idnits 2.17.1 draft-williams-exp-tcp-host-id-opt-05.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (April 8, 2015) is 3306 days in the past. Is this intentional? Checking references for intended status: Experimental ---------------------------------------------------------------------------- ** Obsolete normative reference: RFC 793 (Obsoleted by RFC 9293) -- Obsolete informational reference (is this intentional?): RFC 6824 (Obsoleted by RFC 8684) Summary: 1 error (**), 0 flaws (~~), 1 warning (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group B. Williams 3 Internet-Draft Akamai, Inc. 4 Intended status: Experimental M. Boucadair 5 Expires: October 10, 2015 France Telecom 6 D. Wing 7 Cisco Systems, Inc. 8 April 8, 2015 10 Experimental Option for TCP Host Identification 11 draft-williams-exp-tcp-host-id-opt-05 13 Abstract 15 Recent IETF proposals have identified benefits to more distinctly 16 identifying the hosts that are hidden behind a shared address/prefix 17 sharing device or application-layer proxy. Analysis indicates that 18 the use of a TCP option for this purpose can be successfully applied 19 to a broad range of use cases. This document describes a common 20 experimental TCP option format for host identification. 22 Status of this Memo 24 This Internet-Draft is submitted in full conformance with the 25 provisions of BCP 78 and BCP 79. 27 Internet-Drafts are working documents of the Internet Engineering 28 Task Force (IETF). Note that other groups may also distribute 29 working documents as Internet-Drafts. The list of current Internet- 30 Drafts is at http://datatracker.ietf.org/drafts/current/. 32 Internet-Drafts are draft documents valid for a maximum of six months 33 and may be updated, replaced, or obsoleted by other documents at any 34 time. It is inappropriate to use Internet-Drafts as reference 35 material or to cite them other than as "work in progress." 37 This Internet-Draft will expire on October 10, 2015. 39 Copyright Notice 41 Copyright (c) 2015 IETF Trust and the persons identified as the 42 document authors. All rights reserved. 44 This document is subject to BCP 78 and the IETF Trust's Legal 45 Provisions Relating to IETF Documents 46 (http://trustee.ietf.org/license-info) in effect on the date of 47 publication of this document. Please review these documents 48 carefully, as they describe your rights and restrictions with respect 49 to this document. Code Components extracted from this document must 50 include Simplified BSD License text as described in Section 4.e of 51 the Trust Legal Provisions and are provided without warranty as 52 described in the Simplified BSD License. 54 Table of Contents 56 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 57 1.1. Important Use Cases . . . . . . . . . . . . . . . . . . . 3 58 1.2. Experiment Goals . . . . . . . . . . . . . . . . . . . . . 5 59 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 5 60 3. Option Format . . . . . . . . . . . . . . . . . . . . . . . . 6 61 4. Option Use . . . . . . . . . . . . . . . . . . . . . . . . . . 6 62 4.1. Option Values . . . . . . . . . . . . . . . . . . . . . . 6 63 4.2. Sending Host Requirements . . . . . . . . . . . . . . . . 7 64 4.2.1. Alternative SYN Cookie Support . . . . . . . . . . . . 8 65 4.2.2. Persistent TCP Connections . . . . . . . . . . . . . . 8 66 4.2.3. Packet Fragmentation . . . . . . . . . . . . . . . . . 8 67 4.3. Multiple In-Path HOST_ID Senders . . . . . . . . . . . . . 9 68 4.4. Option Interpretation . . . . . . . . . . . . . . . . . . 10 69 5. Interaction with Other TCP Options . . . . . . . . . . . . . . 11 70 5.1. Multipath TCP (MPTCP) . . . . . . . . . . . . . . . . . . 11 71 5.2. Authentication Option (TCP-AO) . . . . . . . . . . . . . . 11 72 5.3. TCP Fast Open (TFO) . . . . . . . . . . . . . . . . . . . 11 73 6. Security Considerations . . . . . . . . . . . . . . . . . . . 12 74 7. Privacy Considerations . . . . . . . . . . . . . . . . . . . . 13 75 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 14 76 9. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 14 77 10. References . . . . . . . . . . . . . . . . . . . . . . . . . . 14 78 10.1. Normative References . . . . . . . . . . . . . . . . . . . 14 79 10.2. Informative References . . . . . . . . . . . . . . . . . . 14 80 Appendix A. Change History . . . . . . . . . . . . . . . . . . . 16 81 A.1. Changes from version 04 to 05 . . . . . . . . . . . . . . 16 82 A.2. Changes from version 03 to 04 . . . . . . . . . . . . . . 17 83 A.3. Changes from version 02 to 03 . . . . . . . . . . . . . . 17 84 A.4. Changes from version 01 to 02 . . . . . . . . . . . . . . 18 85 A.5. Changes from version 00 to 01 . . . . . . . . . . . . . . 18 86 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 18 88 1. Introduction 90 A broad range of issues associated with address sharing have been 91 well documented in [RFC6269] and 92 [I-D.boucadair-intarea-host-identifier-scenarios]. In addition, 93 [RFC6967] provides analysis of various solutions to the problem of 94 revealing the sending host's identifier (HOST_ID) information to the 95 receiver, indicating that a solution using a TCP [RFC0793] option for 96 this purpose is among the possible approaches that could be applied 97 with limited performance impact and a high success ratio. The 98 purpose of this document is to define such a TCP option in order to 99 facilitate further validation of the mechanism. 101 Multiple recent Internet Drafts define TCP options for the purpose of 102 host identification: [I-D.wing-nat-reveal-option], 103 [I-D.abdo-hostid-tcpopt-implementation], and 104 [I-D.williams-overlaypath-ip-tcp-rfc]. Specification of multiple 105 option formats to serve the purpose of host identification increases 106 the burden for potential implementers and presents interoperability 107 challenges as well. This document defines a common TCP option format 108 that supersedes all three of the above proposals. 110 The option defined in this document uses the TCP experimental option 111 codepoint sharing mechanism defined in [RFC6994] and is intended to 112 allow broad deployment of the mechanism on the public Internet in 113 order to validate the utility of this option format for the intended 114 use cases. 116 Section 5 of this document discusses compatibility between this new 117 TCP option and existing commonly deployed TCP options. 119 1.1. Important Use Cases 121 This memo focuses primarily on the following address-sharing 122 scenarios: 124 Carrier Grade NAT (CGN): As defined in [RFC6888], [RFC6333], and 125 other sources, a CGN allows multiple hosts connected to the public 126 Internet to share a single Internet routable IPv4 address. One 127 important characteristic of the CGN use case is that it modifies 128 IP packets in-path, but does not serve as the end point for the 129 associated TCP connections. 131 Application Proxy: As defined in [RFC1919], an application proxy 132 splits a TCP connection into two segments, serving as an endpoint 133 for each of the connections and relaying data flows between the 134 connections. 136 Overlay Network: An overlay network is an Internet based system 137 providing security, optimization, or other services for data flows 138 that transit the system. A network-layer overlay will sometimes 139 act much like a CGN, in that packets transit the system with NAT 140 being applied at the edge of the overlay. A transport-layer or 141 application-layer overlay [RFC3135] will typically act much like 142 an application proxy, in that the TCP connection will be segmented 143 with the overlay network serving as an endpoint for each of the 144 TCP connections. 146 With this set of scenarios, the TCP option could either be applied to 147 an individual TCP packet at the connection endpoint (e.g. an 148 application proxy or a transport layer overlay network) or at an 149 address-sharing middle box (e.g. a CGN or a network layer overlay 150 network). See Section 4 below for additional details about the types 151 of devices that could add the option to a TCP packet, as well as 152 limitations on use of the option when it is to be inserted by an 153 address-sharing middlebox, including issues related to packet 154 fragmentation. 156 The receiver-side use cases considered by this memo include the 157 following: 159 o Differentiating between attack and non-attack traffic when the 160 source of the attack is sharing an address with non-attack 161 traffic. 163 o Application of per-subscriber policies for resource utilization, 164 etc. when multiple subscribers are sharing a common address. 166 o Improving server-side load-balancing decisions by allowing the 167 load for multiple clients behind a shared address to be assigned 168 to different servers, even when session-affinity is required at 169 the application layer. 171 In all of the above cases, differentiation between address-sharing 172 clients commonly needs to be performed by a network function that 173 does not process the application layer protocol (e.g. HTTP) or the 174 security protocol (e.g. TLS), because the action needs to be 175 performed prior to decryption or parsing the application layer. Due 176 to this, a solution implemented within the application layer or 177 security protocol cannot fully meet the receiver-side requirements. 178 At the same time, as noted in [RFC6967], use of an IP option for this 179 purpose has a low success rate. For these reasons, using a TCP 180 option to deliver the host identifier has been selected as an 181 effective way to satisfy these specific use cases. 183 1.2. Experiment Goals 185 The testing effort documented in 186 [I-D.abdo-hostid-tcpopt-implementation] confirmed that a TCP option 187 could be used for host identification purposes without significant 188 disruption of TCP connectivity to legacy servers and networks that do 189 no support the option. It also showed how mechanisms available in 190 existing TCP implementations could make use of such a TCP option for 191 improved diagnostics and/or packet filtering. 193 Specification of the TCP option described in this memo will allow 194 further experiments to be conducted in order to assess the viability 195 of the option for the receiver-side use cases discussed above: 197 o Differentiate between attack and non-attack traffic. 199 o Enforce per-client policies. 201 o Assist load-balancing decision-making. 203 In particular, real-world deployment of the option is expected to 204 provide opportunities for engagement with a broader range of both 205 application and middleware implementations in order to develop a more 206 complete picture of how well the option meets the use-case 207 requirements. 209 Continued experimentation on the public Internet following 210 publication of this memo is expected to allow further refinement of 211 requirements related to the values used to populate the option and 212 how those values can be interpreted by the receiver. There is a 213 tradeoff between providing the expected functionality to the receiver 214 and protecting the privacy of the sender, and additional work is 215 necessary in order to find the right balance. See Section 7 for 216 additional discussion. 218 Continued experimentation on the public Internet is also expected to 219 support improved guidance on TCP option interoperability, especially 220 in the context of Multipath TCP [RFC6824] and TCP Fast Open 221 [RFC7413]. See Section 5 for additional discussion. 223 2. Terminology 225 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 226 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 227 document are to be interpreted as described in [RFC2119]. 229 3. Option Format 231 When used for host identification, the TCP experimental option uses 232 the experiment identification mechanism described in [RFC6994] and 233 has the following format and content. 235 0 1 2 3 236 01234567 89012345 67890123 45678901 237 +--------+--------+--------+--------+ 238 | Kind | Length | ExID | 239 +--------+--------+--------+--------+ 240 | Host ID ... 241 +--------+--- 243 Kind: The option kind value is 253 245 Length: The length of the option is variable, based on the required 246 size of the host identifier (e.g. a 2 octet host ID will require a 247 length of 6, while a 4 octet host ID will require a length of 8). 249 ExID: The experiment ID value is 0x0348 (840). 251 Host ID: The host identifier is a value that can be used to 252 differentiate among the various hosts sharing a common public IP 253 address. See below for further discussion of this value. 255 4. Option Use 257 This section describes requirements associated with the use of the 258 option, including: expected option values, which hosts are allowed 259 to include the option, and segments that include the option. 261 4.1. Option Values 263 The information conveyed in the HOST_ID option is intended to 264 uniquely identify the sending host to the best capability of the 265 machine that adds the option to the segment, while at the same time 266 avoiding inclusion of information that does not assist this purpose. 267 In addition, the option is not intended to be used to expose 268 information about the sending host that could not be discovered by 269 observing segments in transit on some portion of the Internet path 270 between the sender and the receiver. As noted in Section 1.2, 271 identifying the optimal set of values to use for this purpose is one 272 of the experimental goals for this document. For this reason, the 273 document attempts to provide a high degree of flexibility for the 274 machine that adds the option to TCP segments. 276 The HOST_ID option value MUST correlate to IP addresses and/or TCP 277 port numbers that were changed by the inserting host/device (i.e., 278 some of the IP address and/or port number bits are used to generate 279 the HOST_ID). Example values that satisfy this requirement include 280 the following: 282 Unique ID: An inserting host/device could maintain a pool of locally 283 unique ID values that are dynamically mapped to the unique source 284 IP address values in use behind the host/device as a result of 285 address sharing. This ID value would be meaningful only within 286 the context of a specific shared IP address due to the local 287 uniqueness characteristic. Such an ID value could be smaller than 288 an IP address (e.g. 16-bits) in order to conserve TCP option 289 space. 291 IP Address/Subnet: An inserting host/device could simply populate 292 the option value with the IP address value in use behind the host/ 293 device. In the case of IPv6 addresses, it could be difficult to 294 include the full address due to TCP option space constraints, so 295 the value would likely need to provide only a portion of the 296 address (e.g. the first 64 bits). 298 IP Address and TCP Port: Some networks share public IP addresses 299 among multiple subscribers with a portion of the TCP port number 300 space being assigned to each subscriber [RFC6346]. When such a 301 system is behind an address sharing host/device, inclusion of both 302 the IP address and the TCP port number will more uniquely identify 303 the sending host than just the IP address on its own. 305 When multiple host identifiers are necessary (e.g. an IP address and 306 a port number), the HOST_ID option is included multiple times within 307 the packet, once for each identifier. While this approach 308 significantly increases option space utilization when multiple 309 identifiers are included, cases where only a single identifier is 310 included are expected to be more common and thus it is beneficial to 311 optimize for those cases. 313 See Section 7 below for discussion of privacy considerations related 314 to selection of HOST_ID values. 316 4.2. Sending Host Requirements 318 The HOST_ID option MUST only be added by the sending host or any 319 device involved in the forwarding path that changes IP addresses 320 and/or TCP port numbers (e.g., NAT44 [RFC3022], Layer-2 Aware NAT, 321 DS-Lite AFTR [RFC6333], NPTv6 [RFC6296], NAT64 [RFC6146], Dual-Stack 322 Extra Lite [RFC6619], TCP Proxy, etc.). The HOST_ID option MUST NOT 323 be added or modified en-route by any device that does not modify IP 324 addresses and/or TCP port numbers. 326 The sending host or intermediary device cannot determine whether the 327 option value is used in a stateful manner by the receiver, nor can it 328 determine whether SYN cookies are in use by the receiver. For this 329 reason, the option MUST be included in all segments, both SYN and 330 non-SYN segments, until return segments from the receiver positively 331 indicate that the TCP connection is fully established on the receiver 332 (e.g. the return segment either includes or acknowledges data). 334 4.2.1. Alternative SYN Cookie Support 336 The authors have also considered an alternative approach to SYN 337 cookie support in which the receiving host (i.e. the host that 338 accepts the TCP connection) to echo the option back to the sender in 339 the SYN/ACK segment when a SYN cookie is being sent. This would 340 allow the host sending HOST_ID to determine whether further inclusion 341 of the option is necessary. This approach would have the benefit of 342 not requiring inclusion of the option in non-SYN segments if SYN 343 cookies had not been used. Unfortunately, this approach fails if the 344 responding host itself does not support the option, since an 345 intermediate node would have no way to determine that SYN cookies had 346 been used. 348 4.2.2. Persistent TCP Connections 350 Some types of middleboxes (e.g. application proxy) open and maintain 351 persistent TCP connections to regularly visited destinations in order 352 to minimize connection establishment burden. Such middleboxes might 353 use a single persistent TCP connection for multiple different client 354 hosts over the life of the persistent connection. 356 This specification does not attempt to support the use of persistent 357 TCP connections for multiple client hosts due to the perceived 358 complexity of providing such support. Instead, the HOST_ID option is 359 only allowed to be used at connection initiation. An inserting host/ 360 device that supports both the HOST_ID option and multi-client 361 persistent TCP connections MUST NOT apply the HOST_ID option to TCP 362 connections that could be used for multiple clients over the life of 363 the connection. If the HOST_ID option was sent during connection 364 initiation, the inserting host/device MUST NOT reuse the connection 365 for data flows originating from a client that would require a 366 different HOST_ID value. 368 4.2.3. Packet Fragmentation 370 In order to avoid the overhead associated with in-path IP 371 fragmentation, it is desirable for the inserting host/device to avoid 372 including the HOST_ID option when IP fragmentation might be required. 373 This is not a firm requirement, though, because the HOST_ID option is 374 only included in the first few packets of a TCP connection and thus 375 associated IP fragmentation will have minimal impact. The option 376 SHOULD NOT be included in packets if the resulting packet would 377 require local fragmentation. 379 It can be difficult to determine whether local fragmentation would be 380 required. For example, in cases where multiple interfaces with 381 different MTUs are in use, a local routing decision has to be made 382 before the MTU can be determined and in some systems this decision 383 could be made after TCP option handling is complete. Additionally, 384 it could be true that inclusion of the option causes the packet to 385 violate the path's MTU but that the path's MTU has not been learned 386 yet on the sending host/device. 388 Due to the difficulty of avoiding IP fragmentation entirely, an 389 important experimental goal for this document is to evaluate the 390 impact of IP fragmentation that results from use of the option. 392 4.3. Multiple In-Path HOST_ID Senders 394 The possibility exists that there could be multiple in-path hosts/ 395 devices configured to insert the HOST_ID option. For example, the 396 client's TCP packets might first traverse a CGN device on their way 397 to the edge of a public Internet overlay network. In order for the 398 HOST_ID value to most uniquely identify the sender, it needs to 399 represent both the identity observed by the CGN device (the 400 subscriber's internal IP address, e.g. [RFC6598]) and the identity 401 observed by the overlay network (the shared address of the CGN 402 device). The mechanism for handling the received HOST_ID value could 403 vary depending upon the nature of the new HOST_ID value to be 404 inserted, as described below. 406 An inserting host/device that uses the received packet's source IP 407 address as the HOST_ID value (possibly along with the port) MUST 408 propagate forward the HOST_ID value(s) from the received packet, 409 since the source IP address and port only represent the previous in- 410 path address sharing device and do not represent the original sender. 411 In the CGN-plus-overlay example, this means that the overlay will 412 include both the CGN's HOST_ID value(s) and a HOST_ID with the source 413 IP address received by the overlay. 415 An inserting host/device that sends a unique ID (as described in 416 Section 4.1) has two options for how to handle the HOST_ID value(s) 417 from the received packet. 419 1. A host/device that sends a unique ID MAY strip the received 420 HOST_ID option and insert its own option, provided that it uses 421 the received HOST_ID value as a differentiator for selecting the 422 unique ID. What this means in the CGN-plus-overlay example above 423 is that the overlay is allowed to drop the HOST_ID value inserted 424 by the CGN provided that the HOST_ID value selected by the 425 overlay represents both the CGN itself and the HOST_ID value 426 inserted by the CGN. 428 2. A host/device that sends a unique ID MAY instead select a unique 429 ID that represents only the previous in-path address-sharing 430 host/device and propagate forward the HOST_ID value inserted by 431 the previous host/device. In the CGN-plus-overlay example, this 432 means that the overlay would include both the CGN's HOST_ID value 433 and a HOST_ID with a unique ID of its own that was selected to 434 represent the CGN's shared address. 436 An inserting host/device that sends a unique ID MUST use one of the 437 above two mechanisms. 439 4.4. Option Interpretation 441 Due to the variable nature of the option value, it is not possible 442 for the receiving machine to reliably determine the value type from 443 the option itself. For this reason, a receiving host/device SHOULD 444 interpret the option value as an opaque identifier. 446 This specification allows the inserting host/device to provide 447 multiple HOST_ID options. The order of appearance of TCP options 448 could be modified by some middleboxes, so deployments SHOULD NOT rely 449 on option order to provide additional meaning to the individual 450 options. Instead, when multiple HOST_ID options are present, their 451 values SHOULD be concatenated together in the order in which they 452 appear in the packet and treated as a single large identifier. 454 For both of the receiver requirements discussed above, this 455 specification uses SHOULD rather than MUST because reliable 456 interpretation and ordering of options could be possible if the 457 inserting host and the interpreting host are under common 458 administrative control and integrity protect communication between 459 the inserting host and the interpreting host. Mechanisms for 460 signaling the value type(s) and integrity protection are not provided 461 by this specification, and in their absence the receiving host/device 462 MUST interpret the option value(s) as a single opaque identifier. 464 5. Interaction with Other TCP Options 466 This section details how the HOST_ID option functions in conjunction 467 with other TCP options. 469 5.1. Multipath TCP (MPTCP) 471 TCP provides for a maximum of 40 octets for TCP options. As 472 discussed in Appendix A of MPTCP [RFC6824], a typical SYN from 473 modern, popular operating systems contains several TCP options (MSS, 474 window scale, SACK permitted, and timestamp) which consume 19-24 475 octets depending on word alignment of the options. The initial SYN 476 from a multipath TCP client would consume an additional 16 octets. 478 HOST_ID needs at least 6 octets to be useful, so 9-21 octets are 479 sufficient for many scenarios that benefit from HOST_ID. However, 4 480 octets are not enough space for the HOST_ID option. Thus, a TCP SYN 481 containing all the typical TCP options (MSS, window Scale, SACK 482 permitted, timestamp), and also containing multipath capable or 483 multipath join, and also being word aligned, has insufficient space 484 to accommodate HOST_ID. This means something has to give. The 485 choices are either to avoid word alignment in that case (freeing 5 486 octets) or avoid adding the HOST_ID option. Although option packing 487 seems like the best approach, we expect to learn from deployment 488 experience during the experiment which of these options is most 489 viable in practice. 491 5.2. Authentication Option (TCP-AO) 493 The TCP-AO option [RFC5925] is incompatible with address sharing due 494 to the fact that it provides integrity protection of the source IP 495 address. For this reason, the only use cases where it makes sense to 496 combine TCP-AO and HOST_ID are those where the TCP-AO-NAT extension 497 [RFC6978] is in use. Injecting a HOST_ID TCP option does not 498 interfere with the use of TCP-AO-NAT because the TCP options are not 499 included in the MAC calculation. 501 5.3. TCP Fast Open (TFO) 503 The TFO option [RFC7413] uses a zero length cookie (total option 504 length 2 bytes) to request a TFO cookie for use on future 505 connections. The server-generated TFO cookie is required to be at 506 least 4 bytes long and allowed to be as long as 16 bytes (total 507 option length 6 to 18 bytes). The cookie request form of the option 508 leaves enough room available in a SYN packet with the most commonly 509 used options to accommodate the HOST_ID option, but a valid TFO 510 cookie length of any longer than 13 bytes would prevent even the 511 minimal 6 byte HOST_ID option from being included in the header. 513 There are multiple possibilities for allowing TFO and HOST_ID to be 514 supported for the same connection, including: 516 o If the TFO implementation allows the cookie size to be 517 configurable, the configured cookie size can be specifically 518 selected to leave enough option space available in a typical TFO 519 SYN packet to allow inclusion of the HOST_ID option. 521 o If the TFO implementation provides explicit support for the 522 HOST_ID option, it can be designed to use a shorter cookie length 523 when the HOST_ID option is present in the TFO cookie request SYN. 525 We expect to learn from deployment experience during the experiment 526 whether one of these options is workable, or whether the two 527 mechanisms (TFO and HOST_ID) will be deemed mutually exclusive. In 528 particular, reducing the TFO cookie size in order to include the 529 HOST_ID option could have unacceptable security implications. 531 It should also be noted that the presence of data in a TFO SYN 532 increases the likelihood that there will be no space available in the 533 SYN packet to support inclusion of the HOST_ID option without IP 534 fragmentation, even if there is enough room in the TCP option space. 535 This issue could also lead to the conclusion that TFO and HOST_ID are 536 mutually exclusive. 538 6. Security Considerations 540 Security (including privacy) considerations common to all HOST_ID 541 solutions are discussed in [RFC6967]. 543 The content of the HOST_ID option SHOULD NOT be used for purposes 544 that require a trust relationship between the sender and the receiver 545 (e.g. billing and/or subscriber policy enforcement). This 546 requirement uses SHOULD rather than MUST because reliable 547 interpretation of options could be possible if the inserting host and 548 the interpreting host are under common administrative control and 549 integrity protect communication between the inserting host and the 550 interpreting host. Mechanisms for signaling the value type(s) and 551 integrity protection are not provided by this specification, and in 552 their absence the receiving host/device MUST NOT use the HOST_ID 553 value for purposes that require a trust relationship. 555 Note that the above trust requirement applies equally to HOST_ID 556 option values propagated forward from a previous in-path host as 557 described in Section 4.3. In other words, if the trust mechanism 558 does not apply to all option values in the packet, then none of the 559 HOST_ID values can be considered trusted and the receiving host/ 560 device MUST NOT use any of the HOST_ID values for purposes that 561 require a trust relationship. An inserting host/device that has such 562 a trust relationship MUST NOT propagate forward an untrusted HOST_ID 563 in such a way as to allow it to be considered trusted. 565 When the receiving network uses the values provided by the option in 566 a way that does not require trust (e.g. maintaining session affinity 567 in a load-balancing system), then use of a mechanism to enforce the 568 trust relationship is OPTIONAL. 570 7. Privacy Considerations 572 Sending a TCP SYN across the public Internet necessarily discloses 573 the public IP address of the sending host. When an intermediate 574 address sharing device is deployed on the public Internet, anonymity 575 of the hosts using the device will be increased, with hosts 576 represented by multiple source IP addresses on the ingress side of 577 the device using a single source IP address on the egress side. The 578 HOST_ID TCP option removes that increased anonymity, taking 579 information that was already visible in TCP packets on the public 580 Internet on the ingress side of the address sharing device and making 581 it available on the egress side of the device as well. In some 582 cases, an explicit purpose of the address sharing device is 583 anonymity, in which case use of the HOST_ID TCP option would be 584 incompatible with the purpose of the device. 586 A NAT device used to provide interoperability between a local area 587 network (LAN) using private [RFC1918] IP addresses and the public 588 Internet is sometimes specifically intended to provide anonymity for 589 the LAN clients as described in the above paragraph. For this 590 reason, address sharing devices at the border between a private LAN 591 and the public Internet MUST NOT insert the HOST_ID option. 593 The HOST_ID option MUST NOT be used to provide client geographic or 594 network location information that was not publicly visible in IP 595 packets for the TCP flows processed by the inserting host. For 596 example, the client's IP address MAY be used as the HOST_ID option 597 value, but any geographic or network location information derived 598 from the client's IP address MUST NOT be used as the HOST_ID value. 600 The HOST_ID option MAY provide differentiating information that is 601 locally unique such that individual TCP flows processed by the 602 inserting host can be reliably identified. The HOST_ID option MUST 603 NOT provide client identification information that was not publicly 604 visible in IP packets for the TCP flows processed by the inserting 605 host, such as subscriber information linked to the IP address.. 607 The HOST_ID option MUST be stripped from IP packets traversing middle 608 boxes that provide network-based anonymity services. 610 8. IANA Considerations 612 This document specifies a new TCP option that uses the shared 613 experimental options format [RFC6994], with ExID=0x0348 (840) in 614 network-standard byte order. This ExID has already been registered 615 with IANA. 617 9. Acknowledgements 619 Many thanks to W. Eddy, Y. Nishida, T. Reddy, M. Scharf, J. Touch, 620 and A. Zimmermann for their comments. 622 10. References 624 10.1. Normative References 626 [RFC0793] Postel, J., "Transmission Control Protocol", STD 7, 627 RFC 793, September 1981. 629 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 630 Requirement Levels", BCP 14, RFC 2119, March 1997. 632 [RFC6994] Touch, J., "Shared Use of Experimental TCP Options", 633 RFC 6994, August 2013. 635 10.2. Informative References 637 [I-D.abdo-hostid-tcpopt-implementation] 638 Abdo, E., Boucadair, M., and J. Queiroz, "HOST_ID TCP 639 Options: Implementation & Preliminary Test Results", 640 draft-abdo-hostid-tcpopt-implementation-03 (work in 641 progress), July 2012. 643 [I-D.boucadair-intarea-host-identifier-scenarios] 644 Boucadair, M., Binet, D., Durel, S., Chatras, B., Reddy, 645 T., Williams, B., Sarikaya, B., Xue, L., and R. Wheeldon, 646 "Scenarios with Host Identification Complications", 647 draft-boucadair-intarea-host-identifier-scenarios-11 (work 648 in progress), April 2015. 650 [I-D.williams-overlaypath-ip-tcp-rfc] 651 Williams, B., "Overlay Path Option for IP and TCP", 652 draft-williams-overlaypath-ip-tcp-rfc-04 (work in 653 progress), June 2013. 655 [I-D.wing-nat-reveal-option] 656 Yourtchenko, A. and D. Wing, "Revealing hosts sharing an 657 IP address using TCP option", 658 draft-wing-nat-reveal-option-03 (work in progress), 659 December 2011. 661 [RFC1918] Rekhter, Y., Moskowitz, R., Karrenberg, D., Groot, G., and 662 E. Lear, "Address Allocation for Private Internets", 663 BCP 5, RFC 1918, February 1996. 665 [RFC1919] Chatel, M., "Classical versus Transparent IP Proxies", 666 RFC 1919, March 1996. 668 [RFC3022] Srisuresh, P. and K. Egevang, "Traditional IP Network 669 Address Translator (Traditional NAT)", RFC 3022, 670 January 2001. 672 [RFC3135] Border, J., Kojo, M., Griner, J., Montenegro, G., and Z. 673 Shelby, "Performance Enhancing Proxies Intended to 674 Mitigate Link-Related Degradations", RFC 3135, June 2001. 676 [RFC5925] Touch, J., Mankin, A., and R. Bonica, "The TCP 677 Authentication Option", RFC 5925, June 2010. 679 [RFC6146] Bagnulo, M., Matthews, P., and I. van Beijnum, "Stateful 680 NAT64: Network Address and Protocol Translation from IPv6 681 Clients to IPv4 Servers", RFC 6146, April 2011. 683 [RFC6269] Ford, M., Boucadair, M., Durand, A., Levis, P., and P. 684 Roberts, "Issues with IP Address Sharing", RFC 6269, 685 June 2011. 687 [RFC6296] Wasserman, M. and F. Baker, "IPv6-to-IPv6 Network Prefix 688 Translation", RFC 6296, June 2011. 690 [RFC6333] Durand, A., Droms, R., Woodyatt, J., and Y. Lee, "Dual- 691 Stack Lite Broadband Deployments Following IPv4 692 Exhaustion", RFC 6333, August 2011. 694 [RFC6346] Bush, R., "The Address plus Port (A+P) Approach to the 695 IPv4 Address Shortage", RFC 6346, August 2011. 697 [RFC6598] Weil, J., Kuarsingh, V., Donley, C., Liljenstolpe, C., and 698 M. Azinger, "IANA-Reserved IPv4 Prefix for Shared Address 699 Space", BCP 153, RFC 6598, April 2012. 701 [RFC6619] Arkko, J., Eggert, L., and M. Townsley, "Scalable 702 Operation of Address Translators with Per-Interface 703 Bindings", RFC 6619, June 2012. 705 [RFC6824] Ford, A., Raiciu, C., Handley, M., and O. Bonaventure, 706 "TCP Extensions for Multipath Operation with Multiple 707 Addresses", RFC 6824, January 2013. 709 [RFC6888] Perreault, S., Yamagata, I., Miyakawa, S., Nakagawa, A., 710 and H. Ashida, "Common Requirements for Carrier-Grade NATs 711 (CGNs)", BCP 127, RFC 6888, April 2013. 713 [RFC6967] Boucadair, M., Touch, J., Levis, P., and R. Penno, 714 "Analysis of Potential Solutions for Revealing a Host 715 Identifier (HOST_ID) in Shared Address Deployments", 716 RFC 6967, June 2013. 718 [RFC6978] Touch, J., "A TCP Authentication Option Extension for NAT 719 Traversal", RFC 6978, July 2013. 721 [RFC7413] Cheng, Y., Chu, J., Radhakrishnan, S., and A. Jain, "TCP 722 Fast Open", RFC 7413, December 2014. 724 Appendix A. Change History 726 [Note to RFC Editor: Please remove this section prior to 727 publication.] 729 A.1. Changes from version 04 to 05 731 Make this draft self-contained, rather than referring readers to use- 732 cases and requirements contained in other I.D.s that were never 733 published as RFCs. 735 Add discussion of TCP Fast Open. 737 Correct some discussion of TCP-AO and TCP-AO-NAT. 739 Clarify exactly what the identifier is identifying. 741 Improve discussion on interpretation of multiple instances of the 742 option, including order of interpretation and set interpretation. 744 Evaluated whether use of multiple identifiers should be constrained. 745 This is unclear, and so left for the experiment to determine. 747 Discuss the possibility of the option value changing over the life of 748 the connection (spec now prohibits this). 750 Clarify use cases related to stripping and replacing the option. 752 Add discussion of non-local fragmentation. 754 Evaluate the reliability of attempts to exclude the option when local 755 fragmentation would be required. 757 Clarify the security requirements re: trust relationship. 758 Specifically calls out that common admin control and authentication 759 can allow additional uses. 761 Clarify privacy considerations regarding NATs that separate private 762 and public networks. 764 Remove restatement of requirements from other documents. 766 Justify use of SHOULD rather than MUST throughout. 768 A.2. Changes from version 03 to 04 770 Improve discussion of RFC6967. 772 Don't use "message" to describe TCP segments. 774 Add reference to RFC6994 to section 3. 776 Clarify that this draft supersedes earlier drafts. 778 Improve discussion of SYN cookie handling. 780 Remove lower case uses of keywords (e.g. must, should, etc.) 781 throughout the document. 783 Some stronger privacy guidance, replacing SHOULD with MUST. 785 Add an experiment goal related to optimal option value. 787 Add text related to the identification goals of the option value 788 (still needs more work). 790 A.3. Changes from version 02 to 03 792 Clarification of arguments in favor of this approach. 794 Add discussion of important use cases. 796 Clarification of experiment goals and earlier test results. 798 A.4. Changes from version 01 to 02 800 Add note re: order of appearance. 802 A.5. Changes from version 00 to 01 804 Add discussion of experiment goals. 806 Limit external references to the earlier drafts. 808 Add guidance to limit the types of device that add the option. 810 Improve/correct discussion of TCP-AO and security. 812 Authors' Addresses 814 Brandon Williams 815 Akamai, Inc. 816 8 Cambridge Center 817 Cambridge, MA 02142 818 USA 820 Email: brandon.williams@akamai.com 822 Mohamed Boucadair 823 France Telecom 824 Rennes, 35000 825 Fance 827 Email: mohamed.boucadair@orange.com 829 Dan Wing 830 Cisco Systems, Inc. 831 170 West Tasman Drive 832 San Jose, CA 95134 833 USA 835 Email: dwing@cisco.com