idnits 2.17.00 (12 Aug 2021) /tmp/idnits54656/draft-ietf-tsvwg-nqb-10.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- == The 'Updates: ' line in the draft header should list only the _numbers_ of the RFCs which will be updated by this document (if approved); it should not include the word 'RFC' in the list. Miscellaneous warnings: ---------------------------------------------------------------------------- -- The document date (4 March 2022) is 71 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) ** Downref: Normative reference to an Informational RFC: RFC 2983 == Outdated reference: A later version (-06) exists of draft-briscoe-docsis-q-protection-02 == Outdated reference: A later version (-23) exists of draft-ietf-tsvwg-aqm-dualq-coupled-21 == Outdated reference: A later version (-02) exists of draft-ietf-tsvwg-dscp-considerations-01 == Outdated reference: A later version (-25) exists of draft-ietf-tsvwg-ecn-l4s-id-24 == Outdated reference: A later version (-17) exists of draft-ietf-tsvwg-l4s-arch-16 Summary: 1 error (**), 0 flaws (~~), 6 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Transport Area Working Group G. White 3 Internet-Draft CableLabs 4 Updates: rfc8325 (if approved) T. Fossati 5 Intended status: Standards Track ARM 6 Expires: 5 September 2022 4 March 2022 8 A Non-Queue-Building Per-Hop Behavior (NQB PHB) for Differentiated 9 Services 10 draft-ietf-tsvwg-nqb-10 12 Abstract 14 This document specifies properties and characteristics of a Non- 15 Queue-Building Per-Hop Behavior (NQB PHB). The purpose of this NQB 16 PHB is to provide a separate queue that enables smooth, low-data- 17 rate, application-limited traffic flows, which would ordinarily share 18 a queue with bursty and capacity-seeking traffic, to avoid the 19 latency, latency variation and loss caused by such traffic. This PHB 20 is implemented without prioritization and without rate policing, 21 making it suitable for environments where the use of either these 22 features may be restricted. The NQB PHB has been developed primarily 23 for use by access network segments, where queuing delays and queuing 24 loss caused by Queue-Building protocols are manifested, but its use 25 is not limited to such segments. In particular, applications to 26 cable broadband links, Wi-Fi links, and mobile network radio and core 27 segments are discussed. This document recommends a specific 28 Differentiated Services Code Point (DSCP) to identify Non-Queue- 29 Building flows. 31 Status of This Memo 33 This Internet-Draft is submitted in full conformance with the 34 provisions of BCP 78 and BCP 79. 36 Internet-Drafts are working documents of the Internet Engineering 37 Task Force (IETF). Note that other groups may also distribute 38 working documents as Internet-Drafts. The list of current Internet- 39 Drafts is at https://datatracker.ietf.org/drafts/current/. 41 Internet-Drafts are draft documents valid for a maximum of six months 42 and may be updated, replaced, or obsoleted by other documents at any 43 time. It is inappropriate to use Internet-Drafts as reference 44 material or to cite them other than as "work in progress." 46 This Internet-Draft will expire on 5 September 2022. 48 Copyright Notice 50 Copyright (c) 2022 IETF Trust and the persons identified as the 51 document authors. All rights reserved. 53 This document is subject to BCP 78 and the IETF Trust's Legal 54 Provisions Relating to IETF Documents (https://trustee.ietf.org/ 55 license-info) in effect on the date of publication of this document. 56 Please review these documents carefully, as they describe your rights 57 and restrictions with respect to this document. Code Components 58 extracted from this document must include Revised BSD License text as 59 described in Section 4.e of the Trust Legal Provisions and are 60 provided without warranty as described in the Revised BSD License. 62 Table of Contents 64 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 65 2. Requirements Language . . . . . . . . . . . . . . . . . . . . 4 66 3. Context . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 67 3.1. Non-Queue-Building Behavior . . . . . . . . . . . . . . . 4 68 3.2. Relationship to the Diffserv Architecture . . . . . . . . 4 69 3.3. Relationship to L4S . . . . . . . . . . . . . . . . . . . 6 70 4. DSCP Marking of NQB Traffic . . . . . . . . . . . . . . . . . 6 71 4.1. Non-Queue-Building Sender Requirements . . . . . . . . . 6 72 4.2. Aggregation of the NQB DSCP with other Diffserv PHBs . . 7 73 4.3. End-to-end usage and DSCP Re-marking . . . . . . . . . . 9 74 4.3.1. Unmanaged Networks . . . . . . . . . . . . . . . . . 10 75 4.4. The NQB DSCP and Tunnels . . . . . . . . . . . . . . . . 10 76 5. Non-Queue-Building PHB Requirements . . . . . . . . . . . . . 11 77 5.1. Primary Requirements . . . . . . . . . . . . . . . . . . 11 78 5.2. Traffic Protection . . . . . . . . . . . . . . . . . . . 12 79 5.3. Guidance for Very Low Rate Links . . . . . . . . . . . . 13 80 6. Impact on Higher Layer Protocols . . . . . . . . . . . . . . 13 81 7. Configuration and Management . . . . . . . . . . . . . . . . 14 82 8. Example Use Cases . . . . . . . . . . . . . . . . . . . . . . 14 83 8.1. DOCSIS Access Networks . . . . . . . . . . . . . . . . . 14 84 8.2. Mobile Networks . . . . . . . . . . . . . . . . . . . . . 14 85 8.3. WiFi Networks . . . . . . . . . . . . . . . . . . . . . . 15 86 8.3.1. Interoperability with Existing WiFi Networks . . . . 15 87 9. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 16 88 10. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 16 89 11. Security Considerations . . . . . . . . . . . . . . . . . . . 17 90 12. References . . . . . . . . . . . . . . . . . . . . . . . . . 18 91 12.1. Normative References . . . . . . . . . . . . . . . . . . 18 92 12.2. Informative References . . . . . . . . . . . . . . . . . 18 93 Appendix A. DSCP Remarking Pathologies . . . . . . . . . . . . . 21 94 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 22 96 1. Introduction 98 This document defines a Differentiated Services per-hop behavior 99 (PHB) called "Non-Queue-Building Per-Hop Behavior" (NQB PHB), which 100 isolates traffic flows that are relatively low data rate and that do 101 not themselves materially contribute to queueing delay and loss, 102 allowing them to avoid the queuing delays and losses caused by other 103 traffic. Such Non-Queue-Building flows (for example: interactive 104 voice, gaming, machine-to-machine applications) are application 105 limited flows that are distinguished from traffic flows managed by an 106 end-to-end congestion control algorithm. 108 The vast majority of packets that are carried by broadband access 109 networks are managed by an end-to-end congestion control algorithm, 110 such as Reno, Cubic or BBR. These congestion control algorithms 111 attempt to seek the available capacity of the end-to-end path (which 112 can frequently be the access network link capacity), and in doing so 113 generally overshoot the available capacity, causing a queue to build- 114 up at the bottleneck link. This queue build up results in queuing 115 delay (variable latency) and possibly packet loss that can affect all 116 of the applications that are sharing the bottleneck link. 118 In contrast to traditional congestion-controlled applications, there 119 are a variety of relatively low data rate applications that do not 120 materially contribute to queueing delay and loss, but are nonetheless 121 subjected to it by sharing the same bottleneck link in the access 122 network. Many of these applications may be sensitive to latency or 123 latency variation, as well as packet loss, and thus produce a poor 124 quality of experience in such conditions. 126 Active Queue Management (AQM) mechanisms (such as PIE [RFC8033], 127 DOCSIS-PIE [RFC8034], or CoDel [RFC8289]) can improve the quality of 128 experience for latency sensitive applications, but there are 129 practical limits to the amount of improvement that can be achieved 130 without impacting the throughput of capacity-seeking applications. 131 For example, AQMs generally allow a significant amount of queue depth 132 variation in order to accommodate the behaviors of congestion control 133 algorithms such as Reno and Cubic. If the AQM attempted to control 134 the queue much more tightly, applications using those algorithms 135 would not perform well. Alternatively, flow queueuing systems, such 136 as fq_codel [RFC8290] can be employed to isolate flows from one 137 another, but these are not appropriate for all bottleneck links, due 138 to complexity or other reasons. 140 The NQB PHB supports differentiating between these two classes of 141 traffic in bottleneck links and queuing them separately in order that 142 both classes can deliver satisfactory quality of experience for their 143 applications. 145 To be clear, a network implementing the NQB PHB solely provides 146 isolation for traffic classified as behaving in conformance with the 147 NQB DSCP (and optionally enforces that behavior). It is the NQB 148 senders' behavior itself which results in low latency and low loss. 150 2. Requirements Language 152 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 153 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 154 "OPTIONAL" in this document are to be interpreted as described in BCP 155 14 [RFC2119] [RFC8174] when, and only when, they appear in all 156 capitals, as shown here. 158 3. Context 160 3.1. Non-Queue-Building Behavior 162 There are many applications that send traffic at relatively low data 163 rates and/or in a fairly smooth and consistent manner such that they 164 are highly unlikely to exceed the available capacity of the network 165 path between source and sink. These applications may themselves only 166 cause very small, transient queues to form in network buffers, but 167 nonetheless they can be subjected to packet delay and delay variation 168 as a result of sharing a network buffer with applications that tend 169 to cause large and/or standing queues to form. Many of these 170 applications are negatively affected by excessive packet delay and 171 delay variation. Such applications are ideal candidates to be queued 172 separately from the applications that are the cause of queue buildup, 173 latency and loss. 175 In contrast, Queue-building (QB) flows include those that use TCP or 176 QUIC, with Cubic, Reno or other TCP congestion control algorithms 177 that probe for the link capacity and induce latency and loss as a 178 result. Other types of QB flows include those that frequently send 179 at a high burst rate (e.g. several consecutive packets sent well in 180 excess of 1 Mbps) even if the long-term average data rate is much 181 lower. 183 3.2. Relationship to the Diffserv Architecture 185 The IETF has defined the Differentiated Services architecture 186 [RFC2475] with the intention that it allows traffic to be marked in a 187 manner that conveys the performance requirements of that traffic 188 either quantitatively or in a relative sense (i.e. priority). The 189 architecture defines the use of the Diffserv field [RFC2474] for this 190 purpose, and numerous RFCs have been written that describe 191 recommended interpretations of the values (Diffserv Code Points) of 192 the field, and standardized treatments (traffic conditioning and per- 193 hop-behaviors) that can be implemented to satisfy the performance 194 requirements of traffic so marked. 196 While this architecture is powerful, and can be configured to meet 197 the performance requirements of a variety of applications and traffic 198 categories, or to achieve differentiated service offerings, it has 199 proven problematic to enable its use for these purposes end-to-end 200 across the Internet. 202 This difficulty is in part due to the fact that meeting (in an end- 203 to-end context) the performance requirements of an application 204 involves all of the networks in the path agreeing on what those 205 requirements are, and sharing an interest in meeting them. In many 206 cases this is made more difficult due to the fact that the 207 performance "requirements" are not strict ones (e.g. applications 208 will degrade in some manner as loss/latency/jitter increase), so the 209 importance of meeting them for any particular application in some 210 cases involves a judgment as to the value of avoiding some amount of 211 degradation in quality for that application in exchange for an 212 increase in the degradation of another application. 214 Further, in many cases the implementation of Diffserv PHBs has 215 historically involved prioritization of service classes with respect 216 to one another, which sets up the zero-sum game alluded to in the 217 previous paragraph, and results in the need to limit access to higher 218 priority classes via mechanisms such as access control, admission 219 control, traffic conditioning and rate policing, and/or to meter and 220 bill for carriage of such traffic. These mechanisms can be difficult 221 or impossible to implement in an end-to-end context. 223 Finally, some jurisdictions impose regulations that limit the ability 224 of networks to provide differentiation of services, in large part 225 based on the belief that doing so necessarily involves prioritization 226 or privileged access to bandwidth, and thus a benefit to one class of 227 traffic always comes at the expense of another. 229 In contrast, the NQB PHB has been designed with the goal that it 230 avoids many of these issues, and thus could conceivably be deployed 231 end-to-end across the Internet. The intent of the NQB DSCP is that 232 it signals verifiable behavior that permits the sender to request 233 differentiated treatment. Also, the NQB traffic is to be given a 234 separate queue with priority equal to default traffic, and given no 235 reserved bandwidth other than the bandwidth that it shares with 236 default traffic. As a result, the NQB PHB does not aim to meet 237 specific application performance requirements. Instead the goal of 238 the NQB PHB is to provide statistically better loss, latency, and 239 jitter performance for traffic that is itself only an insignificant 240 contributor to those degradations. The PHB is also designed to 241 minimize any incentives for a sender to mismark its traffic, since 242 neither higher priority nor reserved bandwith are being offered. 243 These attributes eliminate many of the tradeoffs that underlie the 244 handling of differentiated service classes in the Diffserv 245 architecture as it has traditionally been defined. They also 246 significantly simplify access control and admission control 247 functions, reducing them to simple verification of behavior. 249 3.3. Relationship to L4S 251 The NQB DSCP and PHB described in this draft have been defined to 252 operate independently of the experimental L4S Architecture 253 [I-D.ietf-tsvwg-l4s-arch]. Nonetheless, the NQB traffic flows are 254 intended to be compatible with [I-D.ietf-tsvwg-l4s-arch], with the 255 result being that NQB traffic and L4S traffic can share the low- 256 latency queue in an L4S DualQ node 257 [I-D.ietf-tsvwg-aqm-dualq-coupled]. Compliance with the DualQ 258 Coupled AQM requirements (Section 2.5 of 259 [I-D.ietf-tsvwg-aqm-dualq-coupled]) is considered sufficient to 260 support the NQB PHB requirement of fair allocation of bandwidth 261 between the QB and NQB queues (Section 5). 263 4. DSCP Marking of NQB Traffic 265 4.1. Non-Queue-Building Sender Requirements 267 Non-queue-building (NQB) flows are typically UDP flows that don't 268 seek the maximum capacity of the link (examples: online games, voice 269 chat, DNS lookups, real-time IoT analytics data). Here the data rate 270 is limited by the application itself rather than by network capacity 271 - these applications send, at most, the equivalent of a few well- 272 spaced packets per RTT, even if the packets are not actually RTT- 273 clocked. In today's network this corresponds to an instantaneous 274 data rate (packet size divided by packet inter-arrival time) of no 275 more than about 1 Mbps (e.g. no more than one 1250 B packet every 10 276 ms), but there is no precise bound since it depends on the conditions 277 in which the application is operating. 279 Note that, while such flows ordinarily don't implement a traditional 280 congestion control mechanism, they nonetheless are expected to comply 281 with existing guidance for safe deployment on the Internet, for 282 example the requirements in [RFC8085] and Section 2 of [RFC3551] 283 (also see the circuit breaker limits in Section 4.3 of [RFC8083] and 284 the description of inelastic pseudowires in Section 4 of [RFC7893]). 285 To be clear, the description of NQB flows in this document should not 286 be interpreted as suggesting that such flows are in any way exempt 287 from this responsibility. 289 Applications that align with the description of NQB behavior in the 290 preceding paragraphs SHOULD identify themselves to the network using 291 a Diffserv Code Point (DSCP) of 45 (decimal) so that their packets 292 can be queued separately from QB flows. The choice of the value 45 293 is motivated in part by the desire to achieve separate queuing in 294 existing WiFi networks (see Section 8.3) and by the desire to make 295 implementation of the PHB simpler in network gear that has the 296 ability to classify traffic based on ranges of DSCP value (see 297 Section 4.2 for further discussion). In networks where another (e.g. 298 a local-use) codepoint is designated for NQB traffic, or where 299 specialized PHBs are available that can meet specific application 300 requirements (e.g. a guaranteed-latency path for voice traffic), it 301 may be preferred to use another DSCP. In end systems where the 302 choice of using DSCP 45 is not available to the application, the CS5 303 DSCP (40 decimal) could be used as a fallback. See Section 4.2 for 304 rationale as to why this choice could be fruitful. 306 If the application's traffic exceeds more than a few packets per RTT, 307 or exceeds a fairly small fraction of the expected path capacity on 308 an instantaneous (e.g. inter-packet or a suitably short time 309 interval) basis, the application SHOULD NOT mark its traffic with the 310 NQB DSCP. In such a case, the application could instead consider 311 implementing a low latency congestion control mechanism as described 312 in [I-D.ietf-tsvwg-ecn-l4s-id]. At the time of writing, it is 313 believed that 1 Mbps is a reasonable upper bound on instantaneous 314 traffic rate for an NQB-marked application, but this value is of 315 course subject to the context in which the application is expected to 316 be deployed. 318 An application that marks its traffic as NQB but happens to exceed 319 the available path capacity (even on an instantaneous basis) runs the 320 risk of being subjected to a Traffic Protection algorithm (see 321 Section 5.2), which could result in the excess traffic being 322 discarded or queued separately as default traffic (and thus 323 potentially delivered out of order). As a result, applications that 324 aren't clearly beneath the threshold described above would need to 325 weigh the risk of additional loss or out-of-order delivery against 326 the expected latency benefits of NQB treatment in determining whether 327 or not to mark their packets as NQB. 329 4.2. Aggregation of the NQB DSCP with other Diffserv PHBs 331 It is RECOMMENDED that networks and nodes that do not support the NQB 332 PHB be configured to treat NQB marked traffic the same as traffic 333 marked "Default". It is additionally RECOMMENDED that such networks 334 and nodes simply classify the NQB DSCP into the same treatment 335 aggregate as Default traffic, or encapsulate the NQB marked packet, 336 rather than re-marking NQB traffic as Default. This preservation of 337 the NQB marking enables hops further along the path to provide the 338 NQB PHB successfully. Section 4.3 discusses re-marking of NQB 339 traffic to an alternate DSCP value of 5 within core networks as a 340 means to facilitate this for networks where this behavior can be more 341 readily implemented for such a value. 343 In backbone and core network switches (particularly if shallow- 344 buffered), and nodes that do not typically experience congestion, 345 treating NQB marked traffic the same as Default may be sufficient to 346 preserve loss/latency/jitter performance for NQB traffic. In other 347 nodes, treating NQB marked traffic as Default could result in 348 degradation of loss/latency/jitter performance but is recommended 349 nonetheless in order to preserve the incentives described in 350 Section 5. An alternative, in controlled environments where there is 351 no risk of mismarking of traffic, would be to aggregate NQB marked 352 traffic with real-time, latency sensitive traffic. Similarly, 353 networks and nodes that aggregate service classes as discussed in 354 [RFC5127] and [RFC8100] may not be able to provide a PDB/PHB that 355 meets the requirements of this document. In these cases it is 356 RECOMMENDED that NQB-marked traffic be aggregated into the Elastic 357 Treatment Aggregate (for [RFC5127] networks) or the Default / Elastic 358 Treatment Aggregate (for [RFC8100] networks), although in some cases 359 a network operator may instead choose to aggregate NQB traffic into 360 the (Bulk) Real-Time Treatment Aggregate. Either approach comes with 361 trade-offs: when the aggregated traffic encounters a bottleneck, 362 aggregating with Default/Elastic traffic could result in a 363 degradation of loss/latency/jitter performance for NQB traffic, while 364 aggregating with Real-Time (assuming such traffic is provided a 365 prioritized PHB) risks creating an incentive for mismarking of non- 366 compliant traffic as NQB (except in controlled environments). In 367 either case, the NQB DSCP SHOULD be preserved (possibly via 368 encapsulation) in order to limit the negative impact that such 369 networks would have on end-to-end performance for NQB traffic. This 370 aligns with recommendations in [RFC5127]. 372 Nodes that support the NQB PHB may choose to aggregate other service 373 classes into the NQB queue. This is particularly useful in cases 374 where specialized PHBs for these other service classes are not 375 provided. Candidate service classes for this aggregation would 376 include those that carry inelastic traffic that has low to very-low 377 tolerance for loss, latency and/or jitter as discussed in [RFC4594]. 378 These could include Telephony (EF/VA), Signaling (CS5), Real-Time 379 Interactive (CS4) and Broadcast Video (CS3). Or, in some networks, 380 equipment limitations may necessitate aggregating all traffic marked 381 with DSCPs 40-47 (i.e., whose three MSBs are 0b101). As noted in 382 Section 4.1, the choice of the value 45 is motivated in part by the 383 desire to make this aggregation simpler in network equipment that can 384 classify packets via comparing the DSCP value to a range of 385 configured values. 387 4.3. End-to-end usage and DSCP Re-marking 389 In contrast to some existing standard PHBs, many of which are 390 typically only meaningful within a Diffserv Domain (e.g. an AS or an 391 enterprise network), this PHB is expected to be used end-to-end 392 across the Internet, wherever suitable operator agreements apply. 393 Under the [RFC2474] model, this requires that the corresponding DSCP 394 is recognized by all operators and mapped across their boundaries 395 accordingly. 397 To support NQB, networks MUST preserve a DSCP marking distinction 398 between NQB traffic and Default traffic when forwarding via an 399 interconnect from or to another network. To facilitate the default 400 treatment of NQB traffic in backbones and core networks discussed in 401 the previous section (where IP Precedence may be deployed), networks 402 that support NQB SHOULD NOT use the value 45 for NQB at network 403 interconnects unless that usage is explicitly documented in the TCA 404 (Traffic Conditioning Agreement, see [RFC2475]) for that 405 interconnection. Rather, networks SHOULD remap NQB traffic to DSCP 5 406 prior to interconnection, unless agreed otherwise between the 407 interconnecting partners. To be clear, interconnecting networks are 408 not precluded from negotiating (via an SLA, TCA, or some other 409 agreement) a different DSCP to use to signal NQB across an 410 interconnect. Additionally, the fact that this PHB is intended for 411 end-to-end usage does not preclude networks from mapping the NQB DSCP 412 to a value other than 45 or 5 for internal usage, as long as the 413 appropriate NQB DSCP is restored when forwarding to another network. 415 Furthermore, in other network environments where IP Precedence 416 ([RFC0791]) is deployed, it is RECOMMENDED that the network operator 417 re-mark NQB traffic to DSCP 5 in order to ensure that it is 418 aggregated with Default traffic. 420 In order to enable interoperability with WiFi equipment as described 421 in Section 8.3.1, networks SHOULD re-mark NQB traffic (e.g. DSCP 5) 422 to DSCP 45 prior to a customer access link, subject to the safeguards 423 described below and in that section. 425 Thus, this document recommends two DSCPs to designate NQB, the value 426 45 for use by hosts and in WiFi networks, and the value 5 for use 427 across network interconnections. 429 4.3.1. Unmanaged Networks 431 In cases where a network operator is delivering traffic into an 432 unmanaged network outside of their control (e.g. a residential ISP 433 delivering traffic to a customer's home network), the network 434 operator should presume that the existing network equipment in the 435 user network supports the WiFi default DSCP mapping (see Section 8.3) 436 and/or IP Precedence, and does not support the safeguards that are 437 provided by the NQB PHB requirements. As a result, the network 438 operator should take precautions to prevent issues. When the data 439 rate of the access network segment is less than the expected data 440 rate of the user network, this is unlikely to be an issue. However, 441 if the access network rate exceeds the expected rate of the user 442 network, the operator SHOULD deploy a policing function on NQB marked 443 traffic that minimizes the potential for negative impacts on traffic 444 marked Default, for example by limiting the rate of such traffic to a 445 set fraction of the customer's service rate, with excess traffic 446 either dropped or re-marked as Default. 448 As an additional safeguard, and to prevent the inadvertent 449 introduction of problematic traffic into unmanaged user networks, 450 network equipment that is intended to deliver traffic into unmanaged 451 user networks (e.g. an access network gateway for a residential ISP) 452 MUST by default ensure that NQB traffic is marked with a DSCP that 453 selects the WiFi "Best Effort" Access Category and the "Routine" IP 454 Precedence level (i.e. DSCP 0-7). Such equipment MUST support the 455 ability to configure the remapping, so that (when appropriate 456 safeguards are in place) traffic can be delivered as NQB-marked. 458 4.4. The NQB DSCP and Tunnels 460 [RFC2983] discusses tunnel models that support Diffserv. It 461 describes a "uniform model" in which the inner DSCP is copied to the 462 outer header at encapsulation, and the outer DSCP is copied to the 463 inner header at decapsulation. It also describes a "pipe model" in 464 which the outer DSCP is not copied to the inner header at 465 decapsulation. Both models can be used in conjunction with the NQB 466 PHB. In the case of the pipe model, any DSCP manipulation (re- 467 marking) of the outer header by intermediate nodes would be discarded 468 at tunnel egress, potentially improving the possibility of achieving 469 NQB treatment in subsequent nodes. 471 As is discussed in [RFC2983], tunnel protocols that are sensitive to 472 reordering can result in undesirable interactions if multiple DSCP 473 PHBs are signaled for traffic within a tunnel instance. This is true 474 for NQB marked traffic as well. If a tunnel contains a mix of QB and 475 NQB traffic, and this is reflected in the outer DSCP in a network 476 that supports the NQB PHB, it would be necessary to avoid a 477 reordering-sensitive tunnel protocol. 479 5. Non-Queue-Building PHB Requirements 481 It is worthwhile to note again that the NQB designation and marking 482 is intended to convey verifiable traffic behavior, as opposed to 483 simply a desire for differentiated treatment. Also, it is important 484 that incentives are aligned correctly, i.e. that there is a benefit 485 to the application in marking its packets correctly, and a 486 disadvantage (or at least no benefit) to an application in 487 intentionally mismarking its traffic. Thus, a useful property of 488 nodes (i.e. network switches and routers) that support separate 489 queues for NQB and QB flows is that for NQB flows, the NQB queue 490 provides better performance than the QB queue; and for QB flows, the 491 QB queue provides better performance than the NQB queue (this is 492 discussed further in this section and Section 11). By adhering to 493 these principles, there is no incentive for senders to mismark their 494 traffic as NQB, and further, any mismarking can be identified by the 495 network. 497 5.1. Primary Requirements 499 A node supporting the NQB PHB makes no guarantees on latency or data 500 rate for NQB marked flows, but instead aims to provide a bound on 501 queuing delay for as many such marked flows as it can, and shed load 502 when needed. 504 A node supporting the NQB PHB MUST provide a queue for non-queue- 505 building traffic separate from any queue used for queue-building 506 traffic. 508 NQB traffic, in aggregate, SHOULD NOT be rate limited or rate policed 509 separately from queue-building traffic of equivalent importance. 511 The NQB queue SHOULD be given equivalent forwarding preference 512 compared to queue-building traffic of equivalent importance. The 513 node SHOULD provide a scheduler that allows QB and NQB traffic of 514 equivalent importance to share the link in a fair manner, e.g. a 515 deficit round-robin scheduler with equal weights. Compliance with 516 these recommendations helps to ensure that there are no incentives 517 for QB traffic to be mismarked as NQB. In environments where 518 mismarking is not a potential issue (e.g. a network where a marking 519 policy is enforced by other means), these requirements may not be 520 necessary. 522 A node supporting the NQB PHB SHOULD treat traffic marked as Default 523 (DSCP=0) as QB traffic having equivalent importance to the NQB marked 524 traffic. A node supporting the NQB DSCP MUST support the ability to 525 configure the classification criteria that are used to identify QB 526 and NQB traffic of equivalent importance. 528 The NQB queue SHOULD have a buffer size that is significantly smaller 529 than the buffer provided for QB traffic (e.g. single-digit 530 milliseconds). It is expected that most QB traffic is engineered to 531 work well when the network provides a relatively deep buffer (e.g. on 532 the order of tens or hundreds of ms) in nodes where support for the 533 NQB PHB is advantageous (i.e. bottleneck nodes). Providing a 534 similarly deep buffer for the NQB queue would be at cross purposes to 535 providing very low queueing delay, and would erode the incentives for 536 QB traffic to be marked correctly. 538 5.2. Traffic Protection 540 It is possible that due to an implementation error or 541 misconfiguration, a QB flow would end up getting mismarked as NQB, or 542 vice versa. In the case of an NQB flow that isn't marked as NQB and 543 ends up in the QB queue, it would only impact its own quality of 544 service, and so it seems to be of lesser concern. However, a QB flow 545 that is mismarked as NQB would cause queuing delays and/or loss for 546 all of the other flows that are sharing the NQB queue. 548 To prevent this situation from harming the performance of the real 549 NQB flows, network elements that support differentiating NQB traffic 550 SHOULD support a "traffic protection" function that can identify QB 551 flows that are mismarked as NQB, and either reclassify those flows/ 552 packets to the QB queue or discard the offending traffic. Such a 553 function SHOULD be implemented in an objective and verifiable manner, 554 basing its decisions upon the behavior of the flow rather than on 555 application-layer constructs. It is RECOMMENDED that traffic 556 protection algorithms base their decisions on the detection of actual 557 queuing, as opposed to simply packet arrival rates. It may be 558 advantageous for a traffic protection function to employ hysteresis 559 to prevent borderline flows from being reclassified capriciously. 561 One example traffic protection algorithm can be found in 562 [I-D.briscoe-docsis-q-protection]. 564 There are some situations where such function may not be necessary. 565 For example, a network element designed for use in controlled 566 environments (e.g. enterprise LAN) may not require a traffic 567 protection function. Additionally, some networks may prefer to 568 police the application of the NQB DSCP at the ingress edge, so that 569 in-network traffic protection is not needed. 571 5.3. Guidance for Very Low Rate Links 573 The NQB sender requirements in Section 4.1 place responsibility in 574 the hands of the application developer to determine the likelihood 575 that the application's sending behavior could result in a queue 576 forming along the path. These requirements rely on application 577 developers having a reasonable sense for the network context in which 578 their application is to be deployed. Even so, there will undoubtedly 579 be networks that contain links having a data rate that is below the 580 lower end of what is considered "typical", and some of these links 581 may even be below the instantaneous sending rate of some NQB-marked 582 applications. 584 To limit the consequences of this scenario, operators of such 585 networks SHOULD utilize a traffic protection function that is more 586 tolerant of burstiness (i.e. a temporary queue). Alternatively, 587 operators of such networks MAY choose to disable NQB support on these 588 low speed links. In particular, for links that are far below 589 "typical" path rates, it is RECOMMENDED that NQB support be disabled. 591 6. Impact on Higher Layer Protocols 593 Network elements that support the NQB PHB and that support traffic 594 protection as discussed in the previous section introduce the 595 possibility that flows classified into the NQB queue could experience 596 out of order delivery or packet loss if their behavior is not 597 consistent with NQB. This is particularly true if the traffic 598 protection algorithm makes decisions on a packet-by-packet basis. In 599 this scenario, a flow that is (mis)marked as NQB and that causes a 600 queue to form in this bottleneck link could see some of its packets 601 forwarded by the NQB queue, and some of them either discarded or 602 redirected to the QB queue. In the case of redirection, depending on 603 the queueing latency and scheduling within the network element, this 604 could result in packets being delivered out of order. As a result, 605 the use of the NQB DSCP by a higher layer protocol carries some risk 606 that an increased amount of out of order delivery or packet loss will 607 be experienced. This characteristic provides one disincentive for 608 mis-marking of traffic. 610 7. Configuration and Management 612 As required above, nodes supporting the NQB PHB provide for the 613 configuration of classifiers that can be used to differentiate 614 between QB and NQB traffic of equivalent importance. The default for 615 such classifiers is recommended to be the assigned NQB DSCP (to 616 identify NQB traffic) and the Default (0) DSCP (to identify QB 617 traffic). 619 8. Example Use Cases 621 8.1. DOCSIS Access Networks 623 Residential cable broadband Internet services are commonly configured 624 with a single bottleneck link (the access network link) upon which 625 the service definition is applied. The service definition, typically 626 an upstream/downstream data rate tuple, is implemented as a 627 configured pair of rate shapers that are applied to the user's 628 traffic. In such networks, the quality of service that each 629 application receives, and as a result, the quality of experience that 630 it generates for the user is influenced by the characteristics of the 631 access network link. 633 To support the NQB PHB, cable broadband services MUST be configured 634 to provide a separate queue for NQB marked traffic. The NQB queue 635 MUST be configured to share the service's rate shaped bandwidth with 636 the queue for QB traffic. 638 8.2. Mobile Networks 640 Historically, 3GPP mobile networks have utilised "bearers" to 641 encapsulate each user's user plane traffic through the radio and core 642 networks. A "dedicated bearer" may be allocated a Quality of Service 643 (QoS) to apply any prioritisation to its flows at queues and radio 644 schedulers. Typically an LTE operator provides a dedicated bearer 645 for IMS VoLTE (Voice over LTE) traffic, which is prioritised in order 646 to meet regulatory obligations for call completion rates; and a "best 647 effort" default bearer, for Internet traffic. The "best effort" 648 bearer provides no guarantees, and hence its buffering 649 characteristics are not compatible with low-latency traffic. The 5G 650 radio and core systems offer more flexibility over bearer allocation, 651 meaning bearers can be allocated per traffic type (e.g. loss- 652 tolerant, low-latency etc.) and hence support more suitable treatment 653 of Internet real-time flows. 655 To support the NQB PHB, the mobile network SHOULD be configured to 656 give UEs a dedicated, low-latency, non-GBR, EPS bearer, e.g. one with 657 QCI 7, in addition to the default EPS bearer; or a Data Radio Bearer 658 with 5QI 7 in a 5G system (see Table 5.7.4-1: Standardized 5QI to QoS 659 characteristics mapping in [SA-5G]). 661 A packet carrying the NQB DSCP SHOULD be routed through the dedicated 662 low-latency EPS bearer. A packet that has no associated NQB marking 663 SHOULD NOT be routed through the dedicated low-latency EPS bearer. 665 8.3. WiFi Networks 667 WiFi networking equipment compliant with 802.11e/n/ac/ax [IEEE802-11] 668 generally supports either four or eight transmit queues and four sets 669 of associated Enhanced Multimedia Distributed Control Access (EDCA) 670 parameters (corresponding to the four WiFi Multimedia (WMM) Access 671 Categories) that are used to enable differentiated media access 672 characteristics. As discussed in [RFC8325], most existing WiFi 673 implementations use a default DSCP to User Priority mapping that 674 utilizes the most significant three bits of the Diffserv Field to 675 select "User Priority" which is then mapped to the four WMM Access 676 Categories. [RFC8325] also provides an alternative mapping that more 677 closely aligns with the DSCP recommendations provided by the IETF. 679 In addition to the requirements provided in other sections of this 680 document, to support the NQB PHB, WiFi equipment (including equipment 681 compliant with [RFC8325]) SHOULD map the NQB codepoint 45 into a 682 separate queue in the same Access Category as the queue that carries 683 default traffic (i.e. the Best Effort Access Category). 685 8.3.1. Interoperability with Existing WiFi Networks 687 While some existing WiFi equipment may be capable (in some cases via 688 firmware update) of supporting the NQB PHB requirements, many 689 currently deployed devices cannot be configured in this way. As a 690 result the remainder of this section discusses interoperability with 691 these existing WiFi networks, as opposed to PHB compliance. 693 In order to increase the likelihood that NQB traffic is provided a 694 separate queue from QB traffic in existing WiFi equipment that uses 695 the default mapping, the 45 code point is recommended for NQB. This 696 maps NQB to UP_5 which is in the "Video" Access Category. While this 697 DSCP to User Priority mapping enables these WiFi systems to support 698 the NQB PHB requirement for segregated queuing, it does not support 699 the remaining NQB PHB requirements in Section 5. The ramifications 700 of, and remedies for this are discussed further below. 702 Existing WiFi devices are unlikely to support a traffic protection 703 algorithm, so traffic mismarked as NQB is not likely to be detected 704 and remedied by such devices. 706 Furthermore, in their default configuration, existing WiFi devices 707 utilize EDCA parameters that result in statistical prioritization of 708 the "Video" Access Category above the "Best Effort" Access Category. 709 If left unchanged, this would violate the NQB PHB requirement for 710 equal prioritization, and could erode the principle of alignment of 711 incentives. In order to preserve the incentives principle for NQB, 712 WiFi systems SHOULD be configured such that the EDCA parameters for 713 the Video Access Category match those of the Best Effort Access 714 Category. This recommendation is primarily applicable to WiFi 715 systems deployed in a managed environment or those deployed by an 716 ISP, and is intended for situations when the vast majority of traffic 717 that would use AC_VI is NQB. In other situations (e.g. consumer- 718 grade WiFi gear deployed by an ISP's customer) this configuration may 719 not be possible, and the requirements and recommendations in 720 Section 4.3.1 would apply. 722 Similarly, systems that utilize [RFC8325] but that are unable to 723 fully support the PHB requirements, SHOULD map the recommended NQB 724 code point 45 (or the locally determined alternative) to UP_5 in the 725 "Video" Access Category. 727 9. Acknowledgements 729 Thanks to Diego Lopez, Stuart Cheshire, Brian Carpenter, Bob Briscoe, 730 Greg Skinner, Toke Hoeiland-Joergensen, Luca Muscariello, David 731 Black, Sebastian Moeller, Ruediger Geib, Jerome Henry, Steven Blake, 732 Jonathan Morton, Roland Bless, Kevin Smith, Martin Dolly, and Kyle 733 Rose for their review comments. Thanks also to Gorry Fairhurst, Ana 734 Custura, and Ruediger Geib for their input on selection of 735 appropriate DSCPs. 737 10. IANA Considerations 739 This document requests that IANA assign the Differentiated Services 740 Field Codepoint (DSCP) 45 ('0b101101', 0x2D) from the "Differentiated 741 Services Field Codepoints (DSCP)" registry 742 (https://www.iana.org/assignments/dscp-registry/) ("DSCP Pool 3 743 Codepoints", Codepoint Space xxxx01, Standards Action) as the 744 RECOMMENDED codepoint for Non-Queue-Building behavior for end-systems 745 and in edge networks. 747 IANA should update this registry as follows: 749 * Name: NQB-EDGE 750 * Value (Binary): 101101 752 * Value (Decimal): 45 754 * Reference: this document 756 This document requests that IANA assign the Differentiated Services 757 Field Codepoint (DSCP) 5 ('0b000101', 0x05) from the "Differentiated 758 Services Field Codepoints (DSCP)" registry 759 (https://www.iana.org/assignments/dscp-registry/) ("DSCP Pool 3 760 Codepoints", Codepoint Space xxxx01, Standards Action) as the 761 RECOMMENDED codepoint for Non-Queue-Building behavior in core 762 networks and interconnections. 764 IANA should update this registry as follows: 766 * Name: NQB-CORE 768 * Value (Binary): 000101 770 * Value (Decimal): 5 772 * Reference: this document 774 11. Security Considerations 776 When the NQB PHB is fully supported in bottleneck links, there is no 777 incentive for a queue-building application to mismark its packets as 778 NQB (or vice versa). If a queue-building flow were to mark its 779 packets as NQB, it would be unlikely to receive a benefit by doing 780 so, and it could experience excessive packet loss, excessive latency 781 variation and/or excessive out-of-order delivery (depending on the 782 nature of the traffic protection function). If a non-queue-building 783 flow were to fail to mark its packets as NQB, it could suffer the 784 latency and loss typical of sharing a queue with capacity seeking 785 traffic. 787 In order to preserve low latency performance for NQB traffic, 788 networks that support the NQB PHB will need to ensure that mechanisms 789 are in place to prevent malicious NQB-marked traffic from causing 790 excessive queue delays. This document recommends the implementation 791 of a traffic protection mechanism to achieve this goal, but 792 recognizes that other options may be more desirable in certain 793 situations. 795 Notwithstanding the above, the choice of DSCP for NQB does allow 796 existing WiFi networks to readily (and by default) support some of 797 the PHB requirements, but without a traffic protection function, and 798 (when left in the default state) by giving NQB traffic higher 799 priority than QB traffic. This does open up the NQB marking to 800 potential abuse on these WiFi links, but since these existing WiFi 801 networks already give one quarter of the DSCP space this same 802 treatment, and further they give another quarter of the DSCP space 803 even higher priority, the NQB DSCP does not seem to be of any greater 804 risk for abuse than these others. 806 The NQB signal is not integrity protected and could be flipped by an 807 on-path attacker. This might negatively affect the QoS of the 808 tampered flow. 810 12. References 812 12.1. Normative References 814 [RFC0791] Postel, J., "Internet Protocol", STD 5, RFC 791, 815 DOI 10.17487/RFC0791, September 1981, 816 . 818 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 819 Requirement Levels", BCP 14, RFC 2119, 820 DOI 10.17487/RFC2119, March 1997, 821 . 823 [RFC2474] Nichols, K., Blake, S., Baker, F., and D. Black, 824 "Definition of the Differentiated Services Field (DS 825 Field) in the IPv4 and IPv6 Headers", RFC 2474, 826 DOI 10.17487/RFC2474, December 1998, 827 . 829 [RFC2983] Black, D., "Differentiated Services and Tunnels", 830 RFC 2983, DOI 10.17487/RFC2983, October 2000, 831 . 833 [RFC8085] Eggert, L., Fairhurst, G., and G. Shepherd, "UDP Usage 834 Guidelines", BCP 145, RFC 8085, DOI 10.17487/RFC8085, 835 March 2017, . 837 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 838 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 839 May 2017, . 841 [RFC8325] Szigeti, T., Henry, J., and F. Baker, "Mapping Diffserv to 842 IEEE 802.11", RFC 8325, DOI 10.17487/RFC8325, February 843 2018, . 845 12.2. Informative References 847 [Barik] Barik, R., Welzl, M., Elmokashfi, A., Dreibholz, T., and 848 S. Gjessing, "Can WebRTC QoS Work? A DSCP Measurement 849 Study", ITC 30, September 2018. 851 [Custura] Custura, A., Venne, A., and G. Fairhurst, "Exploring DSCP 852 modification pathologies in mobile edge networks", TMA , 853 2017. 855 [I-D.briscoe-docsis-q-protection] 856 Briscoe, B. and G. White, "The DOCSIS(r) Queue Protection 857 Algorithm to Preserve Low Latency", Work in Progress, 858 Internet-Draft, draft-briscoe-docsis-q-protection-02, 31 859 January 2022, . 862 [I-D.ietf-tsvwg-aqm-dualq-coupled] 863 Schepper, K. D., Briscoe, B., and G. White, "DualQ Coupled 864 AQMs for Low Latency, Low Loss and Scalable Throughput 865 (L4S)", Work in Progress, Internet-Draft, draft-ietf- 866 tsvwg-aqm-dualq-coupled-21, 1 February 2022, 867 . 870 [I-D.ietf-tsvwg-dscp-considerations] 871 Custura, A., Fairhurst, G., and R. Secchi, "Considerations 872 for Assigning a new Recommended DiffServ Codepoint 873 (DSCP)", Work in Progress, Internet-Draft, draft-ietf- 874 tsvwg-dscp-considerations-01, 24 January 2022, 875 . 878 [I-D.ietf-tsvwg-ecn-l4s-id] 879 Schepper, K. D. and B. Briscoe, "Explicit Congestion 880 Notification (ECN) Protocol for Very Low Queuing Delay 881 (L4S)", Work in Progress, Internet-Draft, draft-ietf- 882 tsvwg-ecn-l4s-id-24, 1 February 2022, 883 . 886 [I-D.ietf-tsvwg-l4s-arch] 887 Briscoe, B., Schepper, K. D., Bagnulo, M., and G. White, 888 "Low Latency, Low Loss, Scalable Throughput (L4S) Internet 889 Service: Architecture", Work in Progress, Internet-Draft, 890 draft-ietf-tsvwg-l4s-arch-16, 1 February 2022, 891 . 894 [IEEE802-11] 895 IEEE-SA, "IEEE 802.11-2020", IEEE 802, December 2020, 896 . 898 [RFC2475] Blake, S., Black, D., Carlson, M., Davies, E., Wang, Z., 899 and W. Weiss, "An Architecture for Differentiated 900 Services", RFC 2475, DOI 10.17487/RFC2475, December 1998, 901 . 903 [RFC3551] Schulzrinne, H. and S. Casner, "RTP Profile for Audio and 904 Video Conferences with Minimal Control", STD 65, RFC 3551, 905 DOI 10.17487/RFC3551, July 2003, 906 . 908 [RFC4594] Babiarz, J., Chan, K., and F. Baker, "Configuration 909 Guidelines for DiffServ Service Classes", RFC 4594, 910 DOI 10.17487/RFC4594, August 2006, 911 . 913 [RFC5127] Chan, K., Babiarz, J., and F. Baker, "Aggregation of 914 Diffserv Service Classes", RFC 5127, DOI 10.17487/RFC5127, 915 February 2008, . 917 [RFC7893] Stein, Y(J)., Black, D., and B. Briscoe, "Pseudowire 918 Congestion Considerations", RFC 7893, 919 DOI 10.17487/RFC7893, June 2016, 920 . 922 [RFC8033] Pan, R., Natarajan, P., Baker, F., and G. White, 923 "Proportional Integral Controller Enhanced (PIE): A 924 Lightweight Control Scheme to Address the Bufferbloat 925 Problem", RFC 8033, DOI 10.17487/RFC8033, February 2017, 926 . 928 [RFC8034] White, G. and R. Pan, "Active Queue Management (AQM) Based 929 on Proportional Integral Controller Enhanced PIE) for 930 Data-Over-Cable Service Interface Specifications (DOCSIS) 931 Cable Modems", RFC 8034, DOI 10.17487/RFC8034, February 932 2017, . 934 [RFC8083] Perkins, C. and V. Singh, "Multimedia Congestion Control: 935 Circuit Breakers for Unicast RTP Sessions", RFC 8083, 936 DOI 10.17487/RFC8083, March 2017, 937 . 939 [RFC8100] Geib, R., Ed. and D. Black, "Diffserv-Interconnection 940 Classes and Practice", RFC 8100, DOI 10.17487/RFC8100, 941 March 2017, . 943 [RFC8289] Nichols, K., Jacobson, V., McGregor, A., Ed., and J. 944 Iyengar, Ed., "Controlled Delay Active Queue Management", 945 RFC 8289, DOI 10.17487/RFC8289, January 2018, 946 . 948 [RFC8290] Hoeiland-Joergensen, T., McKenney, P., Taht, D., Gettys, 949 J., and E. Dumazet, "The Flow Queue CoDel Packet Scheduler 950 and Active Queue Management Algorithm", RFC 8290, 951 DOI 10.17487/RFC8290, January 2018, 952 . 954 [SA-5G] 3GPP, "System Architecture for 5G", TS 23.501, 2019. 956 Appendix A. DSCP Remarking Pathologies 958 Some network operators typically bleach (zero out) the Diffserv field 959 on ingress into their network 960 [I-D.ietf-tsvwg-dscp-considerations][Custura][Barik], and in some 961 cases apply their own DSCP for internal usage. Bleaching the NQB 962 DSCP is not expected to cause harm to default traffic, but it will 963 severely limit the ability to provide NQB treatment end-to-end. 964 Reports on existing deployments of DSCP manipulation [Custura][Barik] 965 categorize the re-marking behaviors into the following six policies: 966 bleach all traffic (set DSCP to zero), set the top three bits (the 967 former Precedence bits) on all traffic to 0b000, 0b001, or 0b010, set 968 the low three bits on all traffic to 0b000, or remark all traffic to 969 a particular (non-zero) DSCP value. 971 Regarding the DSCP values of 5 & 45, there were no observations of 972 DSCP manipulation reported in which traffic was marked 5 or 45 by any 973 of these policies. Thus it appears that these re-marking policies 974 would be unlikely to result in QB traffic being marked as NQB (45). 975 In terms of the fate of NQB-marked traffic that is subjected to one 976 of these policies, the result would be that NQB marked traffic would 977 be indistinguishable from some subset (possibly all) of other 978 traffic. In the policies where all traffic is remarked using the 979 same (zero or non-zero) DSCP, the ability for a subsequent network 980 hop to differentiate NQB traffic via DSCP would clearly be lost 981 entirely. 983 In the policies where the top three bits are overwritten, both NQB 984 values (5 & 45) would receive the same marking as would the currently 985 unassigned Pool 3 DSCPs 13,21,29,37,53,61, with all of these code 986 points getting mapped to DSCP=5, 13 or 21 (depending on the overwrite 987 value used). Since none of the DSCPs in the preceding lists are 988 currently assigned by IANA, and they all are set aside for Standards 989 Action, it is believed that they are not widely used currently, but 990 this may vary based on local-usage. 992 For the policy in which the low three bits are set to 0b000, the NQB 993 (45) value would be mapped to CS5 and would be indistinguishable from 994 CS5, VA, EF (and the unassigned DSCPs 41, 42, 43). Traffic marked 995 using the existing standardized DSCPs in this list are likely to 996 share the same general properties as NQB traffic (non capacity- 997 seeking, very low data rate or relatively low and consistent data 998 rate). Similarly, any future recommended usage for DSCPs 41, 42, 43 999 would likely be somewhat compatible with NQB treatment, assuming that 1000 IP Precedence compatibility (see Section 1.5.4 of [RFC4594]) is 1001 maintained in the future. Here there may be an opportunity for a 1002 node to provide the NQB PHB or the CS5 PHB to CS5-marked traffic and 1003 retain some of the benefits of NQB marking. This could be another 1004 motivation to (as discussed in Section 4.2) classify CS5-marked 1005 traffic into NQB queue. For this same re-marking policy, the NQB (5) 1006 value would be mapped to CS0/default and would be indistinguishable 1007 from CS0, LE (and the unassigned DSCPs 2,3,4,6,7). In this case, NQB 1008 traffic is likely to be given default treatment in all subsequent 1009 nodes, which would eliminate the ability to provide NQB treatment in 1010 those nodes, but would be relatively harmless otherwise. 1012 Authors' Addresses 1014 Greg White 1015 CableLabs 1016 Email: g.white@cablelabs.com 1018 Thomas Fossati 1019 ARM 1020 Email: Thomas.Fossati@arm.com