idnits 2.17.1 draft-ietf-tsvwg-transport-encrypt-04.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 document seems to lack a both a reference to RFC 2119 and the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords. RFC 2119 keyword, line 685: '... IPv6 "source nodes SHOULD assign each...' Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (February 18, 2019) is 1891 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Outdated reference: A later version (-17) exists of draft-ietf-ippm-ioam-data-03 == Outdated reference: A later version (-34) exists of draft-ietf-quic-transport-14 == Outdated reference: A later version (-12) exists of draft-ietf-taps-transport-security-02 == Outdated reference: A later version (-15) exists of draft-ietf-tcpinc-tcpcrypt-12 -- Obsolete informational reference (is this intentional?): RFC 4995 (Obsoleted by RFC 5795) -- Obsolete informational reference (is this intentional?): RFC 6347 (Obsoleted by RFC 9147) -- Obsolete informational reference (is this intentional?): RFC 7525 (Obsoleted by RFC 9325) Summary: 1 error (**), 0 flaws (~~), 5 warnings (==), 4 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 TSVWG G. Fairhurst 3 Internet-Draft University of Aberdeen 4 Intended status: Informational C. Perkins 5 Expires: August 22, 2019 University of Glasgow 6 February 18, 2019 8 The Impact of Transport Header Confidentiality on Network Operation and 9 Evolution of the Internet 10 draft-ietf-tsvwg-transport-encrypt-04 12 Abstract 14 This document describes implications of applying end-to-end 15 encryption at the transport layer. It identifies in-network uses of 16 transport layer header information. It then reviews the implications 17 of developing end-to-end transport protocols that use authentication 18 to protect the integrity of transport information or encryption to 19 provide confidentiality of the transport protocol header and expected 20 implications of transport protocol design and network operation. 21 Since transport measurement and analysis of the impact of network 22 characteristics have been important to the design of current 23 transport protocols, it also considers the impact on transport and 24 application evolution. 26 Status of This Memo 28 This Internet-Draft is submitted in full conformance with the 29 provisions of BCP 78 and BCP 79. 31 Internet-Drafts are working documents of the Internet Engineering 32 Task Force (IETF). Note that other groups may also distribute 33 working documents as Internet-Drafts. The list of current Internet- 34 Drafts is at https://datatracker.ietf.org/drafts/current/. 36 Internet-Drafts are draft documents valid for a maximum of six months 37 and may be updated, replaced, or obsoleted by other documents at any 38 time. It is inappropriate to use Internet-Drafts as reference 39 material or to cite them other than as "work in progress." 41 This Internet-Draft will expire on August 22, 2019. 43 Copyright Notice 45 Copyright (c) 2019 IETF Trust and the persons identified as the 46 document authors. All rights reserved. 48 This document is subject to BCP 78 and the IETF Trust's Legal 49 Provisions Relating to IETF Documents 50 (https://trustee.ietf.org/license-info) in effect on the date of 51 publication of this document. Please review these documents 52 carefully, as they describe your rights and restrictions with respect 53 to this document. Code Components extracted from this document must 54 include Simplified BSD License text as described in Section 4.e of 55 the Trust Legal Provisions and are provided without warranty as 56 described in the Simplified BSD License. 58 Table of Contents 60 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 61 2. Context and Rationale . . . . . . . . . . . . . . . . . . . . 3 62 3. Current uses of Transport Headers within the Network . . . . 10 63 3.1. Observing Transport Information in the Network . . . . . 10 64 3.2. Transport Measurement . . . . . . . . . . . . . . . . . . 16 65 3.3. Use for Network Diagnostics and Troubleshooting . . . . . 20 66 3.4. Header Compression . . . . . . . . . . . . . . . . . . . 21 67 4. Encryption and Authentication of Transport Headers . . . . . 21 68 5. Addition of Transport Information to Network-Layer Protocol 69 Headers . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 70 6. Implications of Protecting the Transport Headers . . . . . . 26 71 6.1. Independent Measurement . . . . . . . . . . . . . . . . . 26 72 6.2. Characterising "Unknown" Network Traffic . . . . . . . . 28 73 6.3. Accountability and Internet Transport Protocols . . . . . 28 74 6.4. Impact on Operational Cost . . . . . . . . . . . . . . . 29 75 6.5. Impact on Research, Development and Deployment . . . . . 30 76 7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . 30 77 8. Security Considerations . . . . . . . . . . . . . . . . . . . 33 78 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 35 79 10. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 35 80 11. Informative References . . . . . . . . . . . . . . . . . . . 35 81 Appendix A. Revision information . . . . . . . . . . . . . . . . 42 82 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 43 84 1. Introduction 86 There is increased interest in, and deployment of, new protocols that 87 employ end-to-end encryption at the transport layer, including the 88 transport layer headers. An example of such a transport is the QUIC 89 transport protocol [I-D.ietf-quic-transport], currently being 90 standardised in the IETF. Encryption of transport layer headers and 91 payload data has many benefits in terms of protecting user privacy. 92 These benefits have been widely discussed [RFC7258], [RFC7624], and 93 this document strongly supports the increased use of encryption in 94 transport protocols. There are also, however, some costs, in that 95 the widespread use of transport encryption requires changes to 96 network operations, and complicates network measurement for research, 97 operational, and standardisation purposes. 99 This document discusses some consequences of applying end-to-end 100 encryption at the transport layer. It reviews the implications of 101 developing end-to-end transport protocols that use encryption to 102 provide confidentiality of the transport protocol header, and 103 considers the effect of such changes on transport protocol design and 104 network operations. It also considers anticipated implications on 105 transport and application evolution. 107 Transports are increasingly encrypting and authenticating the payload 108 (i.e., the application data carried within the transport connection) 109 end-to-end. Such protection is encouraged, and the implications are 110 not further discussed in this memo. 112 2. Context and Rationale 114 The transport layer provides end-to-end interactions between 115 endpoints (processes) using an Internet path. Transport protocols 116 layer directly over the network-layer service and are sent in the 117 payload of network-layer packets. They support end-to-end 118 communication between applications, supported by higher-layer 119 protocols, running on the end systems (or transport endpoints). This 120 simple architectural view hides one of the core functions of the 121 transport, however, to discover and adapt to the properties of the 122 Internet path that is currently being used. The design of Internet 123 transport protocols is as much about trying to avoid the unwanted 124 side effects of congestion on a flow and other capacity-sharing 125 flows, avoiding congestion collapse, adapting to changes in the path 126 characteristics, etc., as it is about end-to-end feature negotiation, 127 flow control and optimising for performance of a specific 128 application. 130 To achieve stable Internet operations the IETF transport community 131 has to date relied heavily on measurement and insights of the network 132 operations community to understand the trade-offs, and to inform 133 selection of appropriate mechanisms, to ensure a safe, reliable, and 134 robust Internet (e.g., [RFC1273]). In turn, the network operations 135 community relies on being able to understand the pattern and 136 requirements of traffic passing over the Internet, both in aggregate 137 and at the flow level. 139 There are many motivations for deploying encrypted transports 140 [RFC7624] (i.e., transport protocols that use encryption to provide 141 confidentiality of some or all of the transport-layer header 142 information), and encryption of transport payloads (i.e. 143 Confidentiality of the payload data). The increasing public concerns 144 about interference with Internet traffic have led to a rapidly 145 expanding deployment of encryption to protect end-user privacy, e.g., 146 QUIC [I-D.ietf-quic-transport]. Encryption is also expected to form 147 a basis of future transport protocol designs. 149 Some network operators and access providers have come to rely on the 150 in-network measurement of transport properties and the functionality 151 provided by middleboxes to both support network operations and 152 enhance performance. There can therefore be implications when 153 working with encrypted transport protocols that hide transport header 154 information from the network. These present architectural challenges 155 and considerations in the way transport protocols are designed, and 156 ability to characterise and compare different transport solutions 157 [Measure]. Implementations of network devices are encouraged to 158 avoid side-effects when protocols are updated. Introducing 159 cryptographic integrity checks to header fields can also prevent 160 undetected manipulation of the field by network devices, or 161 undetected addition of information to a packet. However, this does 162 not prevent inspection of the information by a device on path, and it 163 is possible that such devices could develop mechanisms that rely on 164 the presence of such a field, or a known value in the field. 166 Reliance on the presence and semantics of specific header information 167 leads to ossification. An endpoint could be required to supply a 168 specific header to receive the network service that it desires. In 169 some cases, this could be benign or advantageous to the protocol 170 (e.g., recognising the start of a connection, or explicitly exposing 171 protocol information can be expected to provide more consistent 172 decisions by on-path devices than the use of diverse methods to infer 173 semantics from other flow properties); in other cases this is not 174 beneficial (e.g., a mechanism implemented in a network device, such 175 as a firewall, that required a header field to have only a specific 176 known set of values could prevent the device from forwarding packets 177 using a different version of a protocol that introduces a new feature 178 that changes the value present in this field, preventing evolution of 179 the protocol). Experience developing Transport Layer Security 180 [RFC8446], required a design that recognised that deployed 181 middleboxes relied on the exposed information in TLS 1.2 183 Examples of the impact of ossification on transport protocol design 184 and ease of deployment can be seen in the case of Multipath TCP 185 (MPTCP) and the TCP Fast Open option. The design of MPTCP had to be 186 revised to account for middleboxes, so called "TCP Normalizers", that 187 monitor the evolution of the window advertised in the TCP headers and 188 that reset connections if the window does not grow as expected. 189 Similarly, TCP Fast Open has had issues with middleboxes that remove 190 unknown TCP options, that drop segments with unknown TCP options, 191 that drop segments that contain data and have the SYN bit set, that 192 drop packets with SYN/ACK that acknowledge data, or that disrupt 193 connections that send data before the three-way handshake completes. 194 In both cases, the issue was caused by middleboxes that had a hard- 195 coded understanding of transport behaviour, and that interacted 196 poorly with transports that tried to change that behaviour. Other 197 examples have included middleboxes that rewrite TCP sequence and 198 acknowledgement numbers but are unaware of the (newer) SACK option 199 and don't correctly rewrite selective acknowledgements to match the 200 changes made to the fixed TCP header. 202 A protocol design that uses header encryption can provide 203 confidentiality of some or all of the protocol header information. 204 Encryption with secure key distribution prevents an on-path device 205 from observing the header field. It therefore prevents mechanisms 206 being built that directly rely on the information or seek to infer 207 semantics of an exposed header field. Using encryption to provide 208 confidentiality of the transport layer brings some well-known privacy 209 and security benefits and can therefore help reduce ossification of 210 the transport layer. In particular, it is important that protocols 211 either do not expose information where the usage could change in 212 future protocols, or that methods that utilise the information are 213 robust to potential changes as protocols evolve over time. To avoid 214 unwanted inspection, a protocol could also intentionally vary the 215 format and/or value of header fields (sometimes known as Greasing 216 [I-D.thomson-quic-grease]). However, while encryption hides the 217 protocol header information, it does not prevent ossification of the 218 network service. People seeking understanding of network traffic 219 could come to rely on pattern inferences and other heuristics as the 220 basis for network decision and to derive measurement data, creating 221 new dependencies on the transport protocol. 223 Specification of non-encrypted transport header fields explicitly 224 allows protocol designers to make specific header information 225 observable in the network. This supports other uses of this 226 information by on-path devices, and at the same time this can be 227 expected to lead to ossification of the transport header, because 228 network forwarding could evolve to depend on the presence and/or 229 value of these fields. The decision about which transport headers 230 fields are made observable offers trade-offs around authentication 231 and confidentiality versus observability, network operations and 232 management, and ossification. For example, a design that provides 233 confidentiality of protocol header information can impact the 234 following activities that rely on measurement and analysis of traffic 235 flows: 237 Network Operations and Research: Observable transport headers enable 238 both operators and the research community to explicitly measure 239 and analyse protocol performance, network anomalies, and failure 240 pathologies. 242 This information can help inform capacity planning, and assist in 243 determining the need for equipment and/or configuration changes by 244 network operators. 246 The data can also inform Internet engineering research, and help 247 in the development of new protocols, methodologies, and 248 procedures. Concealing the transport protocol header information 249 makes the stream performance unavailable to passive observers 250 along the path, and likely leads to the development of alternative 251 methods to collect or infer that data (for example heuristics 252 based on analysis of traffic patterns). 254 Providing confidentiality of the transport payload, but leaving 255 some, or all, of the transport headers unencrypted, possibly with 256 authentication, can provide many of the privacy and security 257 benefits while supporting operations and research, but at the cost 258 of ossifying the transport headers. 260 Protection from Denial of Service: Observable transport headers 261 currently provide useful input to classify traffic and detect 262 anomalous events (e.g., changes in application behaviour, 263 distributed denial of service attacks). To be effective, this 264 protection needs to be able to uniquely disambiguate unwanted 265 traffic. An inability to separate this traffic using packet 266 header information could result in less-efficient identification 267 of unwanted traffic or development of different methods (e.g. 268 rate-limiting of uncharacterised traffic). 270 Network Troubleshooting and Diagnostics: Encrypting transport 271 header information eliminates the incentive for operators to 272 troubleshoot since they cannot interpret the data. A flow 273 experiencing packet loss or jitter looks like an unaffected flow 274 when only observing network layer headers (if transport sequence 275 numbers and flow identifiers are obscured). This limits 276 understanding of the impact of packet loss or latency on the 277 flows, or even localizing the network segment causing the packet 278 loss or latency. Encrypted traffic could imply "don't touch" to 279 some, and could limit a trouble-shooting response to "can't help, 280 no trouble found". Additional mechanisms will need to be 281 introduced to help reconstruct or replace transport-level metrics 282 to support troubleshooting and diagnostics, but these add 283 complexity and operational costs (e.g., in deploying additional 284 functions in equipment or adding traffic overhead). 286 Network Traffic Analysis: Hiding transport protocol header 287 information can make it harder to determine which transport 288 protocols and features are being used across a network segment and 289 to measure trends in the pattern of usage. This could impact the 290 ability for an operator to anticipate the need for network 291 upgrades and roll-out. It can also impact the on-going traffic 292 engineering activities performed by operators (such as determining 293 which parts of the path contribute delay, jitter or loss). While 294 the impact could, in many cases, be small there are scenarios 295 where operators directly support particular services (e.g., to 296 troubleshoot issues relating to Quality of Service, QoS; the 297 ability to perform fast re-routing of critical traffic, or support 298 to mitigate the characteristics of specific radio links). The 299 more complex the underlying infrastructure the more important this 300 impact. 302 Open and Verifiable Network Data: Hiding transport protocol header 303 information can reduce the range of actors that can capture useful 304 measurement data. This limits the information sources available 305 to the Internet community to understand the operation of new 306 transport protocols, so preventing access to the information 307 necessary to inform design decisions and standardisation of the 308 new protocols and related operational practices. 310 The cooperating dependence of network, application, and host to 311 provide communication performance on the Internet is uncertain 312 when only endpoints (i.e., at user devices and within service 313 platforms) can observe performance, and when performance cannot be 314 independently verified by all parties. The ability of other 315 stakeholders to review transport header traces can help develop 316 deeper insight into performance. In the heterogeneous Internet, 317 this helps extend the range of topologies, vendor equipment, and 318 traffic patterns that are evaluated. 320 Independently captured data is important to help ensure the health 321 of the research and development communities. It can provide input 322 and test scenarios to support development of new transport 323 protocol mechanisms, especially when this analysis can be based on 324 the behaviour experienced in a diversity of deployed networks. 326 Independently verifiable performance metrics might also be 327 utilised to demonstrate regulatory compliance in some 328 jurisdictions, and to provide a basis for informing design 329 decisions. 331 The last point leads us to consider the impact of hiding transport 332 headers in the specification and development of protocols and 333 standards. This has potential impact on: 335 o Understanding Feature Interactions: An appropriate vantage point, 336 coupled with timing information about traffic flows, provides a 337 valuable tool for benchmarking equipment, functions, and/or 338 configurations, and to understand complex feature interactions. 339 An inability to observe transport protocol information can limit 340 the ability to diagnose and explore interactions between features 341 at different protocol layers, a side-effect of not allowing a 342 choice of vantage point from which this information is observed. 344 o Supporting Common Specifications: Transmission Control Protocol 345 (TCP) is currently the predominant transport protocol used over 346 Internet paths. Its many variants have broadly consistent 347 approaches to avoiding congestion collapse, and to ensuring the 348 stability of the Internet. Increased use of transport layer 349 encryption can overcome ossification, allowing deployment of new 350 transports and different types of congestion control. This 351 flexibility can be beneficial, but it can come at the cost of 352 fragmenting the ecosystem. There is little doubt that developers 353 will try to produce high quality transports for their intended 354 target uses, but it is not clear there are sufficient incentives 355 to ensure good practice that benefits the wide diversity of 356 requirements for the Internet community as a whole. Increased 357 diversity, and the ability to innovate without public scrutiny, 358 risks point solutions that optimise for specific needs, but 359 accidentally disrupt operations of/in different parts of the 360 network. The social contract that maintains the stability of the 361 Internet relies on accepting common specifications. 363 o Operational Practice: The network operations community relies on 364 being able to understand the pattern and requirements of traffic 365 passing over the Internet, both in aggregate and at the flow 366 level. These operational practices have developed based on the 367 information available from unencrypted transport headers. If this 368 information is only carried in encrypted transport headers, 369 operators will not be able to use this information directly. If 370 operators still wish to use these practices, they may turn to more 371 ambitious ways of discovering this information. For example, if 372 an operator wants to know that traffic is audio traffic, and no 373 longer has access to Session Description Protocol (SDP) session 374 descriptions that would explicitly say a flow "is audio", the 375 operator might use heuristics to guess that short UDP packets with 376 regular spacing are carrying audio traffic. Operational practices 377 aimed at guessing transport parameters are out of scope for this 378 document, and are only mentioned here to recognize that encryption 379 may not prevent operators from attempting to apply the same 380 practices they used with unencrypted transport headers. 382 o Compliance: Published transport specifications allow operators and 383 regulators to check compliance. This can bring assurance to those 384 operating networks, often avoiding the need to deploy complex 385 techniques that routinely monitor and manage TCP/IP traffic flows 386 (e.g., avoiding the capital and operational costs of deploying 387 flow rate-limiting and network circuit-breaker methods [RFC8084]). 388 When it is not possible to observe transport header information, 389 methods are still needed to confirm that the traffic produced 390 conforms to the expectations of the operator or developer. 392 o Restricting research and development: Hiding transport information 393 can impede independent research into new mechanisms, measurement 394 of behaviour, and development initiatives. Experience shows that 395 transport protocols are complicated to design and complex to 396 deploy, and that individual mechanisms need to be evaluated while 397 considering other mechanisms, across a broad range of network 398 topologies and with attention to the impact on traffic sharing the 399 capacity. If this results in reduced availability of open data, 400 it could eliminate the independent self-checks to the 401 standardisation process that have previously been in place from 402 research and academic contributors (e.g., the role of the IRTF 403 Internet Congestion Control Research Groups (ICCRG) and research 404 publications in reviewing new transport mechanisms and assessing 405 the impact of their experimental deployment) 407 In summary, there are trade-offs. On the one hand, transport 408 protocol designers have often ignored the implications of whether the 409 information in transport header fields can or will be used by in- 410 network devices, and the implications this places on protocol 411 evolution. This motivates a design that provides confidentiality of 412 the header information. On the other hand, it can be expected that a 413 lack of visibility of transport header information can impact the 414 ways that protocols are deployed, standardised, and their operational 415 support. 417 To achieve stable Internet operations the IETF transport community 418 has to date relied heavily on measurement and insights of the network 419 operations community to understand the trade-offs, and to inform 420 selection of appropriate mechanisms, to ensure a safe, reliable, and 421 robust Internet (e.g., [RFC1273],[RFC2914]). 423 The choice of whether future transport protocols encrypt their 424 protocol headers therefore needs to be taken based not solely on 425 security and privacy considerations, but also taking into account the 426 impact on operations, standards, and research. As [RFC7258] notes: 427 "Making networks unmanageable to mitigate [pervasive monitoring] is 428 not an acceptable outcome, but ignoring [pervasive monitoring] would 429 go against the consensus documented here. An appropriate balance 430 will emerge over time as real instances of this tension are 431 considered." This balance between information exposed and 432 information concealed ought to be carefully considered when 433 specifying new transport protocols. 435 3. Current uses of Transport Headers within the Network 437 Despite transport headers having end-to-end meaning, some of these 438 transport headers have come to be used in various ways within the 439 Internet. In response to pervasive monitoring [RFC7624] revelations 440 and the IETF consensus that "Pervasive Monitoring is an Attack" 441 [RFC7258], efforts are underway to increase encryption of Internet 442 traffic,. Applying confidentiality to transport header fields would 443 affect how protocol information is used [RFC8404]. To understand 444 these implications, it is first necessary to understand how transport 445 layer headers are currently observed and/or modified by middleboxes 446 within the network. 448 Transport protocols can be designed to encrypt or authenticate 449 transport header fields. Authentication at the transport layer can 450 be used to detect any changes to an immutable header field that were 451 made by a network device along a path. The intentional modification 452 of transport headers by middleboxes (such as Network Address 453 Translation, NAT, or Firewalls) is not considered. Common issues 454 concerning IP address sharing are described in [RFC6269]. 456 3.1. Observing Transport Information in the Network 458 If in-network observation of transport protocol headers is needed, 459 this requires knowledge of the format of the transport header: 461 o Flows need to be identified at the level required to perform the 462 observation; 464 o The protocol and version of the header need to be visible, e.g., 465 by defining the wire image [I-D.trammell-wire-image]. As 466 protocols evolve over time and there could be a need to introduce 467 new transport headers. This could require interpretation of 468 protocol version information or connection setup information; 470 o The location and syntax of any observed transport headers needs to 471 be known. IETF transport protocols can specify this information. 473 The following subsections describe various ways that observable 474 transport information has been utilised. 476 3.1.1. Flow Identification 478 Transport protocol header information (together with information in 479 the network header), has been used to identify a flow and the 480 connection state of the flow, together with the protocol options 481 being used. In some usages, a low-numbered (well-known) transport 482 port number has been used to identify a protocol (although port 483 information alone is not sufficient to guarantee identification of a 484 protocol, since applications can use arbitrary ports, multiple 485 sessions can be multiplexed on a single port, and ports can be re- 486 used by subsequent sessions). 488 Transport protocols, such as TCP and the Stream Control Transport 489 Protocol (SCTP) specify a standard base header that includes sequence 490 number information and other data, with the possibility to negotiate 491 additional headers at connection setup, identified by an option 492 number in the transport header. UDP-based protocols can use, but 493 sometimes do not use, well-known port numbers. Some flows can 494 instead be identified by observing signalling protocol data (e.g., 495 [RFC3261], [I-D.ietf-rtcweb-overview]) or through the use of magic 496 numbers placed in the first byte(s) of the datagram payload 497 [RFC7983]. 499 Flow identification is a common function. For example, performed by 500 measurement activities, QoS classification, firewalls, Denial of 501 Service, DOS, prevention. It becomes more complex and less easily 502 achieved when multiplexing is used at or above the transport layer. 504 3.1.2. Metrics derived from Transport Layer Headers 506 Some actors manage their portion of the Internet by characterizing 507 the performance of link/network segments. Passive monitoring can 508 observe traffic that does not encrypt the transport header 509 information to make inferences from transport headers to derive these 510 performance metrics. A variety of open source and commercial tools 511 have been deployed that utilise this information. The following 512 metrics can be derived from transport header information: 514 Traffic Rate and Volume: Header information (e.g., sequence number 515 and packet size) allows derivation of volume measures per- 516 application, to characterise the traffic that uses a network 517 segment or the pattern of network usage. This can be measured per 518 endpoint or for an aggregate of endpoints (e.g., by an operator to 519 assess subscriber usage). It can also be used to trigger 520 measurement-based traffic shaping and to implement QoS support 521 within the network and lower layers. Volume measures can be 522 valuable for capacity planning and providing detail of trends, 523 rather than the volume per subscriber. 525 Loss Rate and Loss Pattern: Flow loss rate can be derived (e.g., 526 from transport sequence numbers) and has been used as a metric for 527 performance assessment and to characterise transport behaviour. 528 Understanding the location and root cause of loss can help an 529 operator determine whether this requires corrective action. 530 Network operators have used the variation in patterns of loss as a 531 key performance metric, utilising this to detect changes in the 532 offered service. 534 There are various causes of loss, including corruption of link 535 frames (e.g., interference on a radio link), buffer overflow 536 (e.g., due to congestion), policing (traffic management), buffer 537 management (e.g., Active Queue Management, AQM [RFC7567]), and 538 inadequate provision of traffic pre-emption. Understanding flow 539 loss rate requires either maintaining per flow packet counters or 540 by observing sequence numbers in transport headers. Loss can be 541 monitored at the interface level by devices in the network. It is 542 often valuable to understand the conditions under which packet 543 loss occurs. This usually requires relating loss to the traffic 544 flowing on the network node/segment at the time of loss. 546 Observation of transport feedback information (e.g., RTP Control 547 Protocol (RTCP) reception reports [RFC3550], TCP SACK blocks) can 548 increase understanding of the impact of loss and help identify 549 cases where loss could have been wrongly identified, or the 550 transport did not require the lost packet. It is sometimes more 551 helpful to understand the pattern of loss, than the loss rate, 552 because losses can often occur as bursts, rather than randomly- 553 timed events. 555 Throughput and Goodput: The throughput achieved by a flow can be 556 determined even when a flow is encrypted, providing the individual 557 flow can be identified. Goodput [RFC7928] is a measure of useful 558 data exchanged (the ratio of useful/total volume of traffic sent 559 by a flow). This requires ability to differentiate loss and 560 retransmission of packets (e.g., by observing packet sequence 561 numbers in the TCP or the Real-time Transport Protocol, RTP, 562 headers [RFC3550]). 564 Latency: Latency is a key performance metric that impacts 565 application response time and user-perceived response time. It 566 often indirectly impacts throughput and flow completion time. 567 Latency determines the reaction time of the transport protocol 568 itself, impacting flow setup, congestion control, loss recovery, 569 and other transport mechanisms. The observed latency can have 570 many components [Latency]. Of these, unnecessary/unwanted queuing 571 in network buffers has often been observed as a significant factor 573 [bufferbloat]. Once the cause of unwanted latency has been 574 identified, this can often be eliminated. 576 To measure latency across a part of a path, an observation point 577 can measure the experienced round trip time (RTT) using packet 578 sequence numbers, and acknowledgements, or by observing header 579 timestamp information. Such information allows an observation 580 point in the network to determine not only the path RTT, but also 581 to measure the upstream and downstream contribution to the RTT. 582 This could be used to locate a source of latency, e.g., by 583 observing cases where the median RTT is much greater than the 584 minimum RTT for a part of a path. 586 The service offered by network operators can benefit from latency 587 information to understand the impact of deployment and tune 588 deployed services. Latency metrics are key to evaluating and 589 deploying AQM [RFC7567], DiffServ [RFC2474], and Explicit 590 Congestion Notification (ECN) [RFC3168] [RFC8087]. Measurements 591 could identify excessively large buffers, indicating where to 592 deploy or configure AQM. An AQM method is often deployed in 593 combination with other techniques, such as scheduling [RFC7567] 594 [RFC8290] and although parameter-less methods are desired 595 [RFC7567], current methods [RFC8290] [RFC8289] [RFC8033] often 596 cannot scale across all possible deployment scenarios. 598 Variation in delay: Some network applications are sensitive to small 599 changes in packet timing (jitter). Short and long-term delay 600 variation can impact on the latency of a flow and the hence the 601 perceived quality of applications using the network (e.g., jitter 602 metrics are often cited when characterising paths supporting real- 603 time traffic). To assess the performance of such applications, it 604 can be necessary to measure the variation in delay observed along 605 a portion of the path [RFC3393] [RFC5481]. The requirements 606 resemble those for the measurement of latency. 608 Flow Reordering: Significant packet reordering within a flow can 609 impact time-critical applications and can be interpreted as loss 610 by reliable transports. Many transport protocol techniques are 611 impacted by reordering (e.g., triggering TCP retransmission, or 612 re-buffering of real-time applications). Packet reordering can 613 occur for many reasons, from equipment design to misconfiguration 614 of forwarding rules. Since this impacts transport performance, 615 network tools are needed to detect and measure unwanted/excessive 616 reordering. 618 There have been initiatives in the IETF transport area to reduce 619 the impact of reordering within a transport flow, possibly leading 620 to a reduction in the requirements for preserving ordering. These 621 have promise to simplify network equipment design as well as the 622 potential to improve robustness of the transport service. 623 Measurements of reordering can help understand the present level 624 of reordering within deployed infrastructure, and inform decisions 625 about how to progress such mechanisms. Key performance indicators 626 are retransmission rate, packet drop rate, sector utilisation 627 level, a measure of reordering, peak rate, the ECN congestion 628 experienced (CE) marking rate, etc. 630 Metrics have been defined that evaluate whether a network has 631 maintained packet order on a packet-by-packet basis [RFC4737] and 632 [RFC5236]. 634 Techniques for measuring reordering typically observe packet 635 sequence numbers. Some protocols provide in-built monitoring and 636 reporting functions. Transport fields in the RTP header [RFC3550] 637 [RFC4585] can be observed to derive traffic volume measurements 638 and provide information on the progress and quality of a session 639 using RTP. As with other measurement, metadata is often needed to 640 understand the context under which the data was collected, 641 including the time, observation point, and way in which metrics 642 were accumulated. The RTCP protocol directly reports some of this 643 information in a form that can be directly visible in the network. 644 A user of summary measurement data needs to trust the source of 645 this data and the method used to generate the summary information. 647 The above passively monitor transport protocol headers to derive 648 metrics about network layer performance useful for operation and 649 management of a network. 651 3.1.3. Transport use of Network Layer Header Fields 653 Information from the transport protocol can be used by a multi-field 654 classifier as a part of policy framework. Policies are commonly used 655 for management of the QoS or Quality of Experience (QoE) in resource- 656 constrained networks and by firewalls that use the information to 657 implement access rules (see also section 2.2.2 of [RFC8404]). 658 Network-layer classification methods that rely on a multi-field 659 classifier (e.g. Inferring QoS from the 5-tuple or choice of 660 application protocol) are incompatible with transport protocols that 661 encrypt the transport information. Traffic that cannot be 662 classified, will typically receive a default treatment. 664 Transport information can also be explicitly set in network-layer 665 header fields that are not encrypted. This can provide information 666 to enable a different forwarding treatment by the network, even when 667 a transport employs encryption to protect other header information. 669 On the one hand, the user of a transport that multiplexes multiple 670 sub-flows could wish to hide the presence and characteristics of 671 these sub-flows. On the other hand, an encrypted transport could set 672 the network-layer information to indicate the presence of sub-flows 673 and to reflect the network needs of individual sub-flows. There are 674 several ways this could be done: 676 IP Address: Applications expose the addresses used by endpoints, and 677 this is used in the forwarding decisions in network devices. 678 Address and other protocol information can be used by a Multi- 679 Field (MF) classifier to determine how traffic is treated 680 [RFC2475], and hence the quality of experience for a flow. 682 Using the IPv6 Network-Layer Flow Label: A number of Standards Track 683 and Best Current Practice RFCs (e.g., [RFC8085], [RFC6437], 684 [RFC6438]) encourage endpoints to set the IPv6 Flow label field of 685 the network-layer header. IPv6 "source nodes SHOULD assign each 686 unrelated transport connection and application data stream to a 687 new flow" [RFC6437]. A multiplexing transport could choose to use 688 multiple Flow labels to allow the network to independently forward 689 subflows. RFC6437 provides further guidance on choosing a flow 690 label value, stating these "should be chosen such that their bits 691 exhibit a high degree of variability", and chosen so that "third 692 parties should be unlikely to be able to guess the next value that 693 a source of flow labels will choose". To promote privacy, the 694 Flow Label assignment needs to avoid introducing linkability that 695 a network device may observe. Once set, a label can provide 696 information that can help inform network-layer queuing and 697 forwarding [RFC6438](e.g. for Equal Cost Multi-Path, ECMP, 698 routing, and Link Aggregation, LAG) [RFC6294]. [RFC6438] includes 699 describes considerations when used with IPsec. 701 Using the Network-Layer Differentiated Services Code Point: 702 Applications can expose their delivery expectations to the network 703 by setting the Differentiated Services Code Point (DSCP) field of 704 IPv4 and IPv6 packets [RFC2474]. For example, WebRTC applications 705 identify different forwarding treatments for individual sub-flows 706 (audio vs. video) based on the value of the DSCP field 707 [I-D.ietf-tsvwg-rtcweb-qos]). This provides explicit information 708 to inform network-layer queuing and forwarding, rather than an 709 operator inferring traffic requirements from transport and 710 application headers via a multi-field classifier. 712 Since the DSCP value can impact the quality of experience for a 713 flow, observations of service performance need to consider this 714 field when a network path has support for differentiated service 715 treatment. 717 Using Explicit Congestion Marking: ECN [RFC3168] is a transport 718 mechanism that utilises the ECN field in the network-layer header. 719 Use of ECN explicitly informs the network-layer that a transport 720 is ECN-capable, and requests ECN treatment of the flows packets. 721 An ECN-capable transport can offer benefits when used over a path 722 with equipment that implements an AQM method with Congestion 723 Experienced (CE) marking of IP packets [RFC8087], since it can 724 react to congestion without also having to recover from lost 725 packets. 727 ECN exposes the presence of congestion. The reception of CE- 728 marked packets can be used to estimate the level of incipient 729 congestion on the upstream portion of the path from the point of 730 observation (Section 2.5 of [RFC8087]). Interpreting the marking 731 behaviour (i.e., assessing congestion and diagnosing faults) 732 requires context from the transport layer (such as path RTT). 734 AQM and ECN offer a range of algorithms and configuration options. 735 Tools therefore need to be available to network operators and 736 researchers to understand the implication of configuration choices 737 and transport behaviour as use of ECN increases and new methods 738 emerge [RFC7567]. 740 Careful use of the network layer features can therefore help address 741 some of the reasons why the network inspects transport protocol 742 headers. 744 3.2. Transport Measurement 746 The common language between network operators and application/content 747 providers/users is packet transfer performance at a layer that all 748 can view and analyse. For most packets, this has been the transport 749 layer, until the emergence of QUIC, with the obvious exception of 750 Virtual Private Networks (VPNs) and IPsec. 752 When encryption conceals more layers in each packet, people seeking 753 understanding of the network operation rely more on pattern 754 inferences and other heuristics reliance on pattern inferences and 755 accuracy suffers. For example, the traffic patterns between server 756 and browser are dependent on browser supplier and version, even when 757 the sessions use the same server application (e.g., web e-mail 758 access). It remains to be seen whether more complex inferences can 759 be mastered to produce the same monitoring accuracy (see section 760 2.1.1 of [RFC8404]). 762 When measurement datasets are made available by servers or client 763 endpoints, additional metadata, such as the state of the network, is 764 often required to interpret this data. Collecting and coordinating 765 such metadata is more difficult when the observation point is at a 766 different location to the bottleneck/device under evaluation. 768 Packet sampling techniques can be used to scale the processing 769 involved in observing packets on high rate links. This exports only 770 the packet header information of (randomly) selected packets. The 771 utility of these measurements depends on the type of bearer and 772 number of mechanisms used by network devices. Simple routers are 773 relatively easy to manage, a device with more complexity demands 774 understanding of the choice of many system parameters. This level of 775 complexity exists when several network methods are combined. 777 This section discusses topics concerning observation of transport 778 flows, with a focus on transport measurement. 780 3.2.1. Point of Observation 782 On-path measurements are particularly useful for locating the source 783 of problems, or to assess the performance of a network segment or a 784 particular device configuration. Often issues can only be understood 785 in the context of the other flows that share a particular path, 786 common network device, interface port, etc. A simple example is 787 monitoring of a network device that uses a scheduler or active queue 788 management technique [RFC7567], where it could be desirable to 789 understand whether the algorithms are correctly controlling latency, 790 or if overload protection is working. This understanding implies 791 knowledge of how traffic is assigned to any sub-queues used for flow 792 scheduling, but can also require information about how the traffic 793 dynamics impact active queue management, starvation prevention 794 mechanisms, and circuit-breakers. 796 Sometimes multiple on-path observation points are needed. By 797 correlating observations of headers at multiple points along the path 798 (e.g., at the ingress and egress of a network segment), an observer 799 can determine the contribution of a portion of the path to an 800 observed metric, to locate a source of delay, jitter, loss, 801 reordering, congestion marking, etc. 803 3.2.2. Use by Operators to Plan and Provision Networks 805 Traffic measurements (e.g., traffic volume, loss, latency) is used by 806 operators to help plan deployment of new equipment and configurations 807 in their networks. Data is also valuable to equipment vendors who 808 want to understand traffic trends and patterns of usage as inputs to 809 decisions about planning products and provisioning for new 810 deployments. This measurement information can also be correlated 811 with billing information when this is also collected by an operator. 813 A network operator supporting traffic that uses transport header 814 encryption might not have access to per-flow measurement data. 815 Trends in aggregate traffic can be observed and can be related to the 816 endpoint addresses being used, but it may be impossible to correlate 817 patterns in measurements with changes in transport protocols (e.g., 818 the impact of changes in introducing a new transport protocol 819 mechanism). This increases the dependency on other indirect sources 820 of information to inform planning and provisioning. 822 3.2.3. Service Performance Measurement 824 Traffic measurements (e.g., traffic volume, loss, latency) can be 825 used by various actors to help analyse the performance offered to the 826 users of a network segment, and to inform operational practice. 828 While active measurements may be used within a network, passive 829 measurements can have advantages in terms of eliminating unproductive 830 test traffic, reducing the influence of test traffic on the overall 831 traffic mix, and the ability to choose the point of observation (see 832 Section 3.2.1). However, passive measurements can rely on observing 833 transport headers which is not possible if those headers are 834 encrypted. 836 3.2.4. Measuring Transport to Support Network Operations 838 Information provided by tools observing transport headers can help 839 determine whether mechanisms are needed in the network to prevent 840 flows from acquiring excessive network capacity. Operators can 841 implement operational practices to manage traffic flows (e.g., to 842 prevent flows from acquiring excessive network capacity under severe 843 congestion) by deploying rate-limiters, traffic shaping or network 844 transport circuit breakers [RFC8084]. 846 Congestion Control Compliance of Traffic: Congestion control is a 847 key transport function [RFC2914]. Many network operators 848 implicitly accept that TCP traffic complies with a behaviour that 849 is acceptable for use in the shared Internet. TCP algorithms have 850 been continuously improved over decades, and they have reached a 851 level of efficiency and correctness that custom application-layer 852 mechanisms will struggle to easily duplicate [RFC8085]. 854 A standards-compliant TCP stack provides congestion control that 855 may therefore be judged safe for use across the Internet. 856 Applications developed on top of well-designed transports can be 857 expected to appropriately control their network usage, reacting 858 when the network experiences congestion, by back-off and reduce 859 the load placed on the network. This is the normal expected 860 behaviour for IETF-specified transport (e.g., TCP and SCTP). 862 However, when anomalies are detected, tools can interpret the 863 transport protocol header information to help understand the 864 impact of specific transport protocols (or protocol mechanisms) on 865 the other traffic that shares a network. An observation in the 866 network can gain understanding of the dynamics of a flow and its 867 congestion control behaviour. Analysing observed flows can help 868 to build confidence that an application flow backs-off its share 869 of the network load in the face of persistent congestion, and 870 hence to understand whether the behaviour is appropriate for 871 sharing limited network capacity. For example, it is common to 872 visualise plots of TCP sequence numbers versus time for a flow to 873 understand how a flow shares available capacity, deduce its 874 dynamics in response to congestion, etc. The ability to identify 875 sources that contribute excessive congestion is important to safe 876 operation of network infrastructure, and mechanisms can inform 877 configuration of network devices to complement the endpoint 878 congestion avoidance mechanisms [RFC7567] [RFC8084] to avoid a 879 portion of the network being driven into congestion collapse 880 [RFC2914]. 882 Congestion Control Compliance for UDP traffic: UDP provides a 883 minimal message-passing datagram transport that has no inherent 884 congestion control mechanisms. Because congestion control is 885 critical to the stable operation of the Internet, applications and 886 other protocols that choose to use UDP as a transport are required 887 to employ mechanisms to prevent congestion collapse, avoid 888 unacceptable contributions to jitter/latency, and to establish an 889 acceptable share of capacity with concurrent traffic [RFC8085]. 891 A network operator needs tools to understand if datagram flows 892 comply with congestion control expectations and therefore whether 893 there is a need to deploy methods such as rate-limiters, transport 894 circuit breakers or other methods to enforce acceptable usage for 895 the offered service. 897 UDP flows that expose a well-known header by specifying the format 898 of header fields can allow information to be observed to gain 899 understanding of the dynamics of a flow and its congestion control 900 behaviour. For example, tools exist to monitor various aspects of 901 the RTP and RTCP header information of real-time flows (see 902 Section 3.1.2, and the Secure RTP extensions [RFC3711] were 903 explicitly designed to expose header information to enable such 904 observation. 906 3.3. Use for Network Diagnostics and Troubleshooting 908 Transport header information can be useful for a variety of 909 operational tasks [RFC8404]: to diagnose network problems, assess 910 network provider performance, evaluate equipment/protocol 911 performance, capacity planning, management of security threats 912 (including denial of service), and responding to user performance 913 questions. Sections 3.1.2 and 5 of [RFC8404] provide further 914 examples. These tasks seldom involve the need to determine the 915 contents of the transport payload, or other application details. 917 A network operator supporting traffic that uses transport header 918 encryption can see only encrypted transport headers. This prevents 919 deployment of performance measurement tools that rely on transport 920 protocol information. Choosing to encrypt all the information 921 reduces the ability of an operator to observe transport performance, 922 and could limit the ability of network operators to trace problems, 923 make appropriate QoS decisions, or response to other queries about 924 the network service. For some this will be blessing, for others it 925 may be a curse. For example, operational performance data about 926 encrypted flows needs to be determined by traffic pattern analysis, 927 rather than relying on traditional tools. This can impact the 928 ability of the operator to respond to faults, it could require 929 reliance on endpoint diagnostic tools or user involvement in 930 diagnosing and troubleshooting unusual use cases or non-trivial 931 problems. A key need here is for tools to provide useful information 932 during network anomalies (e.g., significant reordering, high or 933 intermittent loss). 935 Measurements can be used to monitor the health of a portion of the 936 Internet, to provide early warning of the need to take action. They 937 can assist in debugging and diagnosing the root causes of faults that 938 concern a particular user's traffic. They can also be used to 939 support post-mortem investigation after an anomaly to determine the 940 root cause of a problem. 942 In some case, measurements may involve active injection of test 943 traffic to perform a measurement. However, most operators do not 944 have access to user equipment, therefore the point of test is 945 normally different from the transport endpoint. Injection of test 946 traffic can incur an additional costs in running such tests (e.g., 947 the implications of capacity tests in a mobile network are obvious). 948 Some active measurements (e.g., response under load or particular 949 workloads) perturb other traffic, and could require dedicated access 950 to the network segment. An alternative approach is to use in-network 951 techniques that observe transport packet headers added while traffic 952 traverses an operational networks to make the measurements. These 953 measurements do not require the cooperation of an endpoint. 955 In other cases, measurement involves dissecting network traffic 956 flows. The observed transport layer information can help identify 957 whether the link/network tuning is effective and alert to potential 958 problems that can be hard to derive from link or device measurements 959 alone. The design trade-offs for radio networks are often very 960 different to those of wired networks. A radio-based network (e.g., 961 cellular mobile, enterprise WiFi, satellite access/back-haul, point- 962 to-point radio) has the complexity of a subsystem that performs radio 963 resource management,s with direct impact on the available capacity, 964 and potentially loss/reordering of packets. The impact of the 965 pattern of loss and congestion, differs for different traffic types, 966 correlation with propagation and interference can all have 967 significant impact on the cost and performance of a provided service. 968 The need for this type of information is expected to increase as 969 operators bring together heterogeneous types of network equipment and 970 seek to deploy opportunistic methods to access radio spectrum. 972 3.4. Header Compression 974 Header compression saves link bandwidth by compressing network and 975 transport protocol headers on a per-hop basis. It was widely used 976 with low bandwidth dial-up access links, and still finds application 977 on wireless links that are subject to capacity constraints. Header 978 compression has been specified for use with TCP/IP and RTP/UDP/IP 979 flows [RFC2507], [RFC2508], [RFC4995]. 981 While it is possible to compress only the network layer headers, 982 significant bandwidth savings can be made if both the network and 983 transport layer headers are compressed together as a single unit. 984 The Secure RTP extensions [RFC3711] were explicitly designed to leave 985 the transport protocol headers unencrypted, but authenticated, since 986 support for header compression was considered important. Encrypting 987 the transport protocol headers does not break such header 988 compression, but does cause it to fall back to compressing only the 989 network layer headers, with a significant reduction in efficiency. 990 This may have operational impact. 992 4. Encryption and Authentication of Transport Headers 994 End-to-end encryption can be applied at various protocol layers. It 995 can be applied above the transport to encrypt the transport payload. 996 Encryption methods can hide information from an eavesdropper in the 997 network. Encryption can also help protect the privacy of a user, by 998 hiding data relating to user/device identity or location. Neither an 999 integrity check nor encryption methods prevent traffic analysis, and 1000 usage needs to reflect that profiling of users, identification of 1001 location and fingerprinting of behaviour can take place even on 1002 encrypted traffic flows. Any header information that has a clear 1003 definition in the protocol's message format(s), or is implied by that 1004 definition, and is not cryptographically confidentiality-protected 1005 can be unambiguously interpreted by on-path observers 1006 [I-D.trammell-wire-image]. 1008 There are several motivations: 1010 o One motive to use encryption is a response to perceptions that the 1011 network has become ossified by over-reliance on middleboxes that 1012 prevent new protocols and mechanisms from being deployed. This 1013 has lead to a perception that there is too much "manipulation" of 1014 protocol headers within the network, and that designing to deploy 1015 in such networks is preventing transport evolution. In the light 1016 of this, a method that authenticates transport headers may help 1017 improve the pace of transport development, by eliminating the need 1018 to always consider deployed middleboxes 1019 [I-D.trammell-plus-abstract-mech], or potentially to only 1020 explicitly enable middlebox use for particular paths with 1021 particular middleboxes that are deliberately deployed to realise a 1022 useful function for the network and/or users[RFC3135]. 1024 o Another motivation stems from increased concerns about privacy and 1025 surveillance. Some Internet users have valued the ability to 1026 protect identity, user location, and defend against traffic 1027 analysis, and have used methods such as IPsec Encapsulated 1028 Security Payload (ESP), Virtual Private Networks (VPNs) and other 1029 encrypted tunnel technologies. Revelations about the use of 1030 pervasive surveillance [RFC7624] have, to some extent, eroded 1031 trust in the service offered by network operators, and following 1032 the Snowden revelation in the USA in 2013 has led to an increased 1033 desire for people to employ encryption to avoid unwanted 1034 "eavesdropping" on their communications. Concerns have also been 1035 voiced about the addition of information to packets by third 1036 parties to provide analytics, customization, advertising, cross- 1037 site tracking of users, to bill the customer, or to selectively 1038 allow or block content. Whatever the reasons, there are now 1039 activities in the IETF to design new protocols that could include 1040 some form of transport header encryption (e.g., QUIC 1041 [I-D.ietf-quic-transport]). 1043 Authentication methods (that provide integrity checks of protocols 1044 fields) have also been specified at the network layer, and this also 1045 protects transport header fields. The network layer itself carries 1046 protocol header fields that are increasingly used to help forwarding 1047 decisions reflect the need of transport protocols, such as the IPv6 1048 Flow Label [RFC6437], DSCP, and ECN fields. 1050 The use of transport layer authentication and encryption exposes a 1051 tussle between middlebox vendors, operators, applications developers 1052 and users. 1054 o On the one hand, future Internet protocols that enable large-scale 1055 encryption assist in the restoration of the end-to-end nature of 1056 the Internet by returning complex processing to the endpoints, 1057 since middleboxes cannot modify what they cannot see. 1059 o On the other hand, encryption of transport layer header 1060 information has implications for people who are responsible for 1061 operating networks and researchers and analysts seeking to 1062 understand the dynamics of protocols and traffic patterns. 1064 Whatever the motives, a decision to use pervasive transport header 1065 encryption will have implications on the way in which design and 1066 evaluation is performed, and which can in turn impact the direction 1067 of evolution of the transport protocol stack. While the IETF can 1068 specify protocols, the success in actual deployment is often 1069 determined by many factors [RFC5218] that are not always clear at the 1070 time when protocols are being defined. 1072 The following briefly reviews some security design options for 1073 transport protocols. A Survey of Transport Security Protocols 1074 [I-D.ietf-taps-transport-security] provides more details concerning 1075 commonly used encryption methods at the transport layer. 1077 Authenticating the Transport Protocol Header: Transport layer header 1078 information can be authenticated. An integrity check that 1079 protects the immutable transport header fields, but can still 1080 expose the transport protocol header information in the clear, 1081 allowing in-network devices to observe these fields. An integrity 1082 check can not prevent in-network modification, but can prevent a 1083 receiving from accepting changes and avoid impact on the transport 1084 protocol operation. 1086 An example transport authentication mechanism is TCP- 1087 Authentication (TCP-AO) [RFC5925]. This TCP option authenticates 1088 the IP pseudo header, TCP header, and TCP data. TCP-AO protects 1089 the transport layer, preventing attacks from disabling the TCP 1090 connection itself and provides replay protection. TCP-AO may 1091 interact with middleboxes, depending on their behaviour [RFC3234]. 1093 The IPsec Authentication Header (AH) [RFC4302] was designed to 1094 work at the network layer and authenticate the IP payload. This 1095 approach authenticates all transport headers, and verifies their 1096 integrity at the receiver, preventing in-network modification. 1098 Secure RTP [RFC3711] is another example of a transport protocol 1099 that allows header authentication. 1101 Greasing: Transport layer header information that is observable can 1102 be observed in the network. Protocols often provide extensibility 1103 features, reserving fields or values for use by future versions of 1104 a specification. The specification of receivers has traditionally 1105 ignored unspecified values, however in-network devices have 1106 emerged that ossify to require a certain value in a field, or re- 1107 use a field for another purpose. When the specification is later 1108 updated, it is impossible to deploy the new use of the field, and 1109 forwarding of the protocol could even become conditional on a 1110 specific header field value. 1112 A protocol can intentionally vary the value, format, and/or 1113 presence of observable transport header fields. This behaviour, 1114 known as GREASE (Generate Random Extensions And Sustain 1115 Extensibility), is designed to avoid a network device ossifying 1116 the use of a specific observable field. Greasing seeks to ease 1117 deployment of new methods. It can be designed to prevent in- 1118 network devices utilising the information in a transport header, 1119 or can make an observation robust to a set of changing values, 1120 rather than a specific set of values. 1122 Encrypting the Transport Payload: The transport layer payload can be 1123 encrypted to protect the content of transport segments. This 1124 leaves transport protocol header information in the clear. The 1125 integrity of immutable transport header fields could be protected 1126 by combining this with an integrity check. 1128 Examples of encrypting the payload include Transport Layer 1129 Security (TLS) over TCP [RFC8446] [RFC7525], Datagram TLS (DTLS) 1130 over UDP [RFC6347] [RFC7525], Secure RTP [RFC3711], and TCPcrypt 1131 [I-D.ietf-tcpinc-tcpcrypt] which permits opportunistic encryption 1132 of the TCP transport payload. 1134 Encrypting the Transport Headers and Payload: The network layer 1135 payload could be encrypted (including the entire transport header 1136 and the payload). This method provides confidentiality of the 1137 entire transport packet. It therefore does not expose any 1138 transport information to devices in the network, which also 1139 prevents modification along a network path. 1141 One example of encryption at the network layer is use of IPsec 1142 Encapsulating Security Payload (ESP) [RFC4303] in tunnel mode. 1143 This encrypts and authenticates all transport headers, preventing 1144 visibility of the transport headers by in-network devices. Some 1145 Virtual Private Network (VPN) methods also encrypt these headers. 1147 Selectively Encrypting Transport Headers and Payload: A transport 1148 protocol design can encrypt selected header fields, while also 1149 choosing to authenticate the entire transport header. This allows 1150 specific transport header fields to be made observable by network 1151 devices. End-to end integrity checks can prevent an endpoint from 1152 undetected modification of the immutable transport headers. 1154 Mutable fields in the transport header provide opportunities for 1155 middleboxes to modify the transport behaviour (e.g., the extended 1156 headers described in [I-D.trammell-plus-abstract-mech]). This 1157 considers only immutable fields in the transport headers, that is, 1158 fields that can be authenticated End-to-End across a path. 1160 An example of a method that encrypts some, but not all, transport 1161 information is GRE-in-UDP [RFC8086] when used with GRE encryption. 1163 Optional Encryption of Header Information: There are implications to 1164 the use of optional header encryption in the design of a transport 1165 protocol, where support of optional mechanisms can increase the 1166 complexity of the protocol and its implementation and in the 1167 management decisions that are required to use variable format 1168 fields. Instead, fields of a specific type ought to always be 1169 sent with the same level of confidentiality or integrity 1170 protection. 1172 As seen, different transports use encryption to protect their header 1173 information to varying degrees. There is, however, a trend towards 1174 increased protection with newer transport protocols. 1176 5. Addition of Transport Information to Network-Layer Protocol Headers 1178 Some measurements can be made by adding additional protocol headers 1179 carrying operations, administration and management (OAM) information 1180 to packets at the ingress to a maintenance domain (e.g., an Ethernet 1181 protocol header with timestamps and sequence number information using 1182 a method such as 802.11ag or in-situ OAM [I-D.ietf-ippm-ioam-data]) 1183 and removing the additional header at the egress of the maintenance 1184 domain. This approach enables some types of measurements, but does 1185 not cover the entire range of measurements described in this 1186 document. In some cases, it can be difficult to position measurement 1187 tools at the required segments/nodes and there can be challenges in 1188 correlating the downsream/upstream information when in-band OAM data 1189 is inserted by an on-path device. This has the advantage that a 1190 single header can support all transport protocols, but there could 1191 also be less desirable implications of separating the operation of 1192 the transport protocol from the measurement framework. 1194 Another example of a network-layer approach is the IPv6 Performance 1195 and Diagnostic Metrics (PDM) Destination Option [RFC8250]. This 1196 allows a sender to optionally include a destination option that 1197 caries header fields that can be used to observe timestamps and 1198 packet sequence numbers. This information could be authenticated by 1199 receiving transport endpoints when the information is added at the 1200 sender and visible at the receiving endpoint, although methods to do 1201 this have not currently been proposed. This method needs to be 1202 explicitly enabled at the sender. 1204 Current measurements suggest it can be undesirable to rely on methods 1205 requiring the presence of network options or extension headers. IPv4 1206 network options are often not supported (or are carried on a slower 1207 processing path) and some IPv6 networks are also known to drop 1208 packets that set an IPv6 header extension (e.g., [RFC7872]). Another 1209 disadvantage is that protocols that separately expose header 1210 information do not necessarily have an advantage to expose the 1211 information that is utilised by the protocol itself, and could 1212 manipulate this header information to gain an advantage from the 1213 network. 1215 6. Implications of Protecting the Transport Headers 1217 The choice of which fields to expose and which to encrypt is a design 1218 choice for the transport protocol. Any selective encryption method 1219 requires trading two conflicting goals for a transport protocol 1220 designer to decide which header fields to encrypt. Security work 1221 typically employs a design technique that seeks to expose only what 1222 is needed. This approach provides incentives to not reveal any 1223 information that is not necessary for the end-to-end communication. 1224 However, there can be performance and operational benefits in 1225 exposing selected information to network tools. 1227 This section explores key implications of working with encrypted 1228 transport protocols. 1230 6.1. Independent Measurement 1232 Independent observation by multiple actors is important for 1233 scientific analysis. Encrypting transport header encryption changes 1234 the ability for other actors to collect and independently analyse 1235 data. Internet transport protocols employ a set of mechanisms. Some 1236 of these need to work in cooperation with the network layer - loss 1237 detection and recovery, congestion detection and congestion control, 1238 some of these need to work only end-to-end (e.g., parameter 1239 negotiation, flow-control). 1241 The majority of present Internet applications use two well-known 1242 transport protocols, TCP and UDP. Although TCP represents the 1243 majority of current traffic, some real-time applications use UDP, and 1244 much of this traffic utilises RTP format headers in the payload of 1245 the UDP datagram. Since these protocol headers have been fixed for 1246 decades, a range of tools and analysis methods have became common and 1247 well-understood. 1249 Protocols that expose the state information used by the transport 1250 protocol in their header information (e.g., timestamps used to 1251 calculate the RTT, packet numbers used to asses congestion and 1252 requests for retransmission) provide an incentive for the sending 1253 endpoint to provide correct information, increasing confidence that 1254 the observer understands the transport interaction with the network. 1255 For example, when TCP is used over an unencrypted network path (i.e., 1256 one that does not use IPsec or other encryption below the transport), 1257 it implicitly exposes header information that can be used for 1258 measurement at any point along the path. This information is 1259 necessary for the protocol's correct operation, therefore there is no 1260 incentive for a TCP implementation to put incorrect information in 1261 this transport header. A network device can have confidence that the 1262 well-known (and ossified) transport information represents the actual 1263 state of the endpoints. 1265 When encryption is used to conceal some or all of the transport 1266 headers, the transport protocol choose what information to reveal to 1267 the network about its internal state, what information to leave 1268 encrypted, and what fields to grease to protect against future 1269 ossification. Such a transport could be designed, for example, to 1270 provide summary data regarding its performance, congestion control 1271 state, etc., or to make an explicit measurement signal available. 1272 For example, a QUIC endpoint could set the spin bit to reflect to 1273 explicitly reveal a session's RTT [I-D.ietf-quic-spin-exp]). 1275 When providing or using such information, it becomes important to 1276 consider the privacy of the user and their incentive for providing 1277 accurate and detailed information. Protocols that selectively reveal 1278 some transport state or measurement signals are choosing to establish 1279 a trust relationship with the network operators. There is no 1280 protocol mechanism that can guarantee that the information provided 1281 represents the actual transport state of the endpoints, since those 1282 endpoints can always send additional information in the encrypted 1283 part of the header, to update to replace whatever they reveal. This 1284 reduces the ability to independently measure and verify that a 1285 protocol is behaving as expected. Some operational uses need the 1286 information to contain sufficient detail to understand, and possibly 1287 reconstruct, the network traffic pattern for further testing; such 1288 operators must gain the trust of transport protocol implementers if 1289 they are to correctly reveal such information. 1291 For some usage a standardised endpoint-based logging format (e.g., 1292 based onQuic-Trace [Quic-Trace]) could offer an alternative to in- 1293 network measurement. Such information will have a diversity of uses 1294 - examples include developers wishing to debug/understand the 1295 transport/applictaion protocols with which they work, to researchers 1296 seeking to spot trends, anomalies and to characterise variants of 1297 protocols. This use will need to establish the validity and 1298 provenance of the logging information (e.g., to establish how and 1299 when traces were captured). 1301 However, endpoint logs do not provide equivalent information to in- 1302 network measurements. In particular, endpoint logs contain only a 1303 part of the information needed to understand the operation of network 1304 devices and identify issues such as link performance or capacity 1305 sharing between multiple flows. Additional information is needed to 1306 determine which equipment/links are used and the configuration of 1307 equipment along the network paths being measured. 1309 6.2. Characterising "Unknown" Network Traffic 1311 The patterns and types of traffic that share Internet capacity change 1312 over time as networked applications, usage patterns and protocols 1313 continue to evolve. 1315 If "unknown" or "uncharacterised" traffic patterns form a small part 1316 of the traffic aggregate passing through a network device or segment 1317 of the network the path, the dynamics of the uncharacterised traffic 1318 may not have a significant collateral impact on the performance of 1319 other traffic that shares this network segment. Once the proportion 1320 of this traffic increases, the need to monitor the traffic and 1321 determine if appropriate safety measures need to be put in place. 1323 Tracking the impact of new mechanisms and protocols requires traffic 1324 volume to be measured and new transport behaviours to be identified. 1325 This is especially true of protocols operating over a UDP substrate. 1326 The level and style of encryption needs to be considered in 1327 determining how this activity is performed. On a shorter timescale, 1328 information may also need to be collected to manage denial of service 1329 attacks against the infrastructure. 1331 6.3. Accountability and Internet Transport Protocols 1333 Information provided by tools observing transport headers can be used 1334 to classify traffic, and to limit the network capacity used by 1335 certain flows, as discussed in Section 3.2.4). Equally, operators 1336 could use analysis of transport headers and transport flow state to 1337 demonstrate that they are not providing differential treatment to 1338 certain flows. Obfuscating or hiding this information using 1339 encryption may lead operators and maintainers of middleboxes 1340 (firewalls, etc.) to seek other methods to classify, and potentially 1341 other mechanisms to condition, network traffic. 1343 A lack of data that reduces the level of precision with which flows 1344 can be classified also reduces the design space for conditioning 1345 mechanisms (e.g., rate limiting, circuit breaker techniques 1346 [RFC8084], or blocking of uncharacterised traffic), and this needs to 1347 be considered when evaluating the impact of designs for transport 1348 encryption [RFC5218]. 1350 6.4. Impact on Operational Cost 1352 Many network operators currently utilise observed transport 1353 information as a part of their operational practice, and have 1354 developed tools and operational practices based around currently 1355 deployed transports and their applications. Encryption of the 1356 transport information prevents tools from directly observing this 1357 information. A variety of open source and commercial tools have been 1358 deployed that utilise this information for a variety of short and 1359 long term measurements. 1361 The network will not break just because transport headers are 1362 encrypted, although alternative diagnostic and troubleshooting tools 1363 would need to be developed and deployed. Introducing a new protocol 1364 or application can require these tool chains and practice to be 1365 updated, and may in turn impact operational mechanisms, and policies. 1366 Each change can introduce associated costs, including the cost of 1367 collecting data, and the tooling needed to handle multiple formats 1368 (possibly as these co-exist in the network, when measurements need to 1369 span time periods during which changes are deployed, or to compare 1370 with historical data). These costs are incurred by an operator to 1371 manage the service and debug network issues. 1373 At the time of writing, the additional operational cost of using 1374 encrypted transports is not yet well understood. Design trade-offs 1375 could mitigate these costs by explicitly choosing to expose selected 1376 information (e.g., header invariants and the spin-bit in 1377 QUIC[I-D.ietf-quic-transport]), the specification of common log 1378 formats and development of alternative approaches. 1380 6.5. Impact on Research, Development and Deployment 1382 Measurement has a critical role in the design of transport protocol 1383 mechanisms and their acceptance by the wider community (e.g., as a 1384 method to judge the safety for Internet deployment) and is 1385 increasingly being used to inform design decisions in networking 1386 research, during development of new mechanisms and protocols and in 1387 standardisation. Observation of pathologies are also important in 1388 understanding the interactions between cooperating protocols and 1389 network mechanism, the implications of sharing capacity with other 1390 traffic and the impact of different patterns of usage. 1392 Evolution and the ability to understand (measure) the impact need to 1393 proceed hand-in-hand. Attention needs to be paid to the expected 1394 scale of deployment of new protocols and protocol mechanisms. 1395 Whatever the mechanism, experience has shown that it is often 1396 difficult to correctly implement combination of mechanisms [RFC8085]. 1397 These mechanisms therefore typically evolve as a protocol matures, or 1398 in response to changes in network conditions, changes in network 1399 traffic or changes to application usage. 1401 New transport protocol formats are expected to facilitate an 1402 increased pace of transport evolution, and with it the possibility to 1403 experiment with and deploy a wide range of protocol mechanisms. 1404 There has been recent interest in a wide range of new transport 1405 methods, e.g., Larger Initial Window, Proportional Rate Reduction 1406 (PRR), congestion control methods based on measuring bottleneck 1407 bandwidth and round-trip propagation time, the introduction of AQM 1408 techniques and new forms of ECN response (e.g., Data Centre TCP, 1409 DCTP, and methods proposed for L4S).The growth and diversity of 1410 applications and protocols using the Internet also continues to 1411 expand. For each new method or application it is desirable to build 1412 a body of data reflecting its behaviour under a wide range of 1413 deployment scenarios, traffic load, and interactions with other 1414 deployed/candidate methods. 1416 Open standards motivate a desire for this evaluation to include 1417 independent observation and evaluation of performance data, which in 1418 turn suggests control over where and when measurement samples are 1419 collected. This requires consideration of the appropriate balance 1420 between encrypting all and no transport information. 1422 7. Conclusions 1424 Confidentiality and strong integrity checks have properties that are 1425 being incorporated into new protocols and that have important 1426 benefits. The pace of development of transports using the WebRTC 1427 data channel and the rapid deployment of QUIC transport protocol can 1428 both be attributed to using the combination of UDP as a substrate 1429 while providing confidentiality and authentication of the 1430 encapsulated transport headers and payload. 1432 The traffic that can be observed by on-path network devices is a 1433 function of transport protocol design/options, network use, 1434 applications, and user characteristics. In general, when only a 1435 small proportion of the traffic has a specific (different) 1436 characteristic, such traffic seldom leads to operational concern, 1437 although the ability to measure and monitor it is less. The desire 1438 to understand the traffic and protocol interactions typically grows 1439 as the proportion of traffic increases in volume. The challenges 1440 increase when multiple instances of an evolving protocol contribute 1441 to the traffic that share network capacity. 1443 An increased pace of evolution therefore needs to be accompanied by 1444 methods that can be successfully deployed and used across operational 1445 networks. This leads to a need for network operators (at various 1446 level (ISPs, enterprises, firewall maintainer, etc) to identify 1447 appropriate operational support functions and procedures. 1449 Protocols that change their transport header format (wire format) or 1450 their behaviour (e.g., algorithms that are needed to classify and 1451 characterise the protocol), will require new tooling to be developed 1452 to catch-up with the changes. If the currently deployed tools and 1453 methods are no longer relevant then it may no longer be possible to 1454 correctly measure performance. This can increase the response-time 1455 after faults, and can impact the ability to manage the network 1456 resulting in traffic causing traffic to be treated inappropriately 1457 (e.g., rate limiting because of being incorrectly classified/ 1458 monitored). 1460 There are benefits in exposing consistent information to the network 1461 that avoids traffic being mis-classified and then receiving a default 1462 treatment by the network. The flow label and DSCP fields provide 1463 examples of how transport information can be made available for 1464 network-layer decisions. Extension headers could also be used to 1465 carry transport information that can inform network-layer decisions. 1467 As a part of its design a new protocol specification therefore needs 1468 to weigh the benefits of ossifying common headers, versus the 1469 potential demerits of exposing specific information that could be 1470 observed along the network path, to provide tools to manage new 1471 variants of protocols. This can be done for the entire transport 1472 header, or by dividing header fields between those that are 1473 observable and mutable; those that are observable, but immutable; and 1474 those that are hidden/obfusticated. 1476 Several scenarios to illustrate different ways this could evolve are 1477 provided below: 1479 o One scenario is when transport protocols provide consistent 1480 information to the network by intentionally exposing a part of the 1481 transport header. The design fixes the format of this information 1482 between versions of the protocol. This ossification of the 1483 transport header allows an operator to establish tooling and 1484 procedures that enable it to provide consistent traffic management 1485 as the protocol evolves. In contrast to TCP (where all protocol 1486 information is exposed), evolution of the transport is facilitated 1487 by providing cryptographic integrity checks of the transport 1488 header fields (preventing undetected middlebox changes) and 1489 encryption of other protocol information (preventing observation 1490 within the network, or providing incentives for the use of the 1491 exposed information, rather than inferring information from other 1492 characteristics of the flow traffic). The exposed transport 1493 information can be used by operators to provide troubleshooting, 1494 measurement and any necessary functions appropriate to the class 1495 of traffic (priority, retransmission, reordering, circuit 1496 breakers, etc). 1498 o An alternative scenario adopts different design goals, with a 1499 different outcome. A protocol that encrypts all header 1500 information forces network operators to act independently from 1501 apps/transport developments to extract the information they need 1502 to manage their network. A range of approaches could proliferate, 1503 as in current networks. Some operators can add a shim header to 1504 each packet as a flow as it crosses the network; other operators/ 1505 managers could develop heuristics and pattern recognition to 1506 derive information that classifies flows and estimates quality 1507 metrics for the service being used; some could decide to rate- 1508 limit or block traffic until new tooling is in place. In many 1509 cases, the derived information can be used by operators to provide 1510 necessary functions appropriate to the class of traffic (priority, 1511 retransmission, reordering, circuit breakers, etc). 1512 Troubleshooting, and measurement becomes more difficult, and more 1513 diverse. This could require additional information beyond that 1514 visible in the packet header and when this information is used to 1515 inform decisions by on-path devices it can lead to dependency on 1516 other characteristics of the flow. In some cases, operators might 1517 need access to keying information to interpret encrypted data that 1518 they observe. Some use cases could demand use of transports that 1519 do not use encryption. 1521 The direction in which this evolves could have significant 1522 implications on the way the Internet architecture develops. It 1523 exposes a risk that significant actors (e.g., developers and 1524 transport designers) achieve more control of the way in which the 1525 Internet architecture develops.In particular, there is a possibility 1526 that designs could evolve to significantly benefit of customers for a 1527 specific vendor, and that communities with very different network, 1528 applications or platforms could then suffer at the expense of 1529 benefits to their vendors own customer base. In such a scenario, 1530 there could be no incentive to support other applications/products or 1531 to work in other networks leading to reduced access for new 1532 approaches. 1534 8. Security Considerations 1536 This document is about design and deployment considerations for 1537 transport protocols. Issues relating to security are discussed in 1538 the various sections of the document. 1540 Authentication, confidentiality protection, and integrity protection 1541 are identified as Transport Features by [RFC8095]. As currently 1542 deployed in the Internet, these features are generally provided by a 1543 protocol or layer on top of the transport protocol 1544 [I-D.ietf-taps-transport-security]. 1546 Confidentiality and strong integrity checks have properties that can 1547 also be incorporated into the design of a transport protocol. 1548 Integrity checks can protect an endpoint from undetected modification 1549 of protocol fields by network devices, whereas encryption and 1550 obfuscation or greasing can further prevent these headers being 1551 utilised by network devices. Hiding headers can therefore provide 1552 the opportunity for greater freedom to update the protocols and can 1553 ease experimentation with new techniques and their final deployment 1554 in endpoints. A protocol specification needs to weigh the benefits 1555 of ossifying common headers, versus the potential demerits of 1556 exposing specific information that could be observed along the 1557 network path to provide tools to manage new variants of protocols. 1559 A protocol design that uses header encryption can provide 1560 confidentiality of some or all of the protocol header information. 1561 This prevents an on-path device from knowledge of the header field. 1562 It therefore prevents mechanisms being built that directly rely on 1563 the information or seeks to infer semantics of an exposed header 1564 field. Hiding headers can limit the ability to measure and 1565 characterise traffic. 1567 Exposed transport headers are sometimes utilised as a part of the 1568 information to detect anomalies in network traffic. This can be used 1569 as the first line of defence yo identify potential threats from DOS 1570 or malware and redirect suspect traffic to dedicated nodes 1571 responsible for DOS analysis, malware detection, or to perform packet 1572 "scrubbing" (the normalization of packets so that there are no 1573 ambiguities in interpretation by the ultimate destination of the 1574 packet). These techniques are currently used by some operators to 1575 also defend from distributed DOS attacks. 1577 Exposed transport header fields are sometimes also utilised as a part 1578 of the information used by the receiver of a transport protocol to 1579 protect the transport layer from data injection by an attacker. In 1580 evaluating this use of exposed header information, it is important to 1581 consider whether it introduces a significant DOS threat. For 1582 example, an attacker could construct a DOS attack by sending packets 1583 with a sequence number that falls within the currently accepted range 1584 of sequence numbers at the receiving endpoint, this would then 1585 introduce additional work at the receiving endpoint, even though the 1586 data in the attacking packet may not finally be delivered by the 1587 transport layer. This is sometimes known as a "shadowing attack". 1588 An attack can, for example, disrupt receiver processing, trigger loss 1589 and retransmission, or make a receiving endpoint perform unproductive 1590 decryption of packets that cannot be successfully decrypted (forcing 1591 a receiver to commit decryption resources, or to update and then 1592 restore protocol state). 1594 One mitigation to off-path attack is to deny knowledge of what header 1595 information is accepted by a receiver or obfuscate the accepted 1596 header information, e.g., setting a non-predictable initial value for 1597 a sequence number during a protocol handshake, as in [RFC3550] and 1598 [RFC6056], or a port value that can not be predicted (see section 5.1 1599 of [RFC8085]). A receiver could also require additional information 1600 to be used as a part of check before accepting packets at the 1601 transport layer (e.g., utilising a part of the sequence number space 1602 that is encrypted; or by verifying an encrypted token not visible to 1603 an attacker). This would also mitigate on-path attacks. An 1604 additional processing cost can be incurred when decryption needs to 1605 be attempted before a receiver is able to discard injected packets. 1607 Open standards motivate a desire for this evaluation to include 1608 independent observation and evaluation of performance data, which in 1609 turn suggests control over where and when measurement samples are 1610 collected. This requires consideration of the appropriate balance 1611 between encrypting all and no transport information. Open data, and 1612 accessibility to tools that can help understand trends in application 1613 deployment, network traffic and usage patterns can all contribute to 1614 understanding security challenges. 1616 9. IANA Considerations 1618 XX RFC ED - PLEASE REMOVE THIS SECTION XXX 1620 This memo includes no request to IANA. 1622 10. Acknowledgements 1624 The authors would like to thank Mohamed Boucadair, Spencer Dawkins, 1625 Tom Herbert, Jana Iyengar, Mirja Kuehlewind, Kyle Rose, Kathleen 1626 Moriarty, Al Morton, Chris Seal, Joe Touch, Brian Trammell, Chris 1627 Wood, and other members of the TSVWG for their comments and feedback. 1629 This work has received funding from the European Union's Horizon 2020 1630 research and innovation programme under grant agreement No 688421.The 1631 opinions expressed and arguments employed reflect only the authors' 1632 view. The European Commission is not responsible for any use that 1633 may be made of that information. 1635 This work has received funding from the UK Engineering and Physical 1636 Sciences Research Council under grant EP/R04144X/1. 1638 11. Informative References 1640 [bufferbloat] 1641 Gettys, J. and K. Nichols, "Bufferbloat: dark buffers in 1642 the Internet. Communications of the ACM, 55(1):57-65", 1643 January 2012. 1645 [I-D.ietf-ippm-ioam-data] 1646 Brockners, F., Bhandari, S., Pignataro, C., Gredler, H., 1647 Leddy, J., Youell, S., Mizrahi, T., Mozes, D., Lapukhov, 1648 P., Chang, R., daniel.bernier@bell.ca, d., and J. Lemon, 1649 "Data Fields for In-situ OAM", draft-ietf-ippm-ioam- 1650 data-03 (work in progress), June 2018. 1652 [I-D.ietf-quic-spin-exp] 1653 Trammell, B. and M. Kuehlewind, "The QUIC Latency Spin 1654 Bit", draft-ietf-quic-spin-exp-01 (work in progress), 1655 October 2018. 1657 [I-D.ietf-quic-transport] 1658 Iyengar, J. and M. Thomson, "QUIC: A UDP-Based Multiplexed 1659 and Secure Transport", draft-ietf-quic-transport-14 (work 1660 in progress), August 2018. 1662 [I-D.ietf-rtcweb-overview] 1663 Alvestrand, H., "Overview: Real Time Protocols for 1664 Browser-based Applications", draft-ietf-rtcweb-overview-19 1665 (work in progress), November 2017. 1667 [I-D.ietf-taps-transport-security] 1668 Pauly, T., Perkins, C., Rose, K., and C. Wood, "A Survey 1669 of Transport Security Protocols", draft-ietf-taps- 1670 transport-security-02 (work in progress), June 2018. 1672 [I-D.ietf-tcpinc-tcpcrypt] 1673 Bittau, A., Giffin, D., Handley, M., Mazieres, D., Slack, 1674 Q., and E. Smith, "Cryptographic protection of TCP Streams 1675 (tcpcrypt)", draft-ietf-tcpinc-tcpcrypt-12 (work in 1676 progress), June 2018. 1678 [I-D.ietf-tsvwg-rtcweb-qos] 1679 Jones, P., Dhesikan, S., Jennings, C., and D. Druta, "DSCP 1680 Packet Markings for WebRTC QoS", draft-ietf-tsvwg-rtcweb- 1681 qos-18 (work in progress), August 2016. 1683 [I-D.thomson-quic-grease] 1684 Thomson, M., "More Apparent Randomization for QUIC", 1685 draft-thomson-quic-grease-00 (work in progress), December 1686 2017. 1688 [I-D.trammell-plus-abstract-mech] 1689 Trammell, B., "Abstract Mechanisms for a Cooperative Path 1690 Layer under Endpoint Control", draft-trammell-plus- 1691 abstract-mech-00 (work in progress), September 2016. 1693 [I-D.trammell-wire-image] 1694 Trammell, B. and M. Kuehlewind, "The Wire Image of a 1695 Network Protocol", draft-trammell-wire-image-04 (work in 1696 progress), April 2018. 1698 [Latency] Briscoe, B., "Reducing Internet Latency: A Survey of 1699 Techniques and Their Merits, IEEE Comm. Surveys & 1700 Tutorials. 26;18(3) p2149-2196", November 2014. 1702 [Measure] Fairhurst, G., Kuehlewind, M., and D. Lopez, "Measurement- 1703 based Protocol Design, Eur. Conf. on Networks and 1704 Communications, Oulu, Finland.", June 2017. 1706 [Quic-Trace] 1707 "https:QUIC trace utilities //github.com/google/quic- 1708 trace". 1710 [RFC1273] Schwartz, M., "Measurement Study of Changes in Service- 1711 Level Reachability in the Global TCP/IP Internet: Goals, 1712 Experimental Design, Implementation, and Policy 1713 Considerations", RFC 1273, DOI 10.17487/RFC1273, November 1714 1991, . 1716 [RFC2474] Nichols, K., Blake, S., Baker, F., and D. Black, 1717 "Definition of the Differentiated Services Field (DS 1718 Field) in the IPv4 and IPv6 Headers", RFC 2474, 1719 DOI 10.17487/RFC2474, December 1998, 1720 . 1722 [RFC2475] Blake, S., Black, D., Carlson, M., Davies, E., Wang, Z., 1723 and W. Weiss, "An Architecture for Differentiated 1724 Services", RFC 2475, DOI 10.17487/RFC2475, December 1998, 1725 . 1727 [RFC2507] Degermark, M., Nordgren, B., and S. Pink, "IP Header 1728 Compression", RFC 2507, DOI 10.17487/RFC2507, February 1729 1999, . 1731 [RFC2508] Casner, S. and V. Jacobson, "Compressing IP/UDP/RTP 1732 Headers for Low-Speed Serial Links", RFC 2508, 1733 DOI 10.17487/RFC2508, February 1999, 1734 . 1736 [RFC2914] Floyd, S., "Congestion Control Principles", BCP 41, 1737 RFC 2914, DOI 10.17487/RFC2914, September 2000, 1738 . 1740 [RFC3135] Border, J., Kojo, M., Griner, J., Montenegro, G., and Z. 1741 Shelby, "Performance Enhancing Proxies Intended to 1742 Mitigate Link-Related Degradations", RFC 3135, 1743 DOI 10.17487/RFC3135, June 2001, 1744 . 1746 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 1747 of Explicit Congestion Notification (ECN) to IP", 1748 RFC 3168, DOI 10.17487/RFC3168, September 2001, 1749 . 1751 [RFC3234] Carpenter, B. and S. Brim, "Middleboxes: Taxonomy and 1752 Issues", RFC 3234, DOI 10.17487/RFC3234, February 2002, 1753 . 1755 [RFC3261] Rosenberg, J., Schulzrinne, H., Camarillo, G., Johnston, 1756 A., Peterson, J., Sparks, R., Handley, M., and E. 1757 Schooler, "SIP: Session Initiation Protocol", RFC 3261, 1758 DOI 10.17487/RFC3261, June 2002, 1759 . 1761 [RFC3393] Demichelis, C. and P. Chimento, "IP Packet Delay Variation 1762 Metric for IP Performance Metrics (IPPM)", RFC 3393, 1763 DOI 10.17487/RFC3393, November 2002, 1764 . 1766 [RFC3550] Schulzrinne, H., Casner, S., Frederick, R., and V. 1767 Jacobson, "RTP: A Transport Protocol for Real-Time 1768 Applications", STD 64, RFC 3550, DOI 10.17487/RFC3550, 1769 July 2003, . 1771 [RFC3711] Baugher, M., McGrew, D., Naslund, M., Carrara, E., and K. 1772 Norrman, "The Secure Real-time Transport Protocol (SRTP)", 1773 RFC 3711, DOI 10.17487/RFC3711, March 2004, 1774 . 1776 [RFC4302] Kent, S., "IP Authentication Header", RFC 4302, 1777 DOI 10.17487/RFC4302, December 2005, 1778 . 1780 [RFC4303] Kent, S., "IP Encapsulating Security Payload (ESP)", 1781 RFC 4303, DOI 10.17487/RFC4303, December 2005, 1782 . 1784 [RFC4585] Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey, 1785 "Extended RTP Profile for Real-time Transport Control 1786 Protocol (RTCP)-Based Feedback (RTP/AVPF)", RFC 4585, 1787 DOI 10.17487/RFC4585, July 2006, 1788 . 1790 [RFC4737] Morton, A., Ciavattone, L., Ramachandran, G., Shalunov, 1791 S., and J. Perser, "Packet Reordering Metrics", RFC 4737, 1792 DOI 10.17487/RFC4737, November 2006, 1793 . 1795 [RFC4995] Jonsson, L-E., Pelletier, G., and K. Sandlund, "The RObust 1796 Header Compression (ROHC) Framework", RFC 4995, 1797 DOI 10.17487/RFC4995, July 2007, 1798 . 1800 [RFC5218] Thaler, D. and B. Aboba, "What Makes for a Successful 1801 Protocol?", RFC 5218, DOI 10.17487/RFC5218, July 2008, 1802 . 1804 [RFC5236] Jayasumana, A., Piratla, N., Banka, T., Bare, A., and R. 1805 Whitner, "Improved Packet Reordering Metrics", RFC 5236, 1806 DOI 10.17487/RFC5236, June 2008, 1807 . 1809 [RFC5481] Morton, A. and B. Claise, "Packet Delay Variation 1810 Applicability Statement", RFC 5481, DOI 10.17487/RFC5481, 1811 March 2009, . 1813 [RFC5925] Touch, J., Mankin, A., and R. Bonica, "The TCP 1814 Authentication Option", RFC 5925, DOI 10.17487/RFC5925, 1815 June 2010, . 1817 [RFC6056] Larsen, M. and F. Gont, "Recommendations for Transport- 1818 Protocol Port Randomization", BCP 156, RFC 6056, 1819 DOI 10.17487/RFC6056, January 2011, 1820 . 1822 [RFC6269] Ford, M., Ed., Boucadair, M., Durand, A., Levis, P., and 1823 P. Roberts, "Issues with IP Address Sharing", RFC 6269, 1824 DOI 10.17487/RFC6269, June 2011, 1825 . 1827 [RFC6294] Hu, Q. and B. Carpenter, "Survey of Proposed Use Cases for 1828 the IPv6 Flow Label", RFC 6294, DOI 10.17487/RFC6294, June 1829 2011, . 1831 [RFC6347] Rescorla, E. and N. Modadugu, "Datagram Transport Layer 1832 Security Version 1.2", RFC 6347, DOI 10.17487/RFC6347, 1833 January 2012, . 1835 [RFC6437] Amante, S., Carpenter, B., Jiang, S., and J. Rajahalme, 1836 "IPv6 Flow Label Specification", RFC 6437, 1837 DOI 10.17487/RFC6437, November 2011, 1838 . 1840 [RFC6438] Carpenter, B. and S. Amante, "Using the IPv6 Flow Label 1841 for Equal Cost Multipath Routing and Link Aggregation in 1842 Tunnels", RFC 6438, DOI 10.17487/RFC6438, November 2011, 1843 . 1845 [RFC7258] Farrell, S. and H. Tschofenig, "Pervasive Monitoring Is an 1846 Attack", BCP 188, RFC 7258, DOI 10.17487/RFC7258, May 1847 2014, . 1849 [RFC7525] Sheffer, Y., Holz, R., and P. Saint-Andre, 1850 "Recommendations for Secure Use of Transport Layer 1851 Security (TLS) and Datagram Transport Layer Security 1852 (DTLS)", BCP 195, RFC 7525, DOI 10.17487/RFC7525, May 1853 2015, . 1855 [RFC7567] Baker, F., Ed. and G. Fairhurst, Ed., "IETF 1856 Recommendations Regarding Active Queue Management", 1857 BCP 197, RFC 7567, DOI 10.17487/RFC7567, July 2015, 1858 . 1860 [RFC7624] Barnes, R., Schneier, B., Jennings, C., Hardie, T., 1861 Trammell, B., Huitema, C., and D. Borkmann, 1862 "Confidentiality in the Face of Pervasive Surveillance: A 1863 Threat Model and Problem Statement", RFC 7624, 1864 DOI 10.17487/RFC7624, August 2015, 1865 . 1867 [RFC7872] Gont, F., Linkova, J., Chown, T., and W. Liu, 1868 "Observations on the Dropping of Packets with IPv6 1869 Extension Headers in the Real World", RFC 7872, 1870 DOI 10.17487/RFC7872, June 2016, 1871 . 1873 [RFC7928] Kuhn, N., Ed., Natarajan, P., Ed., Khademi, N., Ed., and 1874 D. Ros, "Characterization Guidelines for Active Queue 1875 Management (AQM)", RFC 7928, DOI 10.17487/RFC7928, July 1876 2016, . 1878 [RFC7983] Petit-Huguenin, M. and G. Salgueiro, "Multiplexing Scheme 1879 Updates for Secure Real-time Transport Protocol (SRTP) 1880 Extension for Datagram Transport Layer Security (DTLS)", 1881 RFC 7983, DOI 10.17487/RFC7983, September 2016, 1882 . 1884 [RFC8033] Pan, R., Natarajan, P., Baker, F., and G. White, 1885 "Proportional Integral Controller Enhanced (PIE): A 1886 Lightweight Control Scheme to Address the Bufferbloat 1887 Problem", RFC 8033, DOI 10.17487/RFC8033, February 2017, 1888 . 1890 [RFC8084] Fairhurst, G., "Network Transport Circuit Breakers", 1891 BCP 208, RFC 8084, DOI 10.17487/RFC8084, March 2017, 1892 . 1894 [RFC8085] Eggert, L., Fairhurst, G., and G. Shepherd, "UDP Usage 1895 Guidelines", BCP 145, RFC 8085, DOI 10.17487/RFC8085, 1896 March 2017, . 1898 [RFC8086] Yong, L., Ed., Crabbe, E., Xu, X., and T. Herbert, "GRE- 1899 in-UDP Encapsulation", RFC 8086, DOI 10.17487/RFC8086, 1900 March 2017, . 1902 [RFC8087] Fairhurst, G. and M. Welzl, "The Benefits of Using 1903 Explicit Congestion Notification (ECN)", RFC 8087, 1904 DOI 10.17487/RFC8087, March 2017, 1905 . 1907 [RFC8095] Fairhurst, G., Ed., Trammell, B., Ed., and M. Kuehlewind, 1908 Ed., "Services Provided by IETF Transport Protocols and 1909 Congestion Control Mechanisms", RFC 8095, 1910 DOI 10.17487/RFC8095, March 2017, 1911 . 1913 [RFC8250] Elkins, N., Hamilton, R., and M. Ackermann, "IPv6 1914 Performance and Diagnostic Metrics (PDM) Destination 1915 Option", RFC 8250, DOI 10.17487/RFC8250, September 2017, 1916 . 1918 [RFC8289] Nichols, K., Jacobson, V., McGregor, A., Ed., and J. 1919 Iyengar, Ed., "Controlled Delay Active Queue Management", 1920 RFC 8289, DOI 10.17487/RFC8289, January 2018, 1921 . 1923 [RFC8290] Hoeiland-Joergensen, T., McKenney, P., Taht, D., Gettys, 1924 J., and E. Dumazet, "The Flow Queue CoDel Packet Scheduler 1925 and Active Queue Management Algorithm", RFC 8290, 1926 DOI 10.17487/RFC8290, January 2018, 1927 . 1929 [RFC8404] Moriarty, K., Ed. and A. Morton, Ed., "Effects of 1930 Pervasive Encryption on Operators", RFC 8404, 1931 DOI 10.17487/RFC8404, July 2018, 1932 . 1934 [RFC8446] Rescorla, E., "The Transport Layer Security (TLS) Protocol 1935 Version 1.3", RFC 8446, DOI 10.17487/RFC8446, August 2018, 1936 . 1938 Appendix A. Revision information 1940 -00 This is an individual draft for the IETF community. 1942 -01 This draft was a result of walking away from the text for a few 1943 days and then reorganising the content. 1945 -02 This draft fixes textual errors. 1947 -03 This draft follows feedback from people reading this draft. 1949 -04 This adds an additional contributor and includes significant 1950 reworking to ready this for review by the wider IETF community Colin 1951 Perkins joined the author list. 1953 Comments from the community are welcome on the text and 1954 recommendations. 1956 -05 Corrections received and helpful inputs from Mohamed Boucadair. 1958 -06 Updated following comments from Stephen Farrell, and feedback via 1959 email. Added a draft conclusion section to sketch some strawman 1960 scenarios that could emerge. 1962 -07 Updated following comments from Al Morton, Chris Seal, and other 1963 feedback via email. 1965 -08 Updated to address comments sent to the TSVWG mailing list by 1966 Kathleen Moriarty (on 08/05/2018 and 17/05/2018), Joe Touch on 1967 11/05/2018, and Spencer Dawkins. 1969 -09 Updated security considerations. 1971 -10 Updated references, split the Introduction, and added a paragraph 1972 giving some examples of why ossification has been an issue. 1974 -01 This resolved some reference issues. Updated section on 1975 observation by devices on the path. 1977 -02 Comments received from Kyle Rose, Spencer Dawkins and Tom 1978 Herbert. The network-layer information has also been re-organised 1979 after comments at IETF-103. 1981 -03 Added a section on header compression and rewriting of sections 1982 referring to RTP transport. This version contains author editorial 1983 work and removed duplicate section. 1985 -04 Revised following SecDir Review 1986 o Added some text on TLS story (additional input sought on relevant 1987 considerations). 1989 o Section 2, paragraph 8 - changed to be clearer, in particular, 1990 added "Encryption with secure key distribution prevents" 1992 o Flow label description rewritten based on PDS/BCP RFCs. 1994 o Clarify requirements from RFCs concerning the IPv6 flow label and 1995 highlight ways it can be used with encryption. (section 3.1.3) 1997 o Add text on the explicit spin-bit work in the QUIC DT. Added 1998 greasing of spin-bit. (Section 6.1) 2000 o Updated section 6 and added more explanation of impact on 2001 operators. 2003 o Other comments addressed. 2005 Authors' Addresses 2007 Godred Fairhurst 2008 University of Aberdeen 2009 Department of Engineering 2010 Fraser Noble Building 2011 Aberdeen AB24 3UE 2012 Scotland 2014 EMail: gorry@erg.abdn.ac.uk 2015 URI: http://www.erg.abdn.ac.uk/ 2017 Colin Perkins 2018 University of Glasgow 2019 School of Computing Science 2020 Glasgow G12 8QQ 2021 Scotland 2023 EMail: csp@csperkins.org 2024 URI: https://csperkins.org//