 76 4.5. Responder-Initiated Operation Cancellation . . . . . . . 26 77 5. DSO Session Lifecycle and Timers . . . . . . . . . . . . . . 27 78 5.1. DSO Session Initiation . . . . . . . . . . . . . . . . . 27 79 5.2. DSO Session Timeouts . . . . . . . . . . . . . . . . . . 27 80 5.3. Inactive DSO Sessions . . . . . . . . . . . . . . . . . . 28 81 5.4. The Inactivity Timeout . . . . . . . . . . . . . . . . . 29 82 5.4.1. Closing Inactive DSO Sessions . . . . . . . . . . . . 29 83 5.4.2. Values for the Inactivity Timeout . . . . . . . . . . 30 84 5.5. The Keepalive Interval . . . . . . . . . . . . . . . . . 31 85 5.5.1. Keepalive Interval Expiry . . . . . . . . . . . . . . 31 86 5.5.2. Values for the Keepalive Interval . . . . . . . . . . 31 87 5.6. Server-Initiated Session Termination . . . . . . . . . . 33 88 5.6.1. Server-Initiated Retry Delay Message . . . . . . . . 34 89 6. Base TLVs for DNS Stateful Operations . . . . . . . . . . . . 37 90 6.1. Keepalive TLV . . . . . . . . . . . . . . . . . . . . . . 37 91 6.1.1. Client handling of received Session Timeout values . 39 92 6.1.2. Relation to EDNS(0) TCP Keepalive Option . . . . . . 41 93 6.2. Retry Delay TLV . . . . . . . . . . . . . . . . . . . . . 42 94 6.2.1. Retry Delay TLV used as a Primary TLV . . . . . . . . 42 95 6.2.2. Retry Delay TLV used as a Response Additional TLV . . 44 96 6.3. Encryption Padding TLV . . . . . . . . . . . . . . . . . 44 98 7. Summary Highlights . . . . . . . . . . . . . . . . . . . . . 45 99 7.1. QR bit and MESSAGE ID . . . . . . . . . . . . . . . . . . 45 100 7.2. TLV Usage . . . . . . . . . . . . . . . . . . . . . . . . 46 101 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 48 102 8.1. DSO OPCODE Registration . . . . . . . . . . . . . . . . . 48 103 8.2. DSO RCODE Registration . . . . . . . . . . . . . . . . . 48 104 8.3. DSO Type Code Registry . . . . . . . . . . . . . . . . . 48 105 9. Security Considerations . . . . . . . . . . . . . . . . . . . 49 106 10. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 49 107 11. References . . . . . . . . . . . . . . . . . . . . . . . . . 50 108 11.1. Normative References . . . . . . . . . . . . . . . . . . 50 109 11.2. Informative References . . . . . . . . . . . . . . . . . 51 110 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 52 112 1. Introduction 114 The use of transports for DNS other than UDP is being increasingly 115 specified, for example, DNS over TCP [RFC1035] [RFC7766] and DNS over 116 TLS [RFC7858]. Such transports can offer persistent, long-lived 117 sessions and therefore when using them for transporting DNS messages 118 it is of benefit to have a mechanism that can establish parameters 119 associated with those sessions, such as timeouts. In such situations 120 it is also advantageous to support server-initiated messages. 122 The existing EDNS(0) Extension Mechanism for DNS [RFC6891] is 123 explicitly defined to only have "per-message" semantics. While 124 EDNS(0) has been used to signal at least one session-related 125 parameter (the EDNS(0) TCP Keepalive option [RFC7828]) the result is 126 less than optimal due to the restrictions imposed by the EDNS(0) 127 semantics and the lack of server-initiated signalling. For example, 128 a server cannot arbitrarily instruct a client to close a connection 129 because the server can only send EDNS(0) options in responses to 130 queries that contained EDNS(0) options. 132 This document defines a new DNS OPCODE, DSO (tentatively 6), for DNS 133 Stateful Operations. DSO messages are used to communicate operations 134 within persistent stateful sessions, expressed using type-length- 135 value (TLV) syntax. This document defines an initial set of three 136 TLVs, used to manage session timeouts, termination, and encryption 137 padding. 139 The three TLVs defined here are all mandatory for all implementations 140 of DSO. Further TLVs may be defined in additional specifications. 142 The format for DSO messages (Section 4.2) differs somewhat from the 143 traditional DNS message format used for standard queries and 144 responses. The standard twelve-byte header is used, but the four 145 count fields (QDCOUNT, ANCOUNT, NSCOUNT, ARCOUNT) are set to zero and 146 accordingly their corresponding sections are not present. The actual 147 data pertaining to DNS Stateful Operations (expressed in TLV syntax) 148 is appended to the end of the DNS message header. When displayed 149 using packet analyzer tools that have not been updated to recognize 150 the DSO format, this will result in the DSO data being displayed as 151 unknown additional data after the end of the DNS message. It is 152 likely that future updates to these tools will add the ability to 153 recognize, decode, and display the DSO data. 155 This new format has distinct advantages over an RR-based format 156 because it is more explicit and more compact. Each TLV definition is 157 specific to its use case, and as a result contains no redundant or 158 overloaded fields. Importantly, it completely avoids conflating DNS 159 Stateful Operations in any way with normal DNS operations or with 160 existing EDNS(0)-based functionality. A goal of this approach is to 161 avoid the operational issues that have befallen EDNS(0), particularly 162 relating to middlebox behaviour. 164 With EDNS(0), multiple options may be packed into a single OPT 165 pseudo-RR, and there is no generalized mechanism for a client to be 166 able to tell whether a server has processed or otherwise acted upon 167 each individual option within the combined OPT pseudo-RR. The 168 specifications for each individual option need to define how each 169 different option is to be acknowledged, if necessary. 171 In contrast to EDNS(0), with DSO there is no compelling motivation to 172 pack multiple operations into a single message for efficiency 173 reasons, because DSO always operates using a connection-oriented 174 transport protocol. Each DSO operation is communicated in its own 175 separate DNS message, and the transport protocol can take care of 176 packing several DNS messages into a single IP packet if appropriate. 177 For example, TCP can pack multiple small DNS messages into a single 178 TCP segment. This simplification allows for clearer semantics. Each 179 DSO request message communicates just one primary operation, and the 180 RCODE in the corresponding response message indicates the success or 181 failure of that operation. 183 2. Terminology 185 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 186 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 187 "OPTIONAL" in this document are to be interpreted as described in 188 "Key words for use in RFCs to Indicate Requirement Levels", when, and 189 only when, they appear in all capitals, as shown here [RFC2119] 190 [RFC8174]. 192 "DSO" is used to mean DNS Stateful Operation. 194 The term "connection" means a bidirectional byte (or message) stream, 195 where the bytes (or messages) are delivered reliably and in-order, 196 such as provided by using DNS over TCP [RFC1035] [RFC7766] or DNS 197 over TLS [RFC7858]. 199 The unqualified term "session" in the context of this document means 200 the exchange of DNS messages over a connection where: 202 o The connection between client and server is persistent and 203 relatively long-lived (i.e., minutes or hours, rather than 204 seconds). 206 o Either end of the connection may initiate messages to the other. 208 In this document the term "session" is used exclusively as described 209 above. The term has no relationship to the "session layer" of the 210 OSI "seven-layer model" popularized in the 1980s. 212 A "DSO Session" is established between two endpoints that acknowledge 213 persistent DNS state via the exchange of DSO messages over the 214 connection. This is distinct from a DNS-over-TCP session as 215 described in the previous specification for DNS over TCP [RFC7766]. 217 A "DSO Session" is terminated when the underlying connection is 218 closed. The underlying connection can be closed in two ways: 220 Where this specification says, "close gracefully," that means sending 221 a TLS close_notify (if TLS is in use) followed by a TCP FIN, or the 222 equivalents for other protocols. Where this specification requires a 223 connection to be closed gracefully, the requirement to initiate that 224 graceful close is placed on the client, to place the burden of TCP's 225 TIME-WAIT state on the client rather than the server. 227 Where this specification says, "forcibly abort," that means sending a 228 TCP RST, or the equivalent for other protocols. In the BSD Sockets 229 API this is achieved by setting the SO_LINGER option to zero before 230 closing the socket. 232 The term "server" means the software with a listening socket, 233 awaiting incoming connection requests. 235 The term "client" means the software which initiates a connection to 236 the server's listening socket. 238 The terms "initiator" and "responder" correspond respectively to the 239 initial sender and subsequent receiver of a DSO request message or 240 unacknowledged message, regardless of which was the "client" and 241 "server" in the usual DNS sense. 243 The term "sender" may apply to either an initiator (when sending a 244 DSO request message or unacknowledged message) or a responder (when 245 sending a DSO response message). 247 Likewise, the term "receiver" may apply to either a responder (when 248 receiving a DSO request message or unacknowledged message) or an 249 initiator (when receiving a DSO response message). 251 In protocol implementation there are generally two kinds of errors 252 that software writers have to deal with. The first is situations 253 that arise due to factors in the environment, such as temporary loss 254 of connectivity. While undesirable, these situations do not indicate 255 a flaw in the software, and they are situations that software should 256 generally be able to recover from. The second is situations that 257 should never happen when communicating with a correctly-implemented 258 peer. If they do happen, they indicate a serious flaw in the 259 protocol implementation, beyond what it is reasonable to expect 260 software to recover from. This document describes this latter form 261 of error condition as a "fatal error" and specifies that an 262 implementation encountering a fatal error condition "MUST forcibly 263 abort the connection immediately". Given that these fatal error 264 conditions signify defective software, and given that defective 265 software is likely to remain defective for some time until it is 266 fixed, after forcibly aborting a connection, a client SHOULD refrain 267 from automatically reconnecting to that same server instance for at 268 least one hour. 270 This document uses the term "same server instance" as follows: 272 o In cases where a server is specified or configured using an IP 273 address and TCP port number, two different configurations are 274 referring to the same server instance if they contain the same IP 275 address and TCP port number. 277 o In cases where a server is specified or configured using a 278 hostname and TCP port number, such as in the content of a DNS SRV 279 record [RFC2782], two different configurations (or DNS SRV 280 records) are considered to be referring to the same server 281 instance if they contain the same hostname (subject to the usual 282 case insensitive DNS name matching rules [RFC1034] [RFC1035]) and 283 TCP port number. In these cases, configurations with different 284 hostnames are considered to be referring to different server 285 instances, even if those different hostnames happen to be aliases, 286 or happen to resolve to the same IP address(es). Implementations 287 SHOULD NOT resolve hostnames and then perform matching of IP 288 address(es) in order to evaluate whether two entities should be 289 determined to be the "same server instance". 291 The term "long-lived operations" refers to operations such as Push 292 Notification subscriptions [I-D.ietf-dnssd-push], Discovery Relay 293 interface subscriptions [I-D.sctl-dnssd-mdns-relay], and other future 294 long-lived DNS operations that choose to use DSO as their basis, that 295 establish state that persists beyond the lifetime of a traditional 296 brief request/response transaction. This document, the base 297 specification for DNS Stateful Operations, defines a framework for 298 supporting long-lived operations, but does not itself define any 299 long-lived operations. Nonetheless, to appreciate the design 300 rationale behind DNS Stateful Operations, it is helpful to understand 301 the kind of long-lived operations that it is intended to support. 303 DNS Stateful Operations uses three kinds of message: "DSO request 304 messages", "DSO response messages", and "DSO unacknowledged 305 messages". A DSO request message elicits a DSO response message. 306 DSO unacknowledged messages are unidirectional messages and do not 307 generate any response. 309 Both DSO request messages and DSO unacknowledged messages are 310 formatted as DNS request messages (the header QR bit is set to zero, 311 as described in Section 4.2). One difference is that in DSO request 312 messages the MESSAGE ID field is nonzero; in DSO unacknowledged 313 messages it is zero. 315 The content of DSO messages is expressed using type-length-value 316 (TLV) syntax. 318 In a DSO request message or DSO unacknowledged message the first TLV 319 is referred to as the "Primary TLV" and determines the nature of the 320 operation being performed, including whether it is an acknowledged or 321 unacknowledged operation; any other TLVs in a DSO request message or 322 unacknowledged message are referred to as "Additional TLVs" and serve 323 additional non-primary purposes, which may be related to the primary 324 purpose, or not, as in the case of the encryption padding TLV. 326 A DSO response message may contain no TLVs, or it may contain one or 327 more TLVs as appropriate to the information being communicated. In 328 the context of DSO response messages, one or more TLVs with the same 329 DSO-TYPE as the Primary TLV in the corresponding DSO request message 330 are referred to as "Response Primary TLVs". Any other TLVs with 331 different DSO-TYPEs are referred to as "Response Additional TLVs". 332 The Response Primary TLV(s), if present, MUST occur first in the 333 response message, before any Response Additional TLVs. 335 Two timers (elapsed time since an event) are defined in this 336 document: 338 o an inactivity timer (see Section 5.4 and Section 6.1) 340 o a keepalive timer (see Section 5.5 and Section 6.1) 342 The timeouts associated with these timers are called the inactivity 343 timeout and the keepalive interval, respectively. The term "Session 344 Timeouts" is used to refer to this pair of timeout values. 346 Resetting a timer means resetting the timer value to zero and 347 starting the timer again. Clearing a timer means resetting the timer 348 value to zero but NOT starting the timer again. 350 3. Discussion 352 There are several use cases for DNS Stateful operations that can be 353 described here. 355 Firstly, establishing session parameters such as server-defined 356 timeouts is of great use in the general management of persistent 357 connections. For example, using DSO sessions for stub-to-recursive 358 DNS-over-TLS [RFC7858] is more flexible for both the client and the 359 server than attempting to manage sessions using just the EDNS(0) TCP 360 Keepalive option [RFC7828]. The simple set of TLVs defined in this 361 document is sufficient to greatly enhance connection management for 362 this use case. 364 Secondly, DNS-SD [RFC6763] has evolved into a naturally session-based 365 mechanism where, for example, long-lived subscriptions lend 366 themselves to 'push' mechanisms as opposed to polling. Long-lived 367 stateful connections and server-initiated messages align with this 368 use case [I-D.ietf-dnssd-push]. 370 A general use case is that DNS traffic is often bursty but session 371 establishment can be expensive. One challenge with long-lived 372 connections is to maintain sufficient traffic to maintain NAT and 373 firewall state. To mitigate this issue this document introduces a 374 new concept for the DNS, that is DSO "Keepalive traffic". This 375 traffic carries no DNS data and is not considered 'activity' in the 376 classic DNS sense, but serves to maintain state in middleboxes, and 377 to assure client and server that they still have connectivity to each 378 other. 380 4. Protocol Details 382 4.1. DSO Session Establishment 384 DSO messages MUST be carried in only protocols and in environments 385 where a session may be established according to the definition given 386 above in the Terminology section (Section 2). 388 DNS over plain UDP [RFC0768] is not appropriate since it fails on the 389 requirement for in-order message delivery, and, in the presence of 390 NAT gateways and firewalls with short UDP timeouts, it fails to 391 provide a persistent bi-directional communication channel unless an 392 excessive amount of keepalive traffic is used. 394 At the time of publication, DSO is specified only for DNS over TCP 395 [RFC1035] [RFC7766], and for DNS over TLS over TCP [RFC7858]. Any 396 use of DSO over some other connection technology needs to be 397 specified in an appropriate future document. 399 Determining whether a given connection is using DNS over TCP, or DNS 400 over TLS over TCP, is outside the scope of this specification, and 401 must be determined using some out-of-band configuration information. 402 There is no provision within the DSO specification to turn TLS on or 403 off during the lifetime of a connection. For service types where the 404 service instance is discovered using a DNS SRV record [RFC2782], the 405 specification for that service type SRV name [RFC6335] will state 406 whether the connection uses plain TCP, or TLS over TCP. For example, 407 the specification for the "_dns-push-tls._tcp" service 408 [I-D.ietf-dnssd-push], states that it uses TLS. It is a common 409 convention that protocols specified to run over TLS are given IANA 410 service type names ending in "-tls". 412 In some environments it may be known in advance by external means 413 that both client and server support DSO, and in these cases either 414 client or server may initiate DSO messages at any time. 416 However, in the typical case a server will not know in advance 417 whether a client supports DSO, so in general, unless it is known in 418 advance by other means that a client does support DSO, a server MUST 419 NOT initiate DSO request messages or DSO unacknowledged messages 420 until a DSO Session has been mutually established by at least one 421 successful DSO request/response exchange initiated by the client, as 422 described below. Similarly, unless it is known in advance by other 423 means that a server does support DSO, a client MUST NOT initiate DSO 424 unacknowledged messages until after a DSO Session has been mutually 425 established. 427 A DSO Session is established over a connection by the client sending 428 a DSO request message, such as a DSO Keepalive request message 429 (Section 6.1), and receiving a response, with matching MESSAGE ID, 430 and RCODE set to NOERROR (0), indicating that the DSO request was 431 successful. 433 If the RCODE in the response is set to DSOTYPENI ("DSO-TYPE Not 434 Implemented", tentatively RCODE 11) this indicates that the server 435 does support DSO, but does not implement the DSO-TYPE of the primary 436 TLV in this DSO request message. A server implementing DSO MUST NOT 437 return DSOTYPENI for a DSO Keepalive request message, because the 438 Keepalive TLV is mandatory to implement. But in the future, if a 439 client attempts to establish a DSO Session using a response-requiring 440 DSO request message using some newly-defined DSO-TYPE that the server 441 does not understand, that would result in a DSOTYPENI response. If 442 the server returns DSOTYPENI then a DSO Session is not considered 443 established, but the client is permitted to continue sending DNS 444 messages on the connection, including other DSO messages such as the 445 DSO Keepalive, which may result in a successful NOERROR response, 446 yielding the establishment of a DSO Session. 448 If the RCODE is set to any value other than NOERROR (0) or DSOTYPENI 449 (tentatively 11), then the client MUST assume that the server does 450 not implement DSO at all. In this case the client is permitted to 451 continue sending DNS messages on that connection, but the client 452 SHOULD NOT issue further DSO messages on that connection. 454 When the server receives a DSO request message from a client, and 455 transmits a successful NOERROR response to that request, the server 456 considers the DSO Session established. 458 When the client receives the server's NOERROR response to its DSO 459 request message, the client considers the DSO Session established. 461 Once a DSO Session has been established, either end may unilaterally 462 send appropriate DSO messages at any time, and therefore either 463 client or server may be the initiator of a message. 465 Once a DSO Session has been established, clients and servers should 466 behave as described in this specification with regard to inactivity 467 timeouts and session termination, not as previously prescribed in the 468 earlier specification for DNS over TCP [RFC7766]. 470 Note that for clients that implement only the DSO-TYPEs defined in 471 this base specification, sending a DSO Keepalive TLV is the only DSO 472 request message they have available to initiate a DSO Session. Even 473 for clients that do implement other future DSO-TYPEs, for simplicity 474 they MAY elect to always send an initial DSO Keepalive request 475 message as their way of initiating a DSO Session. A future 476 definition of a new response-requiring DSO-TYPE gives implementers 477 the option of using that new DSO-TYPE if they wish, but does not 478 change the fact that sending a DSO Keepalive TLV remains a valid way 479 of initiating a DSO Session. 481 4.1.1. Connection Sharing 483 As previously specified for DNS over TCP [RFC7766]: 485 To mitigate the risk of unintentional server overload, DNS 486 clients MUST take care to minimize the number of concurrent 487 TCP connections made to any individual server. It is RECOMMENDED 488 that for any given client/server interaction there SHOULD be 489 no more than one connection for regular queries, one for zone 490 transfers, and one for each protocol that is being used on top 491 of TCP (for example, if the resolver was using TLS). However, 492 it is noted that certain primary/secondary configurations 493 with many busy zones might need to use more than one TCP 494 connection for zone transfers for operational reasons (for 495 example, to support concurrent transfers of multiple zones). 497 A single server may support multiple services, including DNS Updates 498 [RFC2136], DNS Push Notifications [I-D.ietf-dnssd-push], and other 499 services, for one or more DNS zones. When a client discovers that 500 the target server for several different operations is the same target 501 hostname and port, the client SHOULD use a single shared DSO Session 502 for all those operations. A client SHOULD NOT open multiple 503 connections to the same target host and port just because the names 504 being operated on are different or happen to fall within different 505 zones. This requirement is to reduce unnecessary connection load on 506 the DNS server. 508 However, server implementers and operators should be aware that 509 connection sharing may not be possible in all cases. A single host 510 device may be home to multiple independent client software instances 511 that don't coordinate with each other. Similarly, multiple 512 independent client devices behind the same NAT gateway will also 513 typically appear to the DNS server as different source ports on the 514 same client IP address. Because of these constraints, a DNS server 515 MUST be prepared to accept multiple connections from different source 516 ports on the same client IP address. 518 4.1.2. Zero Round-Trip Operation 520 There is increased awareness today of the performance benefits of 521 eliminating round trips in session establishment. Technologies like 522 TCP Fast Open [RFC7413] and TLS 1.3 [I-D.ietf-tls-tls13] provide 523 mechanisms to reduce or eliminate round trips in session 524 establishment. 526 Similarly, DSO supports zero round-trip operation. 528 Having initiated a connection to a server, possibly using zero round- 529 trip TCP Fast Open and/or zero round-trip TLS 1.3, a client MAY send 530 multiple response-requiring DSO request messages to the server in 531 succession without having to wait for a response to the first request 532 message to confirm successful establishment of a DSO session. 534 However, a client MUST NOT send non-response-requiring DSO request 535 messages until after a DSO Session has been mutually established. 537 Similarly, a server MUST NOT send DSO request messages until it has 538 received a response-requiring DSO request message from a client and 539 transmitted a successful NOERROR response for that request. 541 Caution must be taken to ensure that DSO messages sent before the 542 first round-trip is completed are idempotent, or are otherwise immune 543 to any problems that could be result from the inadvertent replay that 544 can occur with zero round-trip operation. 546 4.1.3. Middlebox Considerations 548 Where an application-layer middlebox (e.g., a DNS proxy, forwarder, 549 or session multiplexer) is in the path, the middlebox MUST NOT 550 blindly forward DSO messages in either direction, and MUST treat the 551 inbound and outbound connections as separate sessions. This does not 552 preclude the use of DSO messages in the presence of an IP-layer 553 middlebox, such as a NAT that rewrites IP-layer and/or transport- 554 layer headers but otherwise preserves the effect of a single session 555 between the client and the server. 557 To illustrate the above, consider a network where a middlebox 558 terminates one or more TCP connections from clients and multiplexes 559 the queries therein over a single TCP connection to an upstream 560 server. The DSO messages and any associated state are specific to 561 the individual TCP connections. A DSO-aware middlebox MAY in some 562 circumstances be able to retain associated state and pass it between 563 the client and server (or vice versa) but this would be highly TLV- 564 specific. For example, the middlebox may be able to maintain a list 565 of which clients have made Push Notification subscriptions 566 [I-D.ietf-dnssd-push] and make its own subscription(s) on their 567 behalf, relaying any subsequent notifications to the client (or 568 clients) that have subscribed to that particular notification. 570 4.2. Message Format 572 A DSO message begins with the standard twelve-byte DNS message header 573 [RFC1035] with the OPCODE field set to the DSO OPCODE (tentatively 574 6). However, unlike standard DNS messages, the question section, 575 answer section, authority records section and additional records 576 sections are not present. The corresponding count fields (QDCOUNT, 577 ANCOUNT, NSCOUNT, ARCOUNT) MUST be set to zero on transmission. 579 If a DSO message is received where any of the count fields are not 580 zero, then a FORMERR MUST be returned, unless a future IETF Standard 581 specifies otherwise. 583 1 1 1 1 1 1 584 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 585 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 586 | MESSAGE ID | 587 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 588 |QR | OPCODE | Z | RCODE | 589 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 590 | QDCOUNT (MUST be zero) | 591 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 592 | ANCOUNT (MUST be zero) | 593 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 594 | NSCOUNT (MUST be zero) | 595 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 596 | ARCOUNT (MUST be zero) | 597 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 598 | | 599 / DSO Data / 600 / / 601 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 603 4.2.1. DNS Header Fields in DSO Messages 605 In an unacknowledged message the MESSAGE ID field MUST be set to 606 zero. In an acknowledged request message the MESSAGE ID field MUST 607 be set to a unique nonzero value, that the initiator is not currently 608 using for any other active operation on this connection. For the 609 purposes here, a MESSAGE ID is in use in this DSO Session if the 610 initiator has used it in a request for which it is still awaiting a 611 response, or if the client has used it to set up a long-lived 612 operation that has not yet been cancelled. For example, a long-lived 613 operation could be a Push Notification subscription 614 [I-D.ietf-dnssd-push] or a Discovery Relay interface subscription 615 [I-D.sctl-dnssd-mdns-relay]. 617 Whether a message is acknowledged or unacknowledged is determined 618 only by the specification for the Primary TLV. An acknowledgment 619 cannot be requested by including a nonzero message ID in a message 620 the primary TLV of which is specified to be unacknowledged, nor can 621 an acknowledgment be prevented by sending a message ID of zero in a 622 message with a primary TLV that is specified to be acknowledged. A 623 responder that receives either such malformed message MUST treat it 624 as a fatal error and forcibly abort the connection immediately. 626 In a request or unacknowledged message the DNS Header QR bit MUST be 627 zero (QR=0). If the QR bit is not zero the message is not a request 628 or unacknowledged message. 630 In a response message the DNS Header QR bit MUST be one (QR=1). 631 If the QR bit is not one the message is not a response message. 633 In a response message (QR=1) the MESSAGE ID field MUST contain a copy 634 of the value of the MESSAGE ID field in the request message being 635 responded to. In a response message (QR=1) the MESSAGE ID field MUST 636 NOT be zero. If a response message (QR=1) is received where the 637 MESSAGE ID is zero this is a fatal error and the recipient MUST 638 forcibly abort the connection immediately. 640 The DNS Header OPCODE field holds the DSO OPCODE value (tentatively 641 6). 643 The Z bits are currently unused in DSO messages, and in both DSO 644 requests and DSO responses the Z bits MUST be set to zero (0) on 645 transmission and MUST be silently ignored on reception, unless a 646 future IETF Standard specifies otherwise. 648 In a DNS request message (QR=0) the RCODE is set according to the 649 definition of the request. For example, in a Retry Delay message 650 (Section 5.6.1) the RCODE indicates the reason for termination. 651 However, in most cases, except where clearly specified otherwise, in 652 a DNS request message (QR=0) the RCODE is set to zero on 653 transmission, and silently ignored on reception. 655 The RCODE value in a response message (QR=1) may be one of the 656 following values: 658 +------+-----------+------------------------------------------------+ 659 | Code | Mnemonic | Description | 660 +------+-----------+------------------------------------------------+ 661 | 0 | NOERROR | Operation processed successfully | 662 | | | | 663 | 1 | FORMERR | Format error | 664 | | | | 665 | 2 | SERVFAIL | Server failed to process request due to a | 666 | | | problem with the server | 667 | | | | 668 | 3 | NXDOMAIN | Name Error -- Named entity does not exist | 669 | | | (TLV-dependent) | 670 | | | | 671 | 4 | NOTIMP | DSO not supported | 672 | | | | 673 | 5 | REFUSED | Operation declined for policy reasons | 674 | | | | 675 | 9 | NOTAUTH | Not Authoritative (TLV-dependent) | 676 | | | | 677 | 11 | DSOTYPENI | Primary TLV's DSO-Type is not implemented | 678 +------+-----------+------------------------------------------------+ 680 Use of the above RCODEs is likely to be common in DSO but does not 681 preclude the definition and use of other codes in future documents 682 that make use of DSO. 684 If a document defining a new DSO-TYPE makes use of NXDOMAIN (Name 685 Error) or NOTAUTH (Not Authoritative) then that document MUST specify 686 the specific interpretation of these RCODE values in the context of 687 that new DSO TLV. 689 4.2.2. DSO Data 691 The standard twelve-byte DNS message header with its zero-valued 692 count fields is followed by the DSO Data, expressed using TLV syntax, 693 as described below Section 4.2.2.1. 695 A DSO message may be a request message, a response message, or an 696 unacknowledged message. 698 A DSO request message or DSO unacknowledged message MUST contain at 699 least one TLV. The first TLV in a DSO request message or DSO 700 unacknowledged message is referred to as the "Primary TLV" and 701 determines the nature of the operation being performed, including 702 whether it is an acknowledged or unacknowledged operation. In some 703 cases it may be appropriate to include other TLVs in a request 704 message or unacknowledged message, such as the Encryption Padding TLV 705 (Section 6.3), and these extra TLVs are referred to as the 706 "Additional TLVs". 708 A DSO response message may contain no TLVs, or it may be specified to 709 contain one or more TLVs appropriate to the information being 710 communicated. 712 A DSO response message may contain one or more TLVs with DSO-TYPE the 713 same as the Primary TLV from the corresponding DSO request message, 714 in which case those TLV(s) are referred to as "Response Primary 715 TLVs". A DSO response message is not required to carry Response 716 Primary TLVs. The MESSAGE ID field in the DNS message header is 717 sufficient to identify the DSO request message to which this response 718 message relates. 720 A DSO response message may contain one or more TLVs with DSO-TYPEs 721 different from the Primary TLV from the corresponding DSO request 722 message, in which case those TLV(s) are referred to as "Response 723 Additional TLVs". 725 Response Primary TLV(s), if present, MUST occur first in the response 726 message, before any Response Additional TLVs. 728 It is anticipated that most DSO operations will be specified to use 729 request messages, which generate corresponding responses. In some 730 specialized high-traffic use cases, it may be appropriate to specify 731 unacknowledged messages. Unacknowledged messages can be more 732 efficient on the network, because they don't generate a stream of 733 corresponding reply messages. Using unacknowledged messages can also 734 simplify software in some cases, by removing need for an initiator to 735 maintain state while it waits to receive replies it doesn't care 736 about. When the specification for a particular TLV states that, when 737 used as a Primary TLV (i.e., first) in an outgoing DNS request 738 message (i.e., QR=0), that message is to be unacknowledged, the 739 MESSAGE ID field MUST be set to zero and the receiver MUST NOT 740 generate any response message corresponding to this unacknowledged 741 message. 743 The previous point, that the receiver MUST NOT generate responses to 744 unacknowledged messages, applies even in the case of errors. When a 745 DSO message is received where both the QR bit and the MESSAGE ID 746 field are zero, the receiver MUST NOT generate any response. For 747 example, if the DSO-TYPE in the Primary TLV is unrecognized, then a 748 DSOTYPENI error MUST NOT be returned; instead the receiver MUST 749 forcibly abort the connection immediately. 751 Unacknowledged messages MUST NOT be used "speculatively" in cases 752 where the sender doesn't know if the receiver supports the Primary 753 TLV in the message, because there is no way to receive any response 754 to indicate success or failure of the request message (the request 755 message does not contain a unique MESSAGE ID with which to associate 756 a response with its corresponding request). Unacknowledged messages 757 are only appropriate in cases where the sender already knows that the 758 receiver supports, and wishes to receive, these messages. 760 For example, after a client has subscribed for Push Notifications 761 [I-D.ietf-dnssd-push], the subsequent event notifications are then 762 sent as unacknowledged messages, and this is appropriate because the 763 client initiated the message stream by virtue of its Push 764 Notification subscription, thereby indicating its support of Push 765 Notifications, and its desire to receive those notifications. 767 Similarly, after an mDNS Relay client has subscribed to receive 768 inbound mDNS traffic from an mDNS Relay, the subsequent stream of 769 received packets is then sent using unacknowledged messages, and this 770 is appropriate because the client initiated the message stream by 771 virtue of its mDNS Relay link subscription, thereby indicating its 772 support of mDNS Relay, and its desire to receive inbound mDNS packets 773 over that DSO session [I-D.sctl-dnssd-mdns-relay]. 775 4.2.2.1. TLV Syntax 777 All TLVs, whether used as "Primary", "Additional", "Response 778 Primary", or "Response Additional", use the same encoding syntax. 780 The specification for a TLV states whether that DSO-TYPE may be used 781 in "Primary", "Additional", "Response Primary", or "Response 782 Additional" TLVs. The specification for a TLV also states whether, 783 when used as the Primary (i.e., first) TLV in a DNS request message 784 (i.e., QR=0), that DSO message is to be acknowledged. If the DSO 785 message is to be acknowledged, the specification also states which 786 TLVs, if any, are to be included in the response. The Primary TLV 787 may or may not be contained in the response, depending on what is 788 stated in the specification for that TLV. 790 1 1 1 1 1 1 791 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 792 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 793 | DSO-TYPE | 794 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 795 | DSO-LENGTH | 796 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 797 | | 798 / DSO-DATA / 799 / / 800 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 802 DSO-TYPE: A 16-bit unsigned integer, in network (big endian) byte 803 order, giving the DSO-TYPE of the current DSO TLV per the IANA DSO 804 Type Code Registry. 806 DSO-LENGTH: A 16-bit unsigned integer, in network (big endian) byte 807 order, giving the size in bytes of the DSO-DATA. 809 DSO-DATA: Type-code specific format. The generic DSO machinery 810 treats the DSO-DATA as an opaque "blob" without attempting to 811 interpret it. Interpretation of the meaning of the DSO-DATA for a 812 particular DSO-TYPE is the responsibility of the software that 813 implements that DSO-TYPE. 815 4.2.2.2. Request TLVs 817 The first TLV in a DSO request message or unacknowledged message is 818 the "Primary TLV" and indicates the operation to be performed. A DSO 819 request message or unacknowledged message MUST contain at at least 820 one TLV, the Primary TLV. 822 Immediately following the Primary TLV, a DSO request message or 823 unacknowledged message MAY contain one or more "Additional TLVs", 824 which specify additional parameters relating to the operation. 826 4.2.2.3. Response TLVs 828 Depending on the operation, a DSO response message MAY contain no 829 TLVs, because it is simply a response to a previous request message, 830 and the MESSAGE ID in the header is sufficient to identify the 831 request in question. Or it may contain a single response TLV, with 832 the same DSO-TYPE as the Primary TLV in the request message. 833 Alternatively it may contain one or more TLVs of other types, or a 834 combination of the above, as appropriate for the information that 835 needs to be communicated. The specification for each DSO TLV 836 determines what TLVs are required in a response to a request using 837 that TLV. 839 If a DSO response is received for an operation where the 840 specification requires that the response carry a particular TLV or 841 TLVs, and the required TLV(s) are not present, then this is a fatal 842 error and the recipient of the defective response message MUST 843 forcibly abort the connection immediately. 845 4.2.2.4. Unrecognized TLVs 847 If DSO request message is received containing an unrecognized Primary 848 TLV, with a nonzero MESSAGE ID (indicating that a response is 849 expected), then the receiver MUST send an error response with 850 matching MESSAGE ID, and RCODE DSOTYPENI (tentatively 11). The error 851 response MUST NOT contain a copy of the unrecognized Primary TLV. 853 If DSO unacknowledged message is received containing an unrecognized 854 Primary TLV, with a zero MESSAGE ID (indicating that no response is 855 expected), then this is a fatal error and the recipient MUST forcibly 856 abort the connection immediately. 858 If a DSO request message or unacknowledged message is received where 859 the Primary TLV is recognized, containing one or more unrecognized 860 Additional TLVs, the unrecognized Additional TLVs MUST be silently 861 ignored, and the remainder of the message is interpreted and handled 862 as if the unrecognized parts were not present. 864 Similarly, if a DSO response message is received containing one or 865 more unrecognized TLVs, the unrecognized TLVs MUST be silently 866 ignored, and the remainder of the message is interpreted and handled 867 as if the unrecognized parts were not present. 869 4.2.3. EDNS(0) and TSIG 871 Since the ARCOUNT field MUST be zero, a DSO message MUST NOT contain 872 an EDNS(0) option in the additional records section. If 873 functionality provided by current or future EDNS(0) options is 874 desired for DSO messages, one or more new DSO TLVs need to be defined 875 to carry the necessary information. 877 For example, the EDNS(0) Padding Option [RFC7830] used for security 878 purposes is not permitted in a DSO message, so if message padding is 879 desired for DSO messages then the Encryption Padding TLV described in 880 Section 6.3 MUST be used. 882 Similarly, a DSO message MUST NOT contain a TSIG record. A TSIG 883 record in a conventional DNS message is added as the last record in 884 the additional records section, and carries a signature computed over 885 the preceding message content. Since DSO data appears *after* the 886 additional records section, it would not be included in the signature 887 calculation. If use of signatures with DSO messages becomes 888 necessary in the future, a new DSO TLV needs to be defined to perform 889 this function. 891 Note however that, while DSO *messages* cannot include EDNS(0) or 892 TSIG records, a DSO *session* is typically used to carry a whole 893 series of DNS messages of different kinds, including DSO messages, 894 and other DNS message types like Query [RFC1034] [RFC1035] and Update 895 [RFC2136], and those messages can carry EDNS(0) and TSIG records. 897 Although messages may contain other EDNS(0) options as appropriate, 898 this specification explicitly prohibits use of the EDNS(0) TCP 899 Keepalive Option [RFC7828] in *any* messages sent on a DSO Session 900 (because it is obsoleted by the functionality provided by the DSO 901 Keepalive operation). If any message sent on a DSO Session contains 902 an EDNS(0) TCP Keepalive Option this is a fatal error and the 903 recipient of the defective message MUST forcibly abort the connection 904 immediately. 906 4.3. Message Handling 908 The initiator MUST set the value of the QR bit in the DNS header to 909 zero (0), and the responder MUST set it to one (1). 911 As described above in Section 4.2.1 whether an outgoing message with 912 QR=0 is unacknowledged or acknowledged is determined by the 913 specification for the Primary TLV, which in turn determines whether 914 the MESSAGE ID field in that outgoing message will be zero or 915 nonzero. 917 A DSO unacknowledged message has both the QR bit and the MESSAGE ID 918 field set to zero, and MUST NOT elicit a response. 920 Every DSO request message (QR=0) with a nonzero MESSAGE ID field is 921 an acknowledged DSO request, and MUST elicit a corresponding response 922 (QR=1), which MUST have the same MESSAGE ID in the DNS message header 923 as in the corresponding request. 925 Valid DSO request messages sent by the client with a nonzero MESSAGE 926 ID field elicit a response from the server, and Valid DSO request 927 messages sent by the server with a nonzero MESSAGE ID field elicit a 928 response from the client. 930 The namespaces of 16-bit MESSAGE IDs are independent in each 931 direction. This means it is *not* an error for both client and 932 server to send request messages at the same time as each other, using 933 the same MESSAGE ID, in different directions. This simplification is 934 necessary in order for the protocol to be implementable. It would be 935 infeasible to require the client and server to coordinate with each 936 other regarding allocation of new unique MESSAGE IDs. It is also not 937 necessary to require the client and server to coordinate with each 938 other regarding allocation of new unique MESSAGE IDs. The value of 939 the 16-bit MESSAGE ID combined with the identity of the initiator 940 (client or server) is sufficient to unambiguously identify the 941 operation in question. This can be thought of as a 17-bit message 942 identifier space, using message identifiers 0x00001-0x0FFFF for 943 client-to-server DSO request messages, and message identifiers 944 0x10001-0x1FFFF for server-to-client DSO request messages. The 945 least-significant 16 bits are stored explicitly in the MESSAGE ID 946 field of the DSO message, and the most-significant bit is implicit 947 from the direction of the message. 949 As described above in Section 4.2.1, an initiator MUST NOT reuse a 950 MESSAGE ID that it already has in use for an outstanding request 951 (unless specified otherwise by the relevant specification for the 952 DSO-TYPE in question). At the very least, this means that a MESSAGE 953 ID MUST NOT be reused in a particular direction on a particular DSO 954 Session while the initiator is waiting for a response to a previous 955 request using that MESSAGE ID on that DSO Session (unless specified 956 otherwise by the relevant specification for the DSO-TYPE in 957 question), and for a long-lived operation the MESSAGE ID for the 958 operation MUST NOT be reused while that operation remains active. 960 If a client or server receives a response (QR=1) where the MESSAGE ID 961 is zero, or is any other value that does not match the MESSAGE ID of 962 any of its outstanding operations, this is a fatal error and the 963 recipient MUST forcibly abort the connection immediately. 965 4.3.1. Error Responses 967 When a DSO unacknowledged message is unsuccessful for some reason, 968 the responder immediately aborts the connection. 970 When a DSO request message is unsuccessful for some reason, the 971 responder returns an error code to the initiator. 973 In the case of a server returning an error code to a client in 974 response to an unsuccessful DSO request message, the server MAY 975 choose to end the DSO Session, or MAY choose to allow the DSO Session 976 to remain open. For error conditions that only affect the single 977 operation in question, the server SHOULD return an error response to 978 the client and leave the DSO Session open for further operations. 980 For error conditions that are likely to make all operations 981 unsuccessful in the immediate future, the server SHOULD return an 982 error response to the client and then end the DSO Session by sending 983 a Retry Delay message, as described in Section 5.6.1. 985 Upon receiving an error response from the server, a client SHOULD NOT 986 automatically close the DSO Session. An error relating to one 987 particular operation on a DSO Session does not necessarily imply that 988 all other operations on that DSO Session have also failed, or that 989 future operations will fail. The client should assume that the 990 server will make its own decision about whether or not to end the DSO 991 Session, based on the server's determination of whether the error 992 condition pertains to this particular operation, or would also apply 993 to any subsequent operations. If the server does not end the DSO 994 Session by sending the client a Retry Delay message (Section 5.6.1) 995 then the client SHOULD continue to use that DSO Session for 996 subsequent operations. 998 4.4. DSO Response Generation 1000 With most TCP implementations, for DSO requests that generate a 1001 response, the TCP data acknowledgement (generated because data has 1002 been received by TCP), the TCP window update (generated because TCP 1003 has delivered that data to the receiving software), and the DSO 1004 response (generated by the receiving application-layer software 1005 itself) are all combined into a single IP packet. Combining these 1006 three elements into a single IP packet can give a significant 1007 improvement in network efficiency. 1009 For DSO requests that do not generate a response, the TCP 1010 implementation generally doesn't have any way to know that no 1011 response will be forthcoming, so it waits fruitlessly for the 1012 application-layer software to generate a response, until the Delayed 1013 ACK timer fires [RFC1122] (typically 200 milliseconds) and only then 1014 does it send the TCP ACK and window update. In conjunction with 1015 Nagle's Algorithm at the sender, this can delay the sender's 1016 transmission of its next (non-full-sized) TCP segment, while the 1017 sender is waiting for its previous (non-full-sized) TCP segment to be 1018 acknowledged, which won't happen until the Delayed ACK timer fires. 1019 Nagle's Algorithm exists to combine multiple small application writes 1020 into more-efficient large TCP segments, to guard against wasteful use 1021 of the network by applications that would otherwise transmit a stream 1022 of small TCP segments, but in this case Nagle's Algorithm (created to 1023 improve network efficiency) can interact badly with TCP's Delayed ACK 1024 feature (also created to improve network efficiency) [NagleDA] with 1025 the result of delaying some messages by up to 200 milliseconds. 1027 Possible mitigations for this problem include: 1029 o Disable Nagle's Algorithm at the sender. This is not great, 1030 because it results in less efficient use of the network. 1032 o Disable Delayed ACK at the receiver. This is not great, 1033 because it results in less efficient use of the network. 1035 o Use a networking API that lets the receiver signal to the TCP 1036 implementation that the receiver has received and processed a 1037 client request for which it will not be generating any immediate 1038 response. This allows the TCP implementation to operate 1039 efficiently in both cases; for requests that generate a response, 1040 the TCP ACK, window update, and DSO response are transmitted 1041 together in a single TCP segment, and for requests that do not 1042 generate a response, the application-layer software informs the 1043 TCP implementation that it should go ahead and send the TCP ACK 1044 and window update immediately, without waiting for the Delayed ACK 1045 timer. Unfortunately it is not known at this time which (if any) 1046 of the widely-available networking APIs currently include this 1047 capability. 1049 4.5. Responder-Initiated Operation Cancellation 1051 This document, the base specification for DNS Stateful Operations, 1052 does not itself define any long-lived operations, but it defines a 1053 framework for supporting long-lived operations, such as Push 1054 Notification subscriptions [I-D.ietf-dnssd-push] and Discovery Relay 1055 interface subscriptions [I-D.sctl-dnssd-mdns-relay]. 1057 Generally speaking, a long-lived operation is initiated by the 1058 initiator, and, if successful, remains active until the initiator 1059 terminates the operation. 1061 However, it is possible that a long-lived operation may be valid at 1062 the time it was initiated, but then a later change of circumstances 1063 may render that previously valid operation invalid. 1065 For example, a long-lived client operation may pertain to a name that 1066 the server is authoritative for, but then the server configuration is 1067 changed such that it is no longer authoritative for that name. 1069 In such cases, instead of terminating the entire session it may be 1070 desirable for the responder to be able to cancel selectively only 1071 those operations that have become invalid. 1073 The responder performs this selective cancellation by sending a new 1074 response message, with the MESSAGE ID field containing the MESSAGE ID 1075 of the long-lived operation that is to be terminated (that it had 1076 previously acknowledged with a NOERROR RCODE), and the RCODE field of 1077 the new response message giving the reason for cancellation. 1079 After a response message with nonzero RCODE has been sent, that 1080 operation has been terminated from the responder's point of view, and 1081 the responder sends no more messages relating to that operation. 1083 After a response message with nonzero RCODE has been received by the 1084 initiator, that operation has been terminated from the initiator's 1085 point of view, and the cancelled operation's MESSAGE ID is now free 1086 for reuse. 1088 5. DSO Session Lifecycle and Timers 1090 5.1. DSO Session Initiation 1092 A DSO Session begins as described in Section 4.1. 1094 The client may perform as many DNS operations as it wishes using the 1095 newly created DSO Session. Operations SHOULD be pipelined (i.e., the 1096 client doesn't need wait for a response before sending the next 1097 message). The server MUST act on messages in the order they are 1098 transmitted, but responses to those messages SHOULD be sent out of 1099 order when appropriate. 1101 5.2. DSO Session Timeouts 1103 Two timeout values are associated with a DSO Session: the inactivity 1104 timeout, and the keepalive interval. Both values are communicated in 1105 the same TLV, the DSO Keepalive TLV (Section 6.1). 1107 The first timeout value, the inactivity timeout, is the maximum time 1108 for which a client may speculatively keep a DSO Session open in the 1109 expectation that it may have future requests to send to that server. 1111 The second timeout value, the keepalive interval, is the maximum 1112 permitted interval between messages if the client wishes to keep the 1113 DSO Session alive. 1115 The two timeout values are independent. The inactivity timeout may 1116 be lower, the same, or higher than the keepalive interval, though in 1117 most cases the inactivity timeout is expected to be shorter than the 1118 keepalive interval. 1120 A shorter inactivity timeout with a longer keepalive interval signals 1121 to the client that it should not speculatively keep an inactive DSO 1122 Session open for very long without reason, but when it does have an 1123 active reason to keep a DSO Session open, it doesn't need to be 1124 sending an aggressive level of keepalive traffic to maintain that 1125 session. 1127 A longer inactivity timeout with a shorter keepalive interval signals 1128 to the client that it may speculatively keep an inactive DSO Session 1129 open for a long time, but to maintain that inactive DSO Session it 1130 should be sending a lot of keepalive traffic. This configuration is 1131 expected to be less common. 1133 In the usual case where the inactivity timeout is shorter than the 1134 keepalive interval, it is only when a client has a very long-lived, 1135 low-traffic, operation that the keepalive interval comes into play, 1136 to ensure that a sufficient residual amount of traffic is generated 1137 to maintain NAT and firewall state and to assure client and server 1138 that they still have connectivity to each other. 1140 On a new DSO Session, if no explicit DSO Keepalive message exchange 1141 has taken place, the default value for both timeouts is 15 seconds. 1143 For both timeouts, lower values of the timeout result in higher 1144 network traffic and higher CPU load on the server. 1146 5.3. Inactive DSO Sessions 1148 At both servers and clients, the generation or reception of any 1149 complete DNS message, including DNS requests, responses, updates, or 1150 DSO messages, resets both timers for that DSO Session, with the 1151 exception that a DSO Keepalive message resets only the keepalive 1152 timer, not the inactivity timeout timer. 1154 In addition, for as long as the client has an outstanding operation 1155 in progress, the inactivity timer remains cleared, and an inactivity 1156 timeout cannot occur. 1158 For short-lived DNS operations like traditional queries and updates, 1159 an operation is considered in progress for the time between request 1160 and response, typically a period of a few hundred milliseconds at 1161 most. At the client, the inactivity timer is cleared upon 1162 transmission of a request and remains cleared until reception of the 1163 corresponding response. At the server, the inactivity timer is 1164 cleared upon reception of a request and remains cleared until 1165 transmission of the corresponding response. 1167 For long-lived DNS Stateful operations (such as a Push Notification 1168 subscription [I-D.ietf-dnssd-push] or a Discovery Relay interface 1169 subscription [I-D.sctl-dnssd-mdns-relay]), an operation is considered 1170 in progress for as long as the operation is active, until it is 1171 cancelled. This means that a DSO Session can exist, with active 1172 operations, with no messages flowing in either direction, for far 1173 longer than the inactivity timeout, and this is not an error. This 1174 is why there are two separate timers: the inactivity timeout, and the 1175 keepalive interval. Just because a DSO Session has no traffic for an 1176 extended period of time does not automatically make that DSO Session 1177 "inactive", if it has an active operation that is awaiting events. 1179 5.4. The Inactivity Timeout 1181 The purpose of the inactivity timeout is for the server to balance 1182 its trade off between the costs of setting up new DSO Sessions and 1183 the costs of maintaining inactive DSO Sessions. A server with 1184 abundant DSO Session capacity can offer a high inactivity timeout, to 1185 permit clients to keep a speculative DSO Session open for a long 1186 time, to save the cost of establishing a new DSO Session for future 1187 communications with that server. A server with scarce memory 1188 resources can offer a low inactivity timeout, to cause clients to 1189 promptly close DSO Sessions whenever they have no outstanding 1190 operations with that server, and then create a new DSO Session later 1191 when needed. 1193 5.4.1. Closing Inactive DSO Sessions 1195 When a connection's inactivity timeout is reached the client MUST 1196 begin closing the idle connection, but a client is NOT REQUIRED to 1197 keep an idle connection open until the inactivity timeout is reached. 1198 A client MAY close a DSO Session at any time, at the client's 1199 discretion. If a client determines that it has no current or 1200 reasonably anticipated future need for a currently inactive DSO 1201 Session, then the client SHOULD gracefully close that connection. 1203 If, at any time during the life of the DSO Session, the inactivity 1204 timeout value (i.e., 15 seconds by default) elapses without there 1205 being any operation active on the DSO Session, the client MUST close 1206 the connection gracefully. 1208 If, at any time during the life of the DSO Session, twice the 1209 inactivity timeout value (i.e., 30 seconds by default), or five 1210 seconds, if twice the inactivity timeout value is less than five 1211 seconds, elapses without there being any operation active on the DSO 1212 Session, the server SHOULD consider the client delinquent, and SHOULD 1213 forcibly abort the DSO Session. 1215 In this context, an operation being active on a DSO Session includes 1216 a query waiting for a response, an update waiting for a response, or 1217 an active long-lived operation, but not a DSO Keepalive message 1218 exchange itself. A DSO Keepalive message exchange resets only the 1219 keepalive interval timer, not the inactivity timeout timer. 1221 If the client wishes to keep an inactive DSO Session open for longer 1222 than the default duration then it uses the DSO Keepalive message to 1223 request longer timeout values, as described in Section 6.1. 1225 5.4.2. Values for the Inactivity Timeout 1227 For the inactivity timeout value, lower values result in more 1228 frequent DSO Session teardown and re-establishment. Higher values 1229 result in lower traffic and lower CPU load on the server, but higher 1230 memory burden to maintain state for inactive DSO Sessions. 1232 A server may dictate any value it chooses for the inactivity timeout 1233 (either in a response to a client-initiated request, or in a server- 1234 initiated message) including values under one second, or even zero. 1236 An inactivity timeout of zero informs the client that it should not 1237 speculatively maintain idle connections at all, and as soon as the 1238 client has completed the operation or operations relating to this 1239 server, the client should immediately begin closing this session. 1241 A server will abort an idle client session after twice the inactivity 1242 timeout value, or five seconds, whichever is greater. In the case of 1243 a zero inactivity timeout value, this means that if a client fails to 1244 close an idle client session then the server will forcibly abort the 1245 idle session after five seconds. 1247 An inactivity timeout of 0xFFFFFFFF represents "infinity" and informs 1248 the client that it may keep an idle connection open as long as it 1249 wishes. Note that after granting an unlimited inactivity timeout in 1250 this way, at any point the server may revise that inactivity timeout 1251 by sending a new Keepalive message dictating new Session Timeout 1252 values to the client. 1254 The largest *finite* inactivity timeout supported by the current DSO 1255 Keepalive TLV is 0xFFFFFFFE (2^32-2 milliseconds, approximately 49.7 1256 days). 1258 5.5. The Keepalive Interval 1260 The purpose of the keepalive interval is to manage the generation of 1261 sufficient messages to maintain state in middleboxes (such at NAT 1262 gateways or firewalls) and for the client and server to periodically 1263 verify that they still have connectivity to each other. This allows 1264 them to clean up state when connectivity is lost, and to establish a 1265 new session if appropriate. 1267 5.5.1. Keepalive Interval Expiry 1269 If, at any time during the life of the DSO Session, the keepalive 1270 interval value (i.e., 15 seconds by default) elapses without any DNS 1271 messages being sent or received on a DSO Session, the client MUST 1272 take action to keep the DSO Session alive, by sending a DSO Keepalive 1273 message (Section 6.1). A DSO Keepalive message exchange resets only 1274 the keepalive timer, not the inactivity timer. 1276 If a client disconnects from the network abruptly, without cleanly 1277 closing its DSO Session, perhaps leaving a long-lived operation 1278 uncancelled, the server learns of this after failing to receive the 1279 required keepalive traffic from that client. If, at any time during 1280 the life of the DSO Session, twice the keepalive interval value 1281 (i.e., 30 seconds by default) elapses without any DNS messages being 1282 sent or received on a DSO Session, the server SHOULD consider the 1283 client delinquent, and SHOULD forcibly abort the DSO Session. 1285 5.5.2. Values for the Keepalive Interval 1287 For the keepalive interval value, lower values result in a higher 1288 volume of keepalive traffic. Higher values of the keepalive interval 1289 reduce traffic and CPU load, but have minimal effect on the memory 1290 burden at the server, because clients keep a DSO Session open for the 1291 same length of time (determined by the inactivity timeout) regardless 1292 of the level of keepalive traffic required. 1294 It may be appropriate for clients and servers to select different 1295 keepalive interval values depending on the nature of the network they 1296 are on. 1298 A corporate DNS server that knows it is serving only clients on the 1299 internal network, with no intervening NAT gateways or firewalls, can 1300 impose a higher keepalive interval, because frequent keepalive 1301 traffic is not required. 1303 A public DNS server that is serving primarily residential consumer 1304 clients, where it is likely there will be a NAT gateway on the path, 1305 may impose a lower keepalive interval, to generate more frequent 1306 keepalive traffic. 1308 A smart client may be adaptive to its environment. A client using a 1309 private IPv4 address [RFC1918] to communicate with a DNS server at an 1310 address outside that IPv4 private address block, may conclude that 1311 there is likely to be a NAT gateway on the path, and accordingly 1312 request a lower keepalive interval. 1314 By default it is RECOMMENDED that clients request, and servers grant, 1315 a keepalive interval of 60 minutes. This keepalive interval provides 1316 for reasonably timely detection if a client abruptly disconnects 1317 without cleanly closing the session, and is sufficient to maintain 1318 state in firewalls and NAT gateways that follow the IETF recommended 1319 Best Current Practice that the "established connection idle-timeout" 1320 used by middleboxes be at least 2 hours 4 minutes [RFC5382]. 1322 Note that the lower the keepalive interval value, the higher the load 1323 on client and server. For example, a hypothetical keepalive interval 1324 value of 100ms would result in a continuous stream of at least ten 1325 messages per second, in both directions, to keep the DSO Session 1326 alive. And, in this extreme example, a single packet loss and 1327 retransmission over a long path could introduce a momentary pause in 1328 the stream of messages, long enough to cause the server to 1329 overzealously abort the connection. 1331 Because of this concern, the server MUST NOT send a Keepalive message 1332 (either a response to a client-initiated request, or a server- 1333 initiated message) with a keepalive interval value less than ten 1334 seconds. If a client receives a Keepalive message specifying a 1335 keepalive interval value less than ten seconds this is a fatal error 1336 and the client MUST forcibly abort the connection immediately. 1338 A keepalive interval value of 0xFFFFFFFF represents "infinity" and 1339 informs the client that it should generate no keepalive traffic. 1340 Note that after signaling that the client should generate no 1341 keepalive traffic in this way, at any point the server may revise 1342 that keepalive traffic requirement by sending a new Keepalive message 1343 dictating new Session Timeout values to the client. 1345 The largest *finite* keepalive interval supported by the current DSO 1346 Keepalive TLV is 0xFFFFFFFE (2^32-2 milliseconds, approximately 49.7 1347 days). 1349 5.6. Server-Initiated Session Termination 1351 In addition to cancelling individual long-lived operations 1352 selectively (Section 4.5) there are also occasions where a server may 1353 need to terminate one or more entire sessions. An entire session may 1354 need to be terminated if the client is defective in some way, or 1355 departs from the network without closing its session. Sessions may 1356 also need to be terminated if the server becomes overloaded, or if 1357 the server is reconfigured and lacks the ability to be selective 1358 about which operations need to be cancelled. 1360 This section discusses various reasons a session may be terminated, 1361 and the mechanisms for doing so. 1363 Normally a server MUST NOT close a DSO Session with a client. A 1364 server only causes a DSO Session to be ended in the exceptional 1365 circumstances outlined below. In normal operation, closing a DSO 1366 Session is the client's responsibility. The client makes the 1367 determination of when to close a DSO Session based on an evaluation 1368 of both its own needs, and the inactivity timeout value dictated by 1369 the server. 1371 Some of the exceptional situations in which a server may terminate a 1372 DSO Session include: 1374 o The server application software or underlying operating system is 1375 shutting down or restarting. 1377 o The server application software terminates unexpectedly (perhaps 1378 due to a bug that makes it crash). 1380 o The server is undergoing a reconfiguration or maintenance 1381 procedure, that, due to the way the server software is 1382 implemented, requires clients to be disconnected. For example, 1383 some software is implemented such that it reads a configuration 1384 file at startup, and changing the server's configuration entails 1385 modifying the configuration file and then killing and restarting 1386 the server software, which generally entails a loss of network 1387 connections. 1389 o The client fails to meets its obligation to generate the required 1390 keepalive traffic, or to close an inactive session by the 1391 prescribed time (twice the time interval dictated by the server, 1392 or five seconds, whichever is greater, as described in 1393 Section 5.2). 1395 o The client sends a grossly invalid or malformed request that is 1396 indicative of a seriously defective client implementation. 1398 o The server is over capacity and needs to shed some load. 1400 5.6.1. Server-Initiated Retry Delay Message 1402 In the cases described above where a server elects to terminate a DSO 1403 Session, it could do so simply by forcibly aborting the connection. 1404 However, if it did this the likely behavior of the client might be 1405 simply to to treat this as a network failure and reconnect 1406 immediately, putting more burden on the server. 1408 Therefore, to avoid this reconnection implosion, a server SHOULD 1409 instead choose to shed client load by sending a Retry Delay message, 1410 with an appropriate RCODE value informing the client of the reason 1411 the DSO Session needs to be terminated. The format of the Retry 1412 Delay TLV, and the interpretations of the various RCODE values, are 1413 described in Section 6.2. After sending a Retry Delay message, the 1414 server MUST NOT send any further messages on that DSO Session. 1416 Upon receipt of a Retry Delay message from the server, the client 1417 MUST make note of the reconnect delay for this server, and then 1418 immediately close the connection gracefully. 1420 After sending a Retry Delay message the server SHOULD allow the 1421 client five seconds to close the connection, and if the client has 1422 not closed the connection after five seconds then the server SHOULD 1423 forcibly abort the connection. 1425 A Retry Delay message MUST NOT be initiated by a client. If a server 1426 receives a Retry Delay message this is a fatal error and the server 1427 MUST forcibly abort the connection immediately. 1429 5.6.1.1. Outstanding Operations 1431 At the instant a server chooses to initiate a Retry Delay message 1432 there may be DNS requests already in flight from client to server on 1433 this DSO Session, which will arrive at the server after its Retry 1434 Delay message has been sent. The server MUST silently ignore such 1435 incoming requests, and MUST NOT generate any response messages for 1436 them. When the Retry Delay message from the server arrives at the 1437 client, the client will determine that any DNS requests it previously 1438 sent on this DSO Session, that have not yet received a response, now 1439 will certainly not be receiving any response. Such requests should 1440 be considered failed, and should be retried at a later time, as 1441 appropriate. 1443 In the case where some, but not all, of the existing operations on a 1444 DSO Session have become invalid (perhaps because the server has been 1445 reconfigured and is no longer authoritative for some of the names), 1446 but the server is terminating all affected DSO Sessions en masse by 1447 sending them all a Retry Delay message, the RECONNECT DELAY MAY be 1448 zero, indicating that the clients SHOULD immediately attempt to re- 1449 establish operations. 1451 It is likely that some of the attempts will be successful and some 1452 will not, depending on the nature of the reconfiguration. 1454 In the case where a server is terminating a large number of DSO 1455 Sessions at once (e.g., if the system is restarting) and the server 1456 doesn't want to be inundated with a flood of simultaneous retries, it 1457 SHOULD send different RECONNECT delay values to each client. These 1458 adjustments MAY be selected randomly, pseudorandomly, or 1459 deterministically (e.g., incrementing the time value by one tenth of 1460 a second for each successive client, yielding a post-restart 1461 reconnection rate of ten clients per second). 1463 5.6.1.2. Client Reconnection 1465 After a DSO Session is ended by the server (either by sending the 1466 client a Retry Delay message, or by forcibly aborting the underlying 1467 transport connection) the client SHOULD try to reconnect, to that 1468 server instance, or to another suitable server instance, if more than 1469 one is available. If reconnecting to the same server instance, the 1470 client MUST respect the indicated delay, if available, before 1471 attempting to reconnect. 1473 If the server instance will only be out of service for a short 1474 maintenance period, it should use a value a little longer that the 1475 expected maintenance window. It should not default to a very large 1476 delay value, or clients may not attempt to reconnect after it resumes 1477 service. 1479 If a particular server instance does not want a client to reconnect 1480 ever (perhaps the server instance is being de-commissioned), it 1481 SHOULD set the retry delay to the maximum value 0xFFFFFFFF (2^32-1 1482 milliseconds, approximately 49.7 days). It is not possible to 1483 instruct a client to stay away for longer than 49.7 days. If, after 1484 49.7 days, the DNS or other configuration information still indicates 1485 that this is the valid server instance for a particular service, then 1486 clients MAY attempt to reconnect. In reality, if a client is 1487 rebooted or otherwise lose state, it may well attempt to reconnect 1488 before 49.7 days elapses, for as long as the DNS or other 1489 configuration information continues to indicate that this is the 1490 server instance the client should use. 1492 6. Base TLVs for DNS Stateful Operations 1494 This section describes the three base TLVs for DNS Stateful 1495 Operations: Keepalive, Retry Delay, and Encryption Padding. 1497 6.1. Keepalive TLV 1499 The Keepalive TLV (DSO-TYPE=1) performs two functions: to reset the 1500 keepalive timer for the DSO Session, and to establish the values for 1501 the Session Timeouts. 1503 The DSO-DATA for the the Keepalive TLV is as follows: 1505 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 3 3 1506 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 1507 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1508 | INACTIVITY TIMEOUT (32 bits) | 1509 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1510 | KEEPALIVE INTERVAL (32 bits) | 1511 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1513 INACTIVITY TIMEOUT: The inactivity timeout for the current DSO 1514 Session, specified as a 32-bit unsigned integer, in network (big 1515 endian) byte order, in units of milliseconds. This is the timeout 1516 at which the client MUST begin closing an inactive DSO Session. 1517 The inactivity timeout can be any value of the server's choosing. 1518 If the client does not gracefully close an inactive DSO Session, 1519 then after twice this interval, or five seconds, whichever is 1520 greater, the server will forcibly abort the connection. 1522 KEEPALIVE INTERVAL: The keepalive interval for the current DSO 1523 Session, specified as a 32-bit unsigned integer, in network (big 1524 endian) byte order, in units of milliseconds. This is the 1525 interval at which a client MUST generate keepalive traffic to 1526 maintain connection state. The keepalive interval MUST NOT be 1527 less than ten seconds. If the client does not generate the 1528 mandated keepalive traffic, then after twice this interval the 1529 server will forcibly abort the connection. Since the minimum 1530 allowed keepalive interval is ten seconds, the minimum time at 1531 which a server will forcibly disconnect a client for failing to 1532 generate the mandated keepalive traffic is twenty seconds. 1534 The transmission or reception of DSO Keepalive messages (i.e., 1535 messages where the Keepalive TLV is the first TLV) reset only the 1536 keepalive timer, not the inactivity timer. The reason for this is 1537 that periodic Keepalive messages are sent for the sole purpose of 1538 keeping a DSO Session alive, when that DSO Session has current or 1539 recent non-maintenance activity that warrants keeping that DSO 1540 Session alive. Sending keepalive traffic itself is not considered a 1541 client activity; it is considered a maintenance activity that is 1542 performed in service of other client activities. If keepalive 1543 traffic itself were to reset the inactivity timer, then that would 1544 create a circular livelock where keepalive traffic would be sent 1545 indefinitely to keep a DSO Session alive, where the only activity on 1546 that DSO Session would be the keepalive traffic keeping the DSO 1547 Session alive so that further keepalive traffic can be sent. For a 1548 DSO Session to be considered active, it must be carrying something 1549 more than just keepalive traffic. This is why merely sending or 1550 receiving a Keepalive message does not reset the inactivity timer. 1552 When sent by a client, the Keepalive request message MUST be sent as 1553 an acknowledged request, with a nonzero MESSAGE ID. If a server 1554 receives a Keepalive DSO message with a zero MESSAGE ID then this is 1555 a fatal error and the server MUST forcibly abort the connection 1556 immediately. The Keepalive request message resets a DSO Session's 1557 keepalive timer, and at the same time communicates to the server the 1558 the client's requested Session Timeout values. In a server response 1559 to a client-initiated Keepalive request message, the Session Timeouts 1560 contain the server's chosen values from this point forward in the DSO 1561 Session, which the client MUST respect. This is modeled after the 1562 DHCP protocol, where the client requests a certain lease lifetime 1563 using DHCP option 51 [RFC2132], but the server is the ultimate 1564 authority for deciding what lease lifetime is actually granted. 1566 When a client is sending its second and subsequent Keepalive DSO 1567 requests to the server, the client SHOULD continue to request its 1568 preferred values each time. This allows flexibility, so that if 1569 conditions change during the lifetime of a DSO Session, the server 1570 can adapt its responses to better fit the client's needs. 1572 Once a DSO Session is in progress (Section 4.1) a Keepalive message 1573 MAY be initiated by a server. When sent by a server, the Keepalive 1574 message MUST be sent as an unacknowledged message, with the MESSAGE 1575 ID set to zero. The client MUST NOT generate a response to a server- 1576 initiated DSO Keepalive message. If a client receives a Keepalive 1577 request message with a nonzero MESSAGE ID then this is a fatal error 1578 and the client MUST forcibly abort the connection immediately. The 1579 Keepalive unacknowledged message from the server resets a DSO 1580 Session's keepalive timer, and at the same time unilaterally informs 1581 the client of the new Session Timeout values to use from this point 1582 forward in this DSO Session. No client DSO response message to this 1583 unilateral declaration is required or allowed. 1585 The Keepalive TLV is not used as an Additional TLV. 1587 In response messages the Keepalive TLV is used only as a Response 1588 Primary TLV, replying to a Keepalive request message from the client. 1589 A Keepalive TLV MUST NOT be added as to other responses a Response 1590 Additional TLV. If the server wishes to update a client's Session 1591 Timeout values other than in response to a Keepalive request message 1592 from the client, then it does so by sending an unacknowledged 1593 Keepalive message of its own, as described above. 1595 It is not required that the Keepalive TLV be used in every DSO 1596 Session. While many DNS Stateful operations will be used in 1597 conjunction with a long-lived session state, not all DNS Stateful 1598 operations require long-lived session state, and in some cases the 1599 default 15-second value for both the inactivity timeout and keepalive 1600 interval may be perfectly appropriate. However, note that for 1601 clients that implement only the DSO-TYPEs defined in this document, a 1602 Keepalive request message is the only way for a client to initiate a 1603 DSO Session. 1605 6.1.1. Client handling of received Session Timeout values 1607 When a client receives a response to its client-initiated DSO 1608 Keepalive message, or receives a server-initiated DSO Keepalive 1609 message, the client has then received Session Timeout values dictated 1610 by the server. The two timeout values contained in the DSO Keepalive 1611 TLV from the server may each be higher, lower, or the same as the 1612 respective Session Timeout values the client previously had for this 1613 DSO Session. 1615 In the case of the keepalive timer, the handling of the received 1616 value is straightforward. The act of receiving the message 1617 containing the DSO Keepalive TLV itself resets the keepalive timer 1618 and updates the keepalive interval for the DSO Session. The new 1619 keepalive interval indicates the maximum time that may elapse before 1620 another message must be sent or received on this DSO Session, if the 1621 DSO Session is to remain alive. 1623 In the case of the inactivity timeout, the handling of the received 1624 value is a little more subtle, though the meaning of the inactivity 1625 timeout remains as specified -- it still indicates the maximum 1626 permissible time allowed without useful activity on a DSO Session. 1627 The act of receiving the message containing the DSO Keepalive TLV 1628 does not itself reset the inactivity timer. The time elapsed since 1629 the last useful activity on this DSO Session is unaffected by 1630 exchange of DSO Keepalive messages. The new inactivity timeout value 1631 in the DSO Keepalive TLV in the received message does update the 1632 timeout associated with the running inactivity timer; that becomes 1633 the new maximum permissible time without activity on a DSO Session. 1635 o If the current inactivity timer value is less than the new 1636 inactivity timeout, then the DSO Session may remain open for now. 1637 When the inactivity timer value reaches the new inactivity 1638 timeout, the client MUST then begin closing the DSO Session, as 1639 described above. 1641 o If the current inactivity timer value is equal to the new 1642 inactivity timeout, then this DSO Session has been inactive for 1643 exactly as long as the server will permit, and now the client MUST 1644 immediately begin closing this DSO Session. 1646 o If the current inactivity timer value is already greater than the 1647 new inactivity timeout, then this DSO Session has already been 1648 inactive for longer than the server permits, and the client MUST 1649 immediately begin closing this DSO Session. 1651 o If the current inactivity timer value is already more than twice 1652 the new inactivity timeout, then the client is immediately 1653 considered delinquent (this DSO Session is immediately eligible to 1654 be forcibly terminated by the server) and the client MUST 1655 immediately begin closing this DSO Session. However if a server 1656 abruptly reduces the inactivity timeout in this way, then, to give 1657 the client time to close the connection gracefully before the 1658 server resorts to forcibly aborting it, the server SHOULD give the 1659 client an additional grace period of one quarter of the new 1660 inactivity timeout, or five seconds, whichever is greater. 1662 6.1.2. Relation to EDNS(0) TCP Keepalive Option 1664 The inactivity timeout value in the Keepalive TLV (DSO-TYPE=1) has 1665 similar intent to the EDNS(0) TCP Keepalive Option [RFC7828]. A 1666 client/server pair that supports DSO MUST NOT use the EDNS(0) TCP 1667 KeepAlive option within any message after a DSO Session has been 1668 established. Once a DSO Session has been established, if either 1669 client or server receives a DNS message over the DSO Session that 1670 contains an EDNS(0) TCP Keepalive option, this is a fatal error and 1671 the receiver of the EDNS(0) TCP Keepalive option MUST forcibly abort 1672 the connection immediately. 1674 6.2. Retry Delay TLV 1676 The Retry Delay TLV (DSO-TYPE=2) can be used as a Primary TLV 1677 (unacknowledged) in a server-to-client message, or as a Response 1678 Additional TLV in either direction. 1680 The DSO-DATA for the the Retry Delay TLV is as follows: 1682 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 3 3 1683 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 1684 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1685 | RETRY DELAY (32 bits) | 1686 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1688 RETRY DELAY: A time value, specified as a 32-bit unsigned integer, 1689 in network (big endian) byte order, in units of milliseconds, 1690 within which the initiator MUST NOT retry this operation, or retry 1691 connecting to this server. Recommendations for the RETRY DELAY 1692 value are given in Section 5.6.1. 1694 6.2.1. Retry Delay TLV used as a Primary TLV 1696 When sent from server to client, the Retry Delay TLV is used as the 1697 Primary TLV in an unacknowledged message. It is used by a server to 1698 instruct a client to close the DSO Session and underlying connection, 1699 and not to reconnect for the indicated time interval. 1701 In this case it applies to the DSO Session as a whole, and the client 1702 MUST begin closing the DSO Session, as described in Section 5.6.1. 1703 The RCODE in the message header SHOULD indicate the principal reason 1704 for the termination: 1706 o NOERROR indicates a routine shutdown or restart. 1708 o FORMERR indicates that the client requests are too badly malformed 1709 for the session to continue. 1711 o SERVFAIL indicates that the server is overloaded due to resource 1712 exhaustion and needs to shed load. 1714 o REFUSED indicates that the server has been reconfigured, and at 1715 this time it is now unable to perform one or more of the long- 1716 lived client operations that were previously being performed on 1717 this DSO Session. 1719 o NOTAUTH indicates that the server has been reconfigured and at 1720 this time it is now unable to perform one or more of the long- 1721 lived client operations that were previously being performed on 1722 this DSO Session because it does not have authority over the names 1723 in question (for example, a DNS Push Notification server could be 1724 reconfigured such that is is no longer accepting DNS Push 1725 Notification requests for one or more of the currently subscribed 1726 names). 1728 This document specifies only these RCODE values for Retry Delay 1729 message. Servers sending Retry Delay messages SHOULD use one of 1730 these values. However, future circumstances may create situations 1731 where other RCODE values are appropriate in Retry Delay messages, so 1732 clients MUST be prepared to accept Retry Delay messages with any 1733 RCODE value. 1735 In some cases, when a server sends a Retry Delay message to a client, 1736 there may be more than one reason for the server wanting to end the 1737 session. Possibly the configuration could have been changed such 1738 that some long-lived client operations can no longer be continued due 1739 to policy (REFUSED), and other long-lived client operations can no 1740 longer be performed due to the server no longer being authoritative 1741 for those names (NOTAUTH). In such cases the server MAY use any of 1742 the applicable RCODE values, or RCODE=NOERROR (routine shutdown or 1743 restart). 1745 Note that the selection of RCODE value in a Retry Delay message is 1746 not critical, since the RCODE value is generally used only for 1747 information purposes, such as writing to a log file for future human 1748 analysis regarding the nature of the disconnection. Generally 1749 clients do not modify their behavior depending on the RCODE value. 1750 The RETRY DELAY in the message tells the client how long it should 1751 wait before attempting a new connection to this server instance. 1753 For clients that do in some way modify their behavior depending on 1754 the RCODE value, they should treat unknown RCODE values the same as 1755 RCODE=NOERROR (routine shutdown or restart). 1757 A Retry Delay message from server to client is an unacknowledged 1758 message; the MESSAGE ID MUST be set to zero in the outgoing message 1759 and the client MUST NOT send a response. 1761 A client MUST NOT send a Retry Delay DSO request message or DSO 1762 unacknowledged message to a server. If a server receives a DNS 1763 request message (i.e., QR=0) where the Primary TLV is the Retry Delay 1764 TLV, this is a fatal error and the server MUST forcibly abort the 1765 connection immediately. 1767 6.2.2. Retry Delay TLV used as a Response Additional TLV 1769 In the case of a request that returns a nonzero RCODE value, the 1770 responder MAY append a Retry Delay TLV to the response, indicating 1771 the time interval during which the initiator SHOULD NOT attempt this 1772 operation again. 1774 The indicated time interval during which the initiator SHOULD NOT 1775 retry applies only to the failed operation, not to the DSO Session as 1776 a whole. 1778 6.3. Encryption Padding TLV 1780 The Encryption Padding TLV (DSO-TYPE=3) can only be used as an 1781 Additional or Response Additional TLV. It is only applicable when 1782 the DSO Transport layer uses encryption such as TLS. 1784 The DSO-DATA for the the Padding TLV is optional and is a variable 1785 length field containing non-specified values. A DSO-LENGTH of 0 1786 essentially provides for 4 bytes of padding (the minimum amount). 1788 1 1 1 1 1 1 1789 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 1790 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1791 / / 1792 / VARIABLE NUMBER OF BYTES / 1793 / / 1794 +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+ 1796 As specified for the EDNS(0) Padding Option [RFC7830] the PADDING 1797 bytes SHOULD be set to 0x00. Other values MAY be used, for example, 1798 in cases where there is a concern that the padded message could be 1799 subject to compression before encryption. PADDING bytes of any value 1800 MUST be accepted in the messages received. 1802 The Encryption Padding TLV may be included in either a DSO request, 1803 response, or both. As specified for the EDNS(0) Padding Option 1804 [RFC7830] if a request is received with an Encryption Padding TLV, 1805 then the response MUST also include an Encryption Padding TLV. 1807 The length of padding is intentionally not specified in this document 1808 and is a function of current best practices with respect to the type 1809 and length of data in the preceding TLVs 1810 [I-D.ietf-dprive-padding-policy]. 1812 7. Summary Highlights 1814 This section summarizes some noteworthy highlights about various 1815 components of the DSO protocol. 1817 7.1. QR bit and MESSAGE ID 1819 In DSO Request Messages the QR bit is 0 and the MESSAGE ID is 1820 nonzero. 1822 In DSO Response Messages the QR bit is 1 and the MESSAGE ID is 1823 nonzero. 1825 In DSO Unacknowledged Messages the QR bit is 0 and the MESSAGE ID is 1826 zero. 1828 The table below illustrates which combinations are legal and how they 1829 are interpreted: 1831 +--------------------------+------------------------+ 1832 | MESSAGE ID zero | MESSAGE ID nonzero | 1833 +--------+--------------------------+------------------------+ 1834 | QR=0 | Unacknowledged Message | Request Message | 1835 +--------+--------------------------+------------------------+ 1836 | QR=1 | Invalid - Fatal Error | Response Message | 1837 +--------+--------------------------+------------------------+ 1839 7.2. TLV Usage 1841 The table below indicates, for each of the three TLVs defined in this 1842 document, whether they are valid in each of ten different contexts. 1844 The first five contexts are requests or unacknowledged messages from 1845 client to server, and the corresponding responses from server back to 1846 client: 1848 o C-P - Primary TLV, sent in DSO Request message, from client to 1849 server, with nonzero MESSAGE ID indicating that this request MUST 1850 generate response message. 1852 o C-U - Primary TLV, sent in DSO Unacknowledged message, from client 1853 to server, with zero MESSAGE ID indicating that this request MUST 1854 NOT generate response message. 1856 o C-A - Additional TLV, optionally added to request message or 1857 unacknowledged message from client to server. 1859 o CRP - Response Primary TLV, included in response message sent to 1860 back the client (in response to a client "C-P" request with 1861 nonzero MESSAGE ID indicating that a response is required) where 1862 the DSO-TYPE of the Response TLV matches the DSO-TYPE of the 1863 Primary TLV in the request. 1865 o CRA - Response Additional TLV, included in response message sent 1866 to back the client (in response to a client "C-P" request with 1867 nonzero MESSAGE ID indicating that a response is required) where 1868 the DSO-TYPE of the Response TLV does not match the DSO-TYPE of 1869 the Primary TLV in the request. 1871 The second five contexts are their counterparts in the opposite 1872 direction: requests or unacknowledged messages from server to client, 1873 and the corresponding responses from client back to server. 1875 +-------------------------+-------------------------+ 1876 | C-P C-U C-A CRP CRA | S-P S-U S-A SRP SRA | 1877 +------------+-------------------------+-------------------------+ 1878 | KeepAlive | X X | X | 1879 +------------+-------------------------+-------------------------+ 1880 | RetryDelay | X | X | 1881 +------------+-------------------------+-------------------------+ 1882 | Padding | X X | X X | 1883 +------------+-------------------------+-------------------------+ 1885 Note that some of the columns in this table are currently empty. The 1886 table provides a template for future TLV definitions to follow. It 1887 is recommended that definitions of future TLVs include a similar 1888 table summarizing the contexts where the new TLV is valid. 1890 8. IANA Considerations 1892 8.1. DSO OPCODE Registration 1894 The IANA is requested to record the value (tentatively) 6 for the 1895 DSO OPCODE in the DNS OPCODE Registry. DSO stands for DNS Stateful 1896 Operations. 1898 8.2. DSO RCODE Registration 1900 The IANA is requested to record the value (tentatively) 11 for the 1901 DSOTYPENI error code in the DNS RCODE Registry. The DSOTYPENI error 1902 code ("DSO-TYPE Not Implemented") indicates that the receiver does 1903 implement DNS Stateful Operations, but does not implement the 1904 specific DSO-TYPE of the primary TLV in the DSO request message. 1906 8.3. DSO Type Code Registry 1908 The IANA is requested to create the 16-bit DSO Type Code Registry, 1909 with initial (hexadecimal) values as shown below: 1911 +-----------+--------------------------------+----------+-----------+ 1912 | Type | Name | Status | Reference | 1913 +-----------+--------------------------------+----------+-----------+ 1914 | 0000 | Reserved | Standard | RFC-TBD | 1915 | | | | | 1916 | 0001 | KeepAlive | Standard | RFC-TBD | 1917 | | | | | 1918 | 0002 | RetryDelay | Standard | RFC-TBD | 1919 | | | | | 1920 | 0003 | EncryptionPadding | Standard | RFC-TBD | 1921 | | | | | 1922 | 0004-003F | Unassigned, reserved for | | | 1923 | | DSO session-management TLVs | | | 1924 | | | | | 1925 | 0040-F7FF | Unassigned | | | 1926 | | | | | 1927 | F800-FBFF | Reserved for | | | 1928 | | experimental/local use | | | 1929 | | | | | 1930 | FC00-FFFF | Reserved for future expansion | | | 1931 +-----------+--------------------------------+----------+-----------+ 1933 DSO Type Code zero is reserved and is not currently intended for 1934 allocation. 1936 Registrations of new DSO Type Codes in the "Reserved for DSO session- 1937 management" range 0004-003F and the "Reserved for future expansion" 1938 range FC00-FFFF require publication of an IETF Standards Action 1939 document [RFC8126]. 1941 Requests to register additional new DSO Type Codes in the 1942 "Unassigned" range 0040-F7FF are to be recorded by IANA after Expert 1943 Review [RFC8126]. At the time of publication of this document, the 1944 Designated Expert for the newly created DSO Type Code registry is 1945 [*TBD*]. 1947 DSO Type Codes in the "experimental/local" range F800-FBFF may be 1948 used as Experimental Use or Private Use values [RFC8126] and may be 1949 used freely for development purposes, or for other purposes within a 1950 single site. No attempt is made to prevent multiple sites from using 1951 the same value in different (and incompatible) ways. There is no 1952 need for IANA to review such assignments (since IANA does not record 1953 them) and assignments are not generally useful for broad 1954 interoperability. It is the responsibility of the sites making use 1955 of "experimental/local" values to ensure that no conflicts occur 1956 within the intended scope of use. 1958 9. Security Considerations 1960 If this mechanism is to be used with DNS over TLS, then these 1961 messages are subject to the same constraints as any other DNS-over- 1962 TLS messages and MUST NOT be sent in the clear before the TLS session 1963 is established. 1965 The data field of the "Encryption Padding" TLV could be used as a 1966 covert channel. 1968 When designing new DSO TLVs, the potential for data in the TLV to be 1969 used as a tracking identifier should be taken into consideration, and 1970 should be avoided when not required. 1972 When used without TLS or similar cryptographic protection, a 1973 malicious entity maybe able to inject a malicious Retry Delay 1974 Unacknowledged Message into the data stream, specifying an 1975 unreasonably large RETRY DELAY, causing a denial-of-service attack 1976 against the client. 1978 10. Acknowledgements 1980 Thanks to Stephane Bortzmeyer, Tim Chown, Ralph Droms, Paul Hoffman, 1981 Jan Komissar, Edward Lewis, Allison Mankin, Rui Paulo, David 1982 Schinazi, Manju Shankar Rao, and Bernie Volz for their helpful 1983 contributions to this document. 1985 11. References 1987 11.1. Normative References 1989 [RFC1034] Mockapetris, P., "Domain names - concepts and facilities", 1990 STD 13, RFC 1034, DOI 10.17487/RFC1034, November 1987, 1991 . 1993 [RFC1035] Mockapetris, P., "Domain names - implementation and 1994 specification", STD 13, RFC 1035, DOI 10.17487/RFC1035, 1995 November 1987, . 1997 [RFC1918] Rekhter, Y., Moskowitz, B., Karrenberg, D., de Groot, G., 1998 and E. Lear, "Address Allocation for Private Internets", 1999 BCP 5, RFC 1918, DOI 10.17487/RFC1918, February 1996, 2000 . 2002 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 2003 Requirement Levels", BCP 14, RFC 2119, 2004 DOI 10.17487/RFC2119, March 1997, 2005 . 2007 [RFC2132] Alexander, S. and R. Droms, "DHCP Options and BOOTP Vendor 2008 Extensions", RFC 2132, DOI 10.17487/RFC2132, March 1997, 2009 . 2011 [RFC2136] Vixie, P., Ed., Thomson, S., Rekhter, Y., and J. Bound, 2012 "Dynamic Updates in the Domain Name System (DNS UPDATE)", 2013 RFC 2136, DOI 10.17487/RFC2136, April 1997, 2014 . 2016 [RFC5382] Guha, S., Ed., Biswas, K., Ford, B., Sivakumar, S., and P. 2017 Srisuresh, "NAT Behavioral Requirements for TCP", BCP 142, 2018 RFC 5382, DOI 10.17487/RFC5382, October 2008, 2019 . 2021 [RFC6891] Damas, J., Graff, M., and P. Vixie, "Extension Mechanisms 2022 for DNS (EDNS(0))", STD 75, RFC 6891, 2023 DOI 10.17487/RFC6891, April 2013, 2024 . 2026 [RFC7766] Dickinson, J., Dickinson, S., Bellis, R., Mankin, A., and 2027 D. Wessels, "DNS Transport over TCP - Implementation 2028 Requirements", RFC 7766, DOI 10.17487/RFC7766, March 2016, 2029 . 2031 [RFC7828] Wouters, P., Abley, J., Dickinson, S., and R. Bellis, "The 2032 edns-tcp-keepalive EDNS0 Option", RFC 7828, 2033 DOI 10.17487/RFC7828, April 2016, 2034 . 2036 [RFC7830] Mayrhofer, A., "The EDNS(0) Padding Option", RFC 7830, 2037 DOI 10.17487/RFC7830, May 2016, 2038 . 2040 [RFC8126] Cotton, M., Leiba, B., and T. Narten, "Guidelines for 2041 Writing an IANA Considerations Section in RFCs", BCP 26, 2042 RFC 8126, DOI 10.17487/RFC8126, June 2017, 2043 . 2045 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2046 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 2047 May 2017, . 2049 11.2. Informative References 2051 [I-D.ietf-dnssd-push] 2052 Pusateri, T. and S. Cheshire, "DNS Push Notifications", 2053 draft-ietf-dnssd-push-13 (work in progress), October 2017. 2055 [I-D.ietf-dprive-padding-policy] 2056 Mayrhofer, A., "Padding Policy for EDNS(0)", draft-ietf- 2057 dprive-padding-policy-04 (work in progress), February 2058 2018. 2060 [I-D.ietf-tls-tls13] 2061 Rescorla, E., "The Transport Layer Security (TLS) Protocol 2062 Version 1.3", draft-ietf-tls-tls13-26 (work in progress), 2063 March 2018. 2065 [I-D.sctl-dnssd-mdns-relay] 2066 Cheshire, S. and T. Lemon, "Multicast DNS Discovery 2067 Relay", draft-sctl-dnssd-mdns-relay-03 (work in progress), 2068 March 2018. 2070 [NagleDA] Cheshire, S., "TCP Performance problems caused by 2071 interaction between Nagle's Algorithm and Delayed ACK", 2072 May 2005, 2073 . 2075 [RFC0768] Postel, J., "User Datagram Protocol", STD 6, RFC 768, 2076 DOI 10.17487/RFC0768, August 1980, 2077 . 2079 [RFC1122] Braden, R., Ed., "Requirements for Internet Hosts - 2080 Communication Layers", STD 3, RFC 1122, 2081 DOI 10.17487/RFC1122, October 1989, 2082 . 2084 [RFC2782] Gulbrandsen, A., Vixie, P., and L. Esibov, "A DNS RR for 2085 specifying the location of services (DNS SRV)", RFC 2782, 2086 DOI 10.17487/RFC2782, February 2000, 2087 . 2089 [RFC6335] Cotton, M., Eggert, L., Touch, J., Westerlund, M., and S. 2090 Cheshire, "Internet Assigned Numbers Authority (IANA) 2091 Procedures for the Management of the Service Name and 2092 Transport Protocol Port Number Registry", BCP 165, 2093 RFC 6335, DOI 10.17487/RFC6335, August 2011, 2094 . 2096 [RFC6763] Cheshire, S. and M. Krochmal, "DNS-Based Service 2097 Discovery", RFC 6763, DOI 10.17487/RFC6763, February 2013, 2098 . 2100 [RFC7413] Cheng, Y., Chu, J., Radhakrishnan, S., and A. Jain, "TCP 2101 Fast Open", RFC 7413, DOI 10.17487/RFC7413, December 2014, 2102 . 2104 [RFC7858] Hu, Z., Zhu, L., Heidemann, J., Mankin, A., Wessels, D., 2105 and P. Hoffman, "Specification for DNS over Transport 2106 Layer Security (TLS)", RFC 7858, DOI 10.17487/RFC7858, May 2107 2016, . 2109 Authors' Addresses 2111 Ray Bellis 2112 Internet Systems Consortium, Inc. 2113 950 Charter Street 2114 Redwood City CA 94063 2115 USA 2117 Phone: +1 650 423 1200 2118 Email: ray@isc.org 2119 Stuart Cheshire 2120 Apple Inc. 2121 1 Infinite Loop 2122 Cupertino CA 95014 2123 USA 2125 Phone: +1 408 974 3207 2126 Email: cheshire@apple.com 2128 John Dickinson 2129 Sinodun Internet Technologies 2130 Magadalen Centre 2131 Oxford Science Park 2132 Oxford OX4 4GA 2133 United Kingdom 2135 Email: jad@sinodun.com 2137 Sara Dickinson 2138 Sinodun Internet Technologies 2139 Magadalen Centre 2140 Oxford Science Park 2141 Oxford OX4 4GA 2142 United Kingdom 2144 Email: sara@sinodun.com 2146 Ted Lemon 2147 Barefoot Consulting 2148 Brattleboro 2149 VT 05301 2150 USA 2152 Email: mellon@fugue.com 2154 Tom Pusateri 2155 Unaffiliated 2156 Raleigh NC 27608 2157 USA 2159 Phone: +1 919 867 1330 2160 Email: pusateri@bangj.com