idnits 2.17.1 draft-ietf-detnet-bounded-latency-01.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 Security Considerations section. ** The document seems to lack an IANA Considerations section. (See Section 2.2 of https://www.ietf.org/id-info/checklist for how to handle the case when there are no actions for IANA.) == There are 2 instances of lines with non-RFC6890-compliant IPv4 addresses in the document. If these are example addresses, they should be changed. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == Line 625 has weird spacing: '...N queue non...' -- The document date (November 4, 2019) is 1625 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- -- Looks like a reference, but probably isn't: '2' on line 278 -- Looks like a reference, but probably isn't: '4' on line 817 == Unused Reference: 'NetCalBook' is defined on line 1206, but no explicit reference was found in the text == Outdated reference: A later version (-13) exists of draft-ietf-detnet-architecture-08 == Outdated reference: A later version (-07) exists of draft-ietf-detnet-ip-00 == Outdated reference: A later version (-13) exists of draft-ietf-detnet-mpls-00 Summary: 2 errors (**), 0 flaws (~~), 7 warnings (==), 3 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 DetNet N. Finn 3 Internet-Draft Huawei Technologies Co. Ltd 4 Intended status: Informational J-Y. Le Boudec 5 Expires: May 7, 2020 E. Mohammadpour 6 EPFL 7 J. Zhang 8 Huawei Technologies Co. Ltd 9 B. Varga 10 J. Farkas 11 Ericsson 12 November 4, 2019 14 DetNet Bounded Latency 15 draft-ietf-detnet-bounded-latency-01 17 Abstract 19 This document presents a timing model for Deterministic Networking 20 (DetNet), so that existing and future standards can achieve the 21 DetNet quality of service features of bounded latency and zero 22 congestion loss. It defines requirements for resource reservation 23 protocols or servers. It calls out queuing mechanisms, defined in 24 other documents, that can provide the DetNet quality of service. 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 May 7, 2020. 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. Terminology and Definitions . . . . . . . . . . . . . . . . . 3 62 3. DetNet bounded latency model . . . . . . . . . . . . . . . . 4 63 3.1. Flow creation . . . . . . . . . . . . . . . . . . . . . . 4 64 3.1.1. Static flow latency calculation . . . . . . . . . . . 4 65 3.1.2. Dynamic flow latency calculation . . . . . . . . . . 5 66 3.2. Relay node model . . . . . . . . . . . . . . . . . . . . 6 67 4. Computing End-to-end Delay Bounds . . . . . . . . . . . . . . 8 68 4.1. Non-queuing delay bound . . . . . . . . . . . . . . . . . 8 69 4.2. Queuing delay bound . . . . . . . . . . . . . . . . . . . 9 70 4.2.1. Per-flow queuing mechanisms . . . . . . . . . . . . . 9 71 4.2.2. Per-class queuing mechanisms . . . . . . . . . . . . 9 72 4.3. Ingress considerations . . . . . . . . . . . . . . . . . 10 73 4.4. Interspersed non-DetNet transit nodes . . . . . . . . . . 11 74 5. Achieving zero congestion loss . . . . . . . . . . . . . . . 11 75 6. Queuing techniques . . . . . . . . . . . . . . . . . . . . . 13 76 6.1. Queuing data model . . . . . . . . . . . . . . . . . . . 13 77 6.2. Preemption . . . . . . . . . . . . . . . . . . . . . . . 15 78 6.3. Time-scheduled queuing . . . . . . . . . . . . . . . . . 15 79 6.4. Credit-Based Shaper with Asynchronous Traffic Shaping . . 16 80 6.4.1. Delay Bound Calculation . . . . . . . . . . . . . . . 18 81 6.4.2. Flow Admission . . . . . . . . . . . . . . . . . . . 19 82 6.5. IntServ . . . . . . . . . . . . . . . . . . . . . . . . . 20 83 6.6. Cyclic Queuing and Forwarding . . . . . . . . . . . . . . 23 84 6.6.1. CQF timing sequence . . . . . . . . . . . . . . . . . 24 85 6.6.2. CQF latency calculation . . . . . . . . . . . . . . . 24 86 7. References . . . . . . . . . . . . . . . . . . . . . . . . . 25 87 7.1. Normative References . . . . . . . . . . . . . . . . . . 25 88 7.2. Informative References . . . . . . . . . . . . . . . . . 26 89 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 27 91 1. Introduction 93 The ability for IETF Deterministic Networking (DetNet) or IEEE 802.1 94 Time-Sensitive Networking (TSN, [IEEE8021TSN]) to provide the DetNet 95 services of bounded latency and zero congestion loss depends upon A) 96 configuring and allocating network resources for the exclusive use of 97 DetNet/TSN flows; B) identifying, in the data plane, the resources to 98 be utilized by any given packet, and C) the detailed behavior of 99 those resources, especially transmission queue selection, so that 100 latency bounds can be reliably assured. Thus, DetNet is an example 101 of an IntServ Guaranteed Quality of Service [RFC2212] 103 As explained in [I-D.ietf-detnet-architecture], DetNet flows are 104 characterized by 1) a maximum bandwidth, guaranteed either by the 105 transmitter or by strict input metering; and 2) a requirement for a 106 guaranteed worst-case end-to-end latency. That latency guarantee, in 107 turn, provides the opportunity for the network to supply enough 108 buffer space to guarantee zero congestion loss. 110 To be of use to the applications identified in [RFC8578], it must be 111 possible to calculate, before the transmission of a DetNet flow 112 commences, both the worst-case end-to-end network latency, and the 113 amount of buffer space required at each hop to ensure against 114 congestion loss. 116 This document references specific queuing mechanisms, defined in 117 other documents, that can be used to control packet transmission at 118 each output port and achieve the DetNet qualities of service. This 119 document presents a timing model for sources, destinations, and the 120 DetNet transit nodes that relay packets that is applicable to all of 121 those referenced queuing mechanisms. 123 Using the model presented in this document, it should be possible for 124 an implementor, user, or standards development organization to select 125 a particular set of queuing mechanisms for each device in a DetNet 126 network, and to select a resource reservation algorithm for that 127 network, so that those elements can work together to provide the 128 DetNet service. 130 This document does not specify any resource reservation protocol or 131 server. It does not describe all of the requirements for that 132 protocol or server. It does describe requirements for such resource 133 reservation methods, and for queuing mechanisms that, if met, will 134 enable them to work together. 136 2. Terminology and Definitions 138 This document uses the terms defined in 139 [I-D.ietf-detnet-architecture]. 141 3. DetNet bounded latency model 143 3.1. Flow creation 145 This document assumes that following paradigm is used for 146 provisioning DetNet flows: 148 1. Perform any configuration required by the DetNet transit nodes in 149 the network for the classes of service to be offered, including 150 one or more classes of DetNet service. This configuration is 151 done beforehand, and not tied to any particular flow. 153 2. Characterize the new DetNet flow, particularly in terms of 154 required bandwidth. 156 3. Establish the path that the DetNet flow will take through the 157 network from the source to the destination(s). This can be a 158 point-to-point or a point-to-multipoint path. 160 4. Select one of the DetNet classes of service for the DetNet flow. 162 5. Compute the worst-case end-to-end latency for the DetNet flow, 163 using one of the methods, below (Section 3.1.1, Section 3.1.2). 164 In the process, determine whether sufficient resources are 165 available for that flow to guarantee the required latency and to 166 provide zero congestion loss. 168 6. Assuming that the resources are available, commit those resources 169 to the flow. This may or may not require adjusting the 170 parameters that control the filtering and/or queuing mechanisms 171 at each hop along the flow's path. 173 This paradigm can be implemented using peer-to-peer protocols or 174 using a central server. In some situations, a lack of resources can 175 require backtracking and recursing through this list. 177 Issues such as un-provisioning a DetNet flow in favor of another, 178 when resources are scarce, are not considered, here. Also not 179 addressed is the question of how to choose the path to be taken by a 180 DetNet flow. 182 3.1.1. Static flow latency calculation 184 The static problem: 185 Given a network and a set of DetNet flows, compute an end-to- 186 end latency bound (if computable) for each flow, and compute 187 the resources, particularly buffer space, required in each 188 DetNet transit node to achieve zero congestion loss. 190 In this calculation, all of the DetNet flows are known before the 191 calculation commences. This problem is of interest to relatively 192 static networks, or static parts of larger networks. It gives the 193 best possible worst-case behavior. The calculations can be extended 194 to provide global optimizations, such as altering the path of one 195 DetNet flow in order to make resources available to another DetNet 196 flow with tighter constraints. 198 The static flow calculation is not limited only to static networks; 199 the entire calculation for all flows can be repeated each time a new 200 DetNet flow is created or deleted. If some already-established flow 201 would be pushed beyond its latency requirements by the new flow, then 202 the new flow can be refused, or some other suitable action taken. 204 This calculation may be more difficult to perform than that of the 205 dynamic calculation (Section 3.1.2), because the flows passing 206 through one port on a DetNet transit node affect each others' 207 latency. The effects can even be circular, from Flow A to B to C and 208 back to A. On the other hand, the static calculation can often 209 accommodate queuing methods, such as transmission selection by strict 210 priority, that are unsuitable for the dynamic calculation. 212 3.1.2. Dynamic flow latency calculation 214 The dynamic problem: 215 Given a network whose maximum capacity for DetNet flows is 216 bounded by a set of static configuration parameters applied 217 to the DetNet transit nodes, and given just one DetNet flow, 218 compute the worst-case end-to-end latency that can be 219 experienced by that flow, no matter what other DetNet flows 220 (within the network's configured parameters) might be created 221 or deleted in the future. Also, compute the resources, 222 particularly buffer space, required in each DetNet transit 223 node to achieve zero congestion loss. 225 This calculation is dynamic, in the sense that flows can be added or 226 deleted at any time, with a minimum of computation effort, and 227 without affecting the guarantees already given to other flows. 229 The choice of queuing methods is critical to the applicability of the 230 dynamic calculation. Some queuing methods (e.g. CQF, Section 6.6) 231 make it easy to configure bounds on the network's capacity, and to 232 make independent calculations for each flow. [[E:The rest of this 233 paragraph should be changed.]] Other queuing methods (e.g., 234 transmission selection by strict priority), make this calculation 235 impossible, because the worst case for one flow cannot be computed 236 without complete knowledge of all other flows. Other queuing methods 237 (e.g. the credit-based shaper defined in [IEEE8021Q] section 8.6.8.2) 238 can be used for dynamic flow creation, but yield poorer latency and 239 buffer space guarantees than when that same queuing method is used 240 for static flow creation (Section 3.1.1). 242 [[E:proposed replacement: Some other queuing methods (e.g. strict 243 priority with the credit-based shaper defined in [IEEE8021Q] section 244 8.6.8.2) can be used for dynamic flow creation, but yield poorer 245 latency and buffer space guarantees than when that same queuing 246 method is used for static flow creation (Section 3.1.1).]] 248 3.2. Relay node model 250 A model for the operation of a DetNet transit node is required, in 251 order to define the latency and buffer calculations. In Figure 1 we 252 see a breakdown of the per-hop latency experienced by a packet 253 passing through a DetNet transit node, in terms that are suitable for 254 computing both hop-by-hop latency and per-hop buffer requirements. 256 DetNet transit node A DetNet transit node B 257 +-------------------------+ +------------------------+ 258 | Queuing | | Queuing | 259 | Regulator subsystem | | Regulator subsystem | 260 | +-+-+-+-+ +-+-+-+-+ | | +-+-+-+-+ +-+-+-+-+ | 261 -->+ | | | | | | | | | + +------>+ | | | | | | | | | + +---> 262 | +-+-+-+-+ +-+-+-+-+ | | +-+-+-+-+ +-+-+-+-+ | 263 | | | | 264 +-------------------------+ +------------------------+ 265 |<->|<------>|<------->|<->|<---->|<->|<------>|<------>|<->|<-- 266 2,3 4 5 6 1 2,3 4 5 6 1 2,3 267 1: Output delay 4: Processing delay 268 2: Link delay 5: Regulation delay 269 3: Preemption delay 6: Queuing delay. 271 Figure 1: Timing model for DetNet or TSN 273 In Figure 1, we see two DetNet transit nodes (typically, bridges or 274 routers), with a wired link between them. In this model, the only 275 queues, that we deal with explicitly, are attached to the output 276 port; other queues are modeled as variations in the other delay 277 times. (E.g., an input queue could be modeled as either a variation 278 in the link delay [2] or the processing delay [4].) There are six 279 delays that a packet can experience from hop to hop. 281 1. Output delay 282 The time taken from the selection of a packet for output from a 283 queue to the transmission of the first bit of the packet on the 284 physical link. If the queue is directly attached to the physical 285 port, output delay can be a constant. But, in many 286 implementations, the queuing mechanism in a forwarding ASIC is 287 separated from a multi-port MAC/PHY, in a second ASIC, by a 288 multiplexed connection. This causes variations in the output 289 delay that are hard for the forwarding node to predict or control. 291 2. Link delay 292 The time taken from the transmission of the first bit of the 293 packet to the reception of the last bit, assuming that the 294 transmission is not suspended by a preemption event. This delay 295 has two components, the first-bit-out to first-bit-in delay and 296 the first-bit-in to last-bit-in delay that varies with packet 297 size. The former is typically measured by the Precision Time 298 Protocol and is constant (see [I-D.ietf-detnet-architecture]). 299 However, a virtual "link" could exhibit a variable link delay. 301 3. Preemption delay 302 If the packet is interrupted in order to transmit another packet 303 or packets, (e.g. [IEEE8023] clause 99 frame preemption) an 304 arbitrary delay can result. 306 4. Processing delay 307 This delay covers the time from the reception of the last bit of 308 the packet to the time the packet is enqueued in the regulator 309 (Queuing subsystem, if there is no regulation). This delay can be 310 variable, and depends on the details of the operation of the 311 forwarding node. 313 5. Regulator delay 314 This is the time spent from the insertion of the last bit of a 315 packet into a regulation queue until the time the packet is 316 declared eligible according to its regulation constraints. We 317 assume that this time can be calculated based on the details of 318 regulation policy. If there is no regulation, this time is zero. 320 6. Queuing subsystem delay 321 This is the time spent for a packet from being declared eligible 322 until being selected for output on the next link. We assume that 323 this time is calculable based on the details of the queuing 324 mechanism. If there is no regulation, this time is from the 325 insertion of the packet into a queue until it is selected for 326 output on the next link. 328 Not shown in Figure 1 are the other output queues that we presume are 329 also attached to that same output port as the queue shown, and 330 against which this shown queue competes for transmission 331 opportunities. 333 The initial and final measurement point in this analysis (that is, 334 the definition of a "hop") is the point at which a packet is selected 335 for output. In general, any queue selection method that is suitable 336 for use in a DetNet network includes a detailed specification as to 337 exactly when packets are selected for transmission. Any variations 338 in any of the delay times 1-4 result in a need for additional buffers 339 in the queue. If all delays 1-4 are constant, then any variation in 340 the time at which packets are inserted into a queue depends entirely 341 on the timing of packet selection in the previous node. If the 342 delays 1-4 are not constant, then additional buffers are required in 343 the queue to absorb these variations. Thus: 345 o Variations in output delay (1) require buffers to absorb that 346 variation in the next hop, so the output delay variations of the 347 previous hop (on each input port) must be known in order to 348 calculate the buffer space required on this hop. 350 o Variations in processing delay (4) require additional output 351 buffers in the queues of that same DetNet transit node. Depending 352 on the details of the queueing subsystem delay (6) calculations, 353 these variations need not be visible outside the DetNet transit 354 node. 356 4. Computing End-to-end Delay Bounds 358 4.1. Non-queuing delay bound 360 End-to-end delay bounds can be computed using the delay model in 361 Section 3.2. Here, it is important to be aware that for several 362 queuing mechanisms, the end-to-end delay bound is less than the sum 363 of the per-hop delay bounds. An end-to-end delay bound for one 364 DetNet flow can be computed as 366 end_to_end_delay_bound = non_queuing_delay_bound + 367 queuing_delay_bound 369 The two terms in the above formula are computed as follows. 371 First, at the h-th hop along the path of this DetNet flow, obtain an 372 upperbound per-hop_non_queuing_delay_bound[h] on the sum of the 373 bounds over the delays 1,2,3,4 of Figure 1. These upper bounds are 374 expected to depend on the specific technology of the DetNet transit 375 node at the h-th hop but not on the T-SPEC of this DetNet flow. Then 376 set non_queuing_delay_bound = the sum of per- 377 hop_non_queuing_delay_bound[h] over all hops h. 379 Second, compute queuing_delay_bound as an upper bound to the sum of 380 the queuing delays along the path. The value of queuing_delay_bound 381 depends on the T-SPEC of this flow and possibly of other flows in the 382 network, as well as the specifics of the queuing mechanisms deployed 383 along the path of this flow. The computation of queuing_delay_bound 384 is described in Section 4.2 as a separate section. 386 4.2. Queuing delay bound 388 For several queuing mechanisms, queuing_delay_bound is less than the 389 sum of upper bounds on the queuing delays (5,6) at every hop. This 390 occurs with (1) per-flow queuing, and (2) per-class queuing with 391 regulators, as explained in Section 4.2.1, Section 4.2.2, and 392 Section 6. 394 For other queuing mechanisms the only available value of 395 queuing_delay_bound is the sum of the per-hop queuing delay bounds. 396 In such cases, the computation of per-hop queuing delay bounds must 397 account for the fact that the T-SPEC of a DetNet flow is no longer 398 satisfied at the ingress of a hop, since burstiness increases as one 399 flow traverses one DetNet transit node. 401 4.2.1. Per-flow queuing mechanisms 403 With such mechanisms, each flow uses a separate queue inside every 404 node. The service for each queue is abstracted with a guaranteed 405 rate and a latency. For every flow, a per-node delay bound as well 406 as an end-to-end delay bound can be computed from the traffic 407 specification of this flow at its source and from the values of rates 408 and latencies at all nodes along its path. The per-flow queuing is 409 used in IntServ. Details of calculation for IntServ are described in 410 Section 6.5. 412 4.2.2. Per-class queuing mechanisms 414 With such mechanisms, the flows that have the same class share the 415 same queue. A practical example is the credit-based shaper defined 416 in section 8.6.8.2 of [IEEE8021Q]. One key issue in this context is 417 how to deal with the burstiness cascade: individual flows that share 418 a resource dedicated to a class may see their burstiness increase, 419 which may in turn cause increased burstiness to other flows 420 downstream of this resource. Computing delay upper bounds for such 421 cases is difficult, and in some conditions impossible 422 [charny2000delay][bennett2002delay]. Also, when bounds are obtained, 423 they depend on the complete configuration, and must be recomputed 424 when one flow is added. (The dynamic calculation, Section 3.1.2.) 426 A solution to deal with this issue is to reshape the flows at every 427 hop. This can be done with per-flow regulators (e.g. leaky bucket 428 shapers), but this requires per-flow queuing and defeats the purpose 429 of per-class queuing. An alternative is the interleaved regulator, 430 which reshapes individual flows without per-flow queuing 431 ([Specht2016UBS], [IEEE8021Qcr]). With an interleaved regulator, the 432 packet at the head of the queue is regulated based on its (flow) 433 regulation constraints; it is released at the earliest time at which 434 this is possible without violating the constraint. One key feature 435 of per-flow or interleaved regulator is that, it does not increase 436 worst-case latency bounds [le_boudec_theory_2018]. Specifically, 437 when an interleaved regulator is appended to a FIFO subsystem, it 438 does not increase the worst-case delay of the latter. 440 Figure 2 shows an example of a network with 5 nodes, per-class 441 queuing mechanism and interleaved regulators as in Figure 1. An end- 442 to-end delay bound for flow f, traversing nodes 1 to 5, is calculated 443 as follows: 445 end_to_end_latency_bound_of_flow_f = C12 + C23 + C34 + S4 447 In the above formula, Cij is a bound on the delay of the queuing 448 subsystem in node i and interleaved regulator of node j, and S4 is a 449 bound on the delay of the queuing subsystem in node 4 for flow f. In 450 fact, using the delay definitions in Section 3.2, Cij is a bound on 451 sum of the delays 1,2,3,6 of node i and 4,5 of node j. Similarly, S4 452 is a bound on sum of the delays 1,2,3,6 of node 4. A practical 453 example of queuing model and delay calculation is presented 454 Section 6.4. 456 f 457 -----------------------------> 458 +---+ +---+ +---+ +---+ +---+ 459 | 1 |---| 2 |---| 3 |---| 4 |---| 5 | 460 +---+ +---+ +---+ +---+ +---+ 461 \__C12_/\__C23_/\__C34_/\_S4_/ 463 Figure 2: End-to-end delay computation example 465 REMARK: The end-to-end delay bound calculation provided here gives a 466 much better upper bound in comparison with end-to-end delay bound 467 computation by adding the delay bounds of each node in the path of a 468 flow [TSNwithATS]. 470 4.3. Ingress considerations 472 A sender can be a DetNet node which uses exactly the same queuing 473 methods as its adjacent DetNet transit node, so that the delay and 474 buffer bounds calculations at the first hop are indistinguishable 475 from those at a later hop within the DetNet domain. On the other 476 hand, the sender may be DetNet unaware, in which case some 477 conditioning of the flow may be necessary at the ingress DetNet 478 transit node. 480 This ingress conditioning typically consists of a FIFO with an output 481 regulator that is compatible with the queuing employed by the DetNet 482 transit node on its output port(s). For some queuing methods, simply 483 requires added extra buffer space in the queuing subsystem. Ingress 484 conditioning requirements for different queuing methods are mentioned 485 in the sections, below, describing those queuing methods. 487 4.4. Interspersed non-DetNet transit nodes 489 It is sometimes desirable to build a network that has both DetNet 490 aware transit nodes and DetNet non-aware transit nodes, and for a 491 DetNet flow to traverse an island of non-DetNet transit nodes, while 492 still allowing the network to offer delay and congestion loss 493 guarantees. This is possible under certain conditions. 495 In general, when passing through a non-DetNet island, the island 496 causes delay variation in excess of what would be caused by DetNet 497 nodes. That is, the DetNet flow is "lumpier" after traversing the 498 non-DetNet island. DetNet guarantees for delay and buffer 499 requirements can still be calculated and met if and only if the 500 following are true: 502 1. The latency variation across the non-DetNet island must be 503 bounded and calculable. 505 2. An ingress conditioning function (Section 4.3) may be required at 506 the re-entry to the DetNet-aware domain. This will, at least, 507 require some extra buffering to accommodate the additional delay 508 variation, and thus further increases the delay bound. 510 The ingress conditioning is exactly the same problem as that of a 511 sender at the edge of the DetNet domain. The requirement for bounds 512 on the latency variation across the non-DetNet island is typically 513 the most difficult to achieve. Without such a bound, it is obvious 514 that DetNet cannot deliver its guarantees, so a non-DetNet island 515 that cannot offer bounded latency variation cannot be used to carry a 516 DetNet flow. 518 5. Achieving zero congestion loss 520 When the input rate to an output queue exceeds the output rate for a 521 sufficient length of time, the queue must overflow. This is 522 congestion loss, and this is what deterministic networking seeks to 523 avoid. 525 To avoid congestion losses, an upper bound on the backlog present in 526 the regulator and queuing subsystem of Figure 1 must be computed 527 during resource reservation. This bound depends on the set of flows 528 that use these queues, the details of the specific queuing mechanism 529 and an upper bound on the processing delay (4). The queue must 530 contain the packet in transmission plus all other packets that are 531 waiting to be selected for output. 533 A conservative backlog bound, that applies to all systems, can be 534 derived as follows. 536 The backlog bound is counted in data units (bytes, or words of 537 multiple bytes) that are relevant for buffer allocation. For every 538 class we need one buffer space for the packet in transmission, plus 539 space for the packets that are waiting to be selected for output. 540 Excluding transmission and preemption times, the packets are waiting 541 in the queue since reception of the last bit, for a duration equal to 542 the processing delay (4) plus the queuing delays (5,6). 544 Let 546 o total_in_rate be the sum of the line rates of all input ports that 547 send traffic of any class to this output port. The value of 548 total_in_rate is in data units (e.g. bytes) per second. 550 o nb_input_ports be the number input ports that send traffic of any 551 class to this output port 553 o max_packet_length be the maximum packet size for packets of any 554 class that may be sent to this output port. This is counted in 555 data units. 557 o max_delay456 be an upper bound, in seconds, on the sum of the 558 processing delay (4) and the queuing delays (5,6) for a packet of 559 any class at this output port. 561 Then a bound on the backlog of traffic of all classes in the queue at 562 this output port is 564 [[E: The formula is not right; why do we need nb_classes to compute 565 backlog bound?]] 567 backlog_bound = ( nb_classes + nb_input_ports ) * 568 max_packet_length + total_in_rate* max_delay456 570 [[E: proposed general backlog bound:]] 571 backlog_bound = nb_input_ports * max_packet_length + 572 total_in_rate* max_delay456 574 6. Queuing techniques 576 6.1. Queuing data model 578 Sophisticated queuing mechanisms are available in Layer 3 (L3, see, 579 e.g., [RFC7806] for an overview). In general, we assume that "Layer 580 3" queues, shapers, meters, etc., are precisely the "regulators" 581 shown in Figure 1. The "queuing subsystems" in this figure are not 582 the province solely of bridges; they are an essential part of any 583 DetNet transit node. As illustrated by numerous implementation 584 examples, some of the "Layer 3" mechanisms described in documents 585 such as [RFC7806] are often integrated, in an implementation, with 586 the "Layer 2" mechanisms also implemented in the same node. An 587 integrated model is needed in order to successfully predict the 588 interactions among the different queuing mechanisms needed in a 589 network carrying both DetNet flows and non-DetNet flows. 591 Figure 3 shows the general model for the flow of packets through the 592 queues of a DetNet transit node. Packets are assigned to a class of 593 service. The classes of service are mapped to some number of 594 regulator queues. Only DetNet/TSN packets pass through regulators. 595 Queues compete for the selection of packets to be passed to queues in 596 the queuing subsystem. Packets again are selected for output from 597 the queuing subsystem. 599 | 600 +--------------------------------V----------------------------------+ 601 | Class of Service Assignment | 602 +--+------+----------+---------+-----------+-----+-------+-------+--+ 603 | | | | | | | | 604 +--V-+ +--V-+ +--V--+ +--V--+ +--V--+ | | | 605 |Flow| |Flow| |Flow | |Flow | |Flow | | | | 606 | 0 | | 1 | ... | i | | i+1 | ... | n | | | | 607 | reg| | reg| | reg | | reg | | reg | | | | 608 +--+-+ +--+-+ +--+--+ +--+--+ +--+--+ | | | 609 | | | | | | | | 610 +--V------V----------V--+ +--V-----------V--+ | | | 611 | Trans. selection | | Trans. select. | | | | 612 +----------+------------+ +-----+-----------+ | | | 613 | | | | | 614 +--V--+ +--V--+ +--V--+ +--V--+ +--V--+ 615 | out | | out | | out | | out | | out | 616 |queue| |queue| |queue| |queue| |queue| 617 | 1 | | 2 | | 3 | | 4 | | 5 | 618 +--+--+ +--+--+ +--+--+ +--+--+ +--+--+ 619 | | | | | 620 +----------V----------------------V--------------V-------V-------V--+ 621 | Transmission selection | 622 +----------+----------------------+--------------+-------+-------+--+ 623 | | | | | 624 V V V V V 625 DetNet/TSN queue DetNet/TSN queue non-DetNet/TSN queues 627 Figure 3: IEEE 802.1Q Queuing Model: Data flow 629 Some relevant mechanisms are hidden in this figure, and are performed 630 in the queue boxes: 632 o Discarding packets because a queue is full. 634 o Discarding packets marked "yellow" by a metering function, in 635 preference to discarding "green" packets. 637 Ideally, neither of these actions are performed on DetNet packets. 638 Full queues for DetNet packets should occur only when a flow is 639 misbehaving, and the DetNet QoS does not include "yellow" service for 640 packets in excess of committed rate. 642 The Class of Service Assignment function can be quite complex, even 643 in a bridge [IEEE8021Q], since the introduction of per-stream 644 filtering and policing ([IEEE8021Q] clause 8.6.5.1). In addition to 645 the Layer 2 priority expressed in the 802.1Q VLAN tag, a DetNet 646 transit node can utilize any of the following information to assign a 647 packet to a particular class of service (queue): 649 o Input port. 651 o Selector based on a rotating schedule that starts at regular, 652 time-synchronized intervals and has nanosecond precision. 654 o MAC addresses, VLAN ID, IP addresses, Layer 4 port numbers, DSCP. 655 ([I-D.ietf-detnet-ip], [I-D.ietf-detnet-mpls]) (Work items are 656 expected to add MPC and other indicators.) 658 o The Class of Service Assignment function can contain metering and 659 policing functions. 661 o MPLS and/or pseudowire ([RFC6658]) labels. 663 The "Transmission selection" function decides which queue is to 664 transfer its oldest packet to the output port when a transmission 665 opportunity arises. 667 6.2. Preemption 669 In [IEEE8021Q] and [IEEE8023], the transmission of a frame can be 670 interrupted by one or more "express" frames, and then the interrupted 671 frame can continue transmission. This frame preemption is modeled as 672 consisting of two MAC/PHY stacks, one for packets that can be 673 interrupted, and one for packets that can interrupt the interruptible 674 packets. The Class of Service (queue) determines which packets are 675 which. Only one layer of preemption is supported -- a transmitter 676 cannot have more than one interrupted frame in progress. DetNet 677 flows typically pass through the interrupting MAC. Best-effort 678 queues pass through the interruptible MAC, and can thus be preempted. 680 6.3. Time-scheduled queuing 682 In [IEEE8021Q], the notion of time-scheduling queue gates is 683 described in section 8.6.8.4. Below every output queue (the lower 684 row of queues in Figure 3) is a gate that permits or denies the queue 685 to present data for transmission selection. The gates are controlled 686 by a rotating schedule that can be locked to a clock that is 687 synchronized with other DetNet transit nodes. The DetNet class of 688 service can be supplied by queuing mechanisms based on time, rather 689 than the regulator model in Figure 3. Generally speaking, this time- 690 aware scheduling can be used as a layer 2 time division multiplexing 691 (TDM) technique. 693 Consider the static configuration of a deterministic network. To 694 provide end-to-end latency guaranteed service, network nodes can 695 support time-based behavior, which is determined by gate control list 696 (GCL). GCL defines the gate operation, in open or closed state, with 697 associated timing for each traffic class queue. A time slice with 698 gate state "open" is called transmission window. The time-based 699 traffic scheduling must be coordinated among the DetNet transit nodes 700 along the path from sender to receiver, to control the transmission 701 of time-sensitive traffic. 703 Ideally all network devices are time synchronized and static GCL 704 configurations on all devices along the routed path are coordinated 705 to ensure that length of transmission window fits the assigned 706 frames, and no two time windows for DetNet traffic on the same port 707 overlap. (DetNet flows' windows can overlap with best-effort 708 windows, so that unused DetNet bandwidth is available to best-effort 709 traffic.) The processing delay, link delay and output delay in 710 transmitting are considered in GCL computation. Transmission window 711 for a certain flow may require that a time offset on consecutive hops 712 be selected to reduce queueing delay as much as possible. In this 713 case, TSN/DetNet frames transmit at the assigned transmission window 714 at every node through the routed path, with zero congestion loss and 715 bounded end-to-end latency. Then, the worst-case end-to-end latency 716 of the flow can be derived from GCL configuration. For a TSN or 717 DetNet frame, denote the transmission window on last hop closes at 718 gate_close_time_last_hop. Assuming talker supports scheduled traffic 719 behavior, it starts the transmission at gate_open_time_on_talker. 720 Then worst case end-to-end delay of this flow is bounded by 721 gate_close_time_last_hop - gate_open_time_on_talker + 722 link_delay_last_hop. 724 It should be noted that scheduled traffic service relies on a 725 synchronized network and coordinated GCL configuration. Synthesis of 726 GCL on multiple nodes in network is a scheduling problem considering 727 all TSN/DetNet flows traversing the network, which is a non- 728 deterministic polynomial-time hard (NP-hard) problem. Also, at this 729 writing, scheduled traffic service supports no more than eight 730 traffic classes, typically using up to seven priority classes and at 731 least one best effort class. 733 6.4. Credit-Based Shaper with Asynchronous Traffic Shaping 735 In the cosidered queuing model, there are four types of flows, 736 namely, control-data traffic (CDT), class A, class B, and best effort 737 (BE) in decreasing order of priority. Flows of classes A and B are 738 together referred to AVB flows. This model is a subset of Time- 739 Sensitive Networking as described next. 741 Based on the timing model described in Figure 1, the contention 742 occurs only at the output port of a relay node; therefore, the focus 743 of the rest of this subsection is on the regulator and queuing 744 subsystem in the output port of a relay node. The output port 745 performs per-class scheduling with eight classes (queuing 746 subsystems): one for CDT, one for class A traffic, one for class B 747 traffic, and five for BE traffic denoted as BE0-BE4. The queuing 748 policy for each queuing subsystem is FIFO. In addition, each node 749 output port also performs per-flow regulation for AVB flows using an 750 interleaved regulator (IR), called Asynchronous Traffic Shaper 751 [IEEE8021Qcr]. Thus, at each output port of a node, there is one 752 interleaved regulator per-input port and per-class; the interleaved 753 regulator is mapped to the regulator depicted in Figure 1. The 754 detailed picture of scheduling and regulation architecture at a node 755 output port is given by Figure 4. The packets received at a node 756 input port for a given class are enqueued in the respective 757 interleaved regulator at the output port. Then, the packets from all 758 the flows, including CDT and BE flows, are enqueued in queuing 759 subsytem; there is no regulator for such classes. 761 +--+ +--+ +--+ +--+ 762 | | | | | | | | 763 |IR| |IR| |IR| |IR| 764 | | | | | | | | 765 +-++XXX++-+ +-++XXX++-+ 766 | | | | 767 | | | | 768 +---+ +-v-XXX-v-+ +-v-XXX-v-+ +-----+ +-----+ +-----+ +-----+ +-----+ 769 | | | | | | |Class| |Class| |Class| |Class| |Class| 770 |CDT| | Class A | | Class B | | BE4 | | BE3 | | BE2 | | BE1 | | BE0 | 771 | | | | | | | | | | | | | | | | 772 +-+-+ +----+----+ +----+----+ +--+--+ +--+--+ +--+--+ +--+--+ +--+--+ 773 | | | | | | | | 774 | +-v-+ +-v-+ | | | | | 775 | |CBS| |CBS| | | | | | 776 | +-+-+ +-+-+ | | | | | 777 | | | | | | | | 778 +-v--------v-----------v---------v-------V-------v-------v-------v--+ 779 | Strict Priority selection | 780 +--------------------------------+----------------------------------+ 781 | 782 V 784 Figure 4: The architecture of an output port inside a relay node with 785 interleaved regulators (IRs) and credit-based shaper (CBS) 787 Each of the queuing subsystems for class A and B, contains Credit- 788 Based Shaper (CBS). The CBS serves a packet from a class according 789 to the available credit for that class. The credit for each class A 790 or B increases based on the idle slope, and decreases based on the 791 send slope, both of which are parameters of the CBS (Section 8.6.8.2 792 of [IEEE8021Q]). The CDT and BE0-BE4 flows are served by separate 793 queuing subsystems. Then, packets from all flows are served by a 794 transmission selection subsystem that serves packets from each class 795 based on its priority. All subsystems are non-preemptive. 796 Guarantees for AVB traffic can be provided only if CDT traffic is 797 bounded; it is assumed that the CDT traffic has leaky bucket arrival 798 curve with two parameters r_h as rate and b_h as bucket size, i.e., 799 the amount of bits entering a node within a time interval t is 800 bounded by r_h t + b_h. 802 Additionally, it is assumed that the AVB flows are also regulated at 803 their source according to leaky bucket arrival curve. At the source, 804 the traffic satisfies its regulation constraint, i.e. the delay due 805 to interleaved regulator at source is ignored. 807 At each DetNet transit node implementing an interleaved regulator, 808 packets of multiple flows are processed in one FIFO queue; the packet 809 at the head of the queue is regulated based on its leaky bucket 810 parameters; it is released at the earliest time at which this is 811 possible without violating the constraint. The regulation parameters 812 for a flow (leaky bucket rate and bucket size) are the same at its 813 source and at all DetNet transit nodes along its path. 815 6.4.1. Delay Bound Calculation 817 A delay bound of the queuing subsystem ([4] in Figure 1) for an AVB 818 flow of class A or B can be computed if the following condition 819 holds: 821 sum of leaky bucket rates of all flows of this class at this 822 transit node <= R, where R is given below for every class. 824 If the condition holds, the delay bounds for a flow of class X (A or 825 B) is d_X and calculated as: 827 d_X = T_X + (b_t_X-L_min_X)/R_X - L_min_X/c 829 where L_min_X is the minimum packet lengths of class X (A or B); c is 830 the output link transmission rate; b_t_X is the sum of the b term 831 (bucket size) for all the flows of the class X. Parameters R_X and 832 T_X are calculated as follows for class A and class B, separately: 834 If the flow is of class A: 836 R_A = I_A (c-r_h)/ c 838 T_A = L_nA + b_h + r_h L_n/c)/(c-r_h) 840 where L_nA is the maximum packet length of class B and BE packets; 841 L_n is the maximum packet length of classes A,B, and BE. 843 If the flow is of class B: 845 R_B = I_B (c-r_h)/ c 847 T_B = (L_BE + L_A + L_nA I_A/(c_h-I_A) + b_h + r_h L_n/c)/(c-r_h) 849 where L_A is the maximum packet length of class A; L_BE is the 850 maximum packet length of class BE. 852 Then, an end-to-end delay bound of class X (A or B)is calculated by 853 the formula Section 4.2.2, where for Cij: 855 Cij = d_X 857 More information of delay analysis in such a DetNet transit node is 858 described in [TSNwithATS]. 860 6.4.2. Flow Admission 862 The delay bound calculation requires some information about each 863 node. For each node, it is required to know the idle slope of CBS 864 for each class A and B (I_A and I_B), as well as the transmission 865 rate of the output link (c). Besides, it is necessary to have the 866 information on each class, i.e. maximum packet length of classes A, 867 B, and BE. Moreover, the leaky bucket parameters of CDT (r_h,b_h) 868 should be known. To admit a flow/flows, their delay requirements 869 should be guaranteed not to be violated. As described in 870 Section 3.1, the two problems, static and dynamic, are addressed 871 separately. In either of the problems, the rate and delay should be 872 guaranteed. Thus, 874 The static admission control: 875 The leaky bucket parameters of all flows are known, 876 therefore, for each flow f, a delay bound can be calculated. 877 The computed delay bound for every flow should not be more 878 than its delay requirement. Moreover, the sum of the rate of 879 each flow (r_f) should not be more than the rate allocated to 880 each class (R). If these two conditions hold, the 881 configuration is declared admissible. 883 The dynamic admission control: 885 For dynamic admission control, we allocate to every node and 886 class A or B, static value for rate (R) and maximum 887 burstiness (b_t). In addition, for every node and every 888 class A and B, two counters are maintained: 890 R_acc is equal to the sum of the leaky-bucket rates of all 891 flows of this class already admitted at this node; At all 892 times, we must have: 894 R_acc <=R, (Eq. 1) 896 b_acc is equal to the sum of the bucket sizes of all flows 897 of this class already admitted at this node; At all times, 898 we must have: 900 b_acc <=b_t. (Eq. 2) 902 A new flow is admitted at this node, if Eqs. (1) and (2) 903 continue to be satisfied after adding its leaky bucket rate 904 and bucket size to R_acc and b_acc. A flow is admitted in 905 the network, if it is admitted at all nodes along its path. 906 When this happens, all variables R_acc and b_acc along its 907 path must be incremented to reflect the addition of the flow. 908 Similarly, when a flow leaves the network, all variables 909 R_acc and b_acc along its path must be decremented to reflect 910 the removal of the flow. 912 The choice of the static values of R and b_t at all nodes and classes 913 must be done in a prior configuration phase; R controls the bandwidth 914 allocated to this class at this node, b_t affects the delay bound and 915 the buffer requirement. R must satisfy the constraints given in 916 Annex L.1 of [IEEE8021Q]. 918 6.5. IntServ 920 Integrated service (IntServ) is an architecture that specifies the 921 elements to guarantee quality of service (QoS) on networks. [[E: The 922 rest of this paragraph is better not to be placed here; these should 923 be mentioned (is mentioned) in the introduction.]] To satisfied 924 guaranteed service, a flow must conform to a traffic specification 925 (T-spec), and reservation is made along a path, only if routers are 926 able to guarantee the required bandwidth and buffer. 928 [[E: The information about arrival and service curves can be shorter 929 with less detail. I put a proposed text after description of 930 these.]] 932 Consider the traffic model which conforms to token bucket regulator 933 (r, b), with 935 o Token bucket depth (b). 937 o Token bucket rate (r). 939 The traffic specification can be described as an arrival curve: 941 alpha(t) = b + rt 943 This token bucket regulator requires that, during any time window t, 944 the number of bit for the flow is limited by alpha(t) = b + rt. 946 If resource reservation on a path is applied, IntServ model of a 947 router can be described as a rate-latency service curve beta(t). 949 beta(t) = max(0, R(t-T)) 951 It describes that bits might have to wait up to T before being served 952 with a rate greater or equal to R. 954 [[E: proposed text: 956 The flow, at the source, has a leaky bucket arrival curve with two 957 parameters r as rate and b as bucket size, i.e., the amount of bits 958 entering a node within a time interval t is bounded by r t + b. 960 If a resource reservation on a path is applied, a node provides a 961 guaranteed rate R and maximum service latency of T. This can be 962 interpreted in a way that the bits might have to wait up to T before 963 being served with a rate greater or equal to R. ]] 965 It should be noted that the guaranteed service rate R is a portion of 966 link's bandwidth. The selection of R is related to the specification 967 of flows traversing through the current node. For example, in strict 968 priority policy, considering a flow with priority i, its guaranteed 969 rate is R=c-sum(r_j), j 0.7 1 (units of T_c) 2 3 1039 DetNet transit node A out port 1 1040 | a <-DT->| b | c | d 1041 +------------+------+-------------------+-------------------+-------- 1042 \_____ \_____ 1043 \_____ \_____ queue-to-queue delay = 1.3 T_c 1044 \_____ \_____ 1045 \_____ \_____ DetNet transit node B 1046 \_ \_ queue assignment, in 1047 | | |<-DT->| port 2 to out 3 | 1048 -------+-------------------+------------+------+-------------------+- 1049 0.3 time--> 1.3 2.0 2.3 3.3 1051 window to transfer 1052 to buffer c ---> VVVVVVVVVVVV 1053 if dead time not window to transfer 1054 excessive VVVVVVVVVVVVVVVVVVV <--- to buffer d 1055 DetNet transit node B out port 3 1056 | a | b | c | d 1057 +-------------------+-------------------+-------------------+-------- 1058 0 time--> 1 2 3 1060 Figure 6: CQF timing diagram 1062 Figure 6 shows two DetNet transit nodes A and B, including three 1063 timelines for: 1065 1. The output queues on port 1 in node A. 1067 2. The input gate function ([IEEE8021Q], 8.6.5.1) that assigns 1068 packets received on port 1 of transit node B to output queues on 1069 port 2 of transit node B. 1071 3. The output queues on port 2 of node B. 1073 In this figure, the output ports on the two nodes are synchronized, 1074 and a new buffer starts transmitting at each tick, shown as 0, 1, 2, 1075 ... The output times shown for timelines 1 and 3 are the times at 1076 which packets are selected for output, which is the start point of 1077 the output time (1) of Figure 1. The queue assignments times on 1078 timeline 3 take place at the beginning of the queuing delay (6) of 1079 Figure 1. Time-based CQF, as described here, does not require any 1080 regulator queues. In the shown in the figure, the total time [[E: 1081 what is meant by total time? Does it mean a delay bound is 1.3 1082 T_C?]] for delays (1) through (6) of Figure 1, is 1.3T_c. Of course, 1083 any value is possible. 1085 6.6.1. CQF timing sequence 1087 In general, as shown in Figure 6, the windows for buffer assignment 1088 do not align perfectly with the windows for buffer transmission. The 1089 input gates (the center timeline in Figure 6) must switch from using 1090 one buffer to using another buffer in sync with the (delayed) 1091 received data, at times offset by the dead time from the output 1092 buffer switching (the bottom timeline in Figure 6). 1094 If the dead time DT in Figure 6 is not excessive, then it is feasible 1095 to subtract the dead time from the cycle time Tc, and use the 1096 remainder as the input window. In the example in Figure 6, packets 1097 from node A buffer a can be transferred from the input port to node 1098 B's buffer "c" during the window shown by the upper row "VVVV...". 1099 Input must cease by time = 2.0, because that is when transit node B 1100 starts transmitting the contents of buffer c. In this case, only two 1101 output buffers are in use, one filling and one outputting. 1103 If the dead time is too large (e.g., if the delays placed the middle 1104 timeline's switching points at n+0.9, instead of n+0.3), three 1105 buffers are used by node B. This case is shown by the lower row 1106 "VVVV..." in Figure 6. In this case, node B places the data received 1107 from node A buffer a into node B buffer d between the times 1.3 and 1108 2.3 in Figure 6. Buffer b starts outputting at time = 2.0, while 1109 buffer d is filling. Thus, three buffers are in use, one filling, 1110 one waiting, and one emptying. 1112 6.6.2. CQF latency calculation 1114 The per-hop latency is trivially determined by the wire delay and the 1115 queuing delay. Since the wire delay is either absorbed into the 1116 queueing delay (dead time is small and two buffers are used) or 1117 padded out to a whole cycle time T_c (three buffers are used) the 1118 per-hop latency is always an integral number of cycle times T_c, with 1119 a latency variation at the output of the final hop of T_c. 1121 Ingress conditioning (Section 4.3) may be required if the source of a 1122 DetNet flow does not, itself, employ CQF. 1124 Note that there are no per-flow parameters in the CQF technique. 1125 Therefore, there is no requirement for per-hop configuration when a 1126 new DetNet flow is added to a network, except perhaps for ingress 1127 checks to see that the transmitter does not exceed the contracted 1128 bandwidth. 1130 7. References 1132 7.1. Normative References 1134 [I-D.ietf-detnet-architecture] 1135 Finn, N., Thubert, P., Varga, B., and J. Farkas, 1136 "Deterministic Networking Architecture", draft-ietf- 1137 detnet-architecture-08 (work in progress), September 2018. 1139 [I-D.ietf-detnet-ip] 1140 Varga, B., Farkas, J., Berger, L., Fedyk, D., Malis, A., 1141 Bryant, S., and J. Korhonen, "DetNet Data Plane: IP", 1142 draft-ietf-detnet-ip-00 (work in progress), May 2019. 1144 [I-D.ietf-detnet-mpls] 1145 Varga, B., Farkas, J., Berger, L., Fedyk, D., Malis, A., 1146 Bryant, S., and J. Korhonen, "DetNet Data Plane: MPLS", 1147 draft-ietf-detnet-mpls-00 (work in progress), May 2019. 1149 [RFC2212] Shenker, S., Partridge, C., and R. Guerin, "Specification 1150 of Guaranteed Quality of Service", RFC 2212, 1151 DOI 10.17487/RFC2212, September 1997, 1152 . 1154 [RFC6658] Bryant, S., Ed., Martini, L., Swallow, G., and A. Malis, 1155 "Packet Pseudowire Encapsulation over an MPLS PSN", 1156 RFC 6658, DOI 10.17487/RFC6658, July 2012, 1157 . 1159 [RFC7806] Baker, F. and R. Pan, "On Queuing, Marking, and Dropping", 1160 RFC 7806, DOI 10.17487/RFC7806, April 2016, 1161 . 1163 [RFC8578] Grossman, E., Ed., "Deterministic Networking Use Cases", 1164 RFC 8578, DOI 10.17487/RFC8578, May 2019, 1165 . 1167 7.2. Informative References 1169 [bennett2002delay] 1170 J.C.R. Bennett, K. Benson, A. Charny, W.F. Courtney, and 1171 J.-Y. Le Boudec, "Delay Jitter Bounds and Packet Scale 1172 Rate Guarantee for Expedited Forwarding", 1173 . 1175 [charny2000delay] 1176 A. Charny and J.-Y. Le Boudec, "Delay Bounds in a Network 1177 with Aggregate Scheduling", . 1180 [IEEE8021Q] 1181 IEEE 802.1, "IEEE Std 802.1Q-2018: IEEE Standard for Local 1182 and metropolitan area networks - Bridges and Bridged 1183 Networks", 2018, 1184 . 1186 [IEEE8021Qcr] 1187 IEEE 802.1, "IEEE P802.1Qcr: IEEE Draft Standard for Local 1188 and metropolitan area networks - Bridges and Bridged 1189 Networks - Amendment: Asynchronous Traffic Shaping", 2017, 1190 . 1192 [IEEE8021TSN] 1193 IEEE 802.1, "IEEE 802.1 Time-Sensitive Networking (TSN) 1194 Task Group", . 1196 [IEEE8023] 1197 IEEE 802.3, "IEEE Std 802.3-2018: IEEE Standard for 1198 Ethernet", 2018, 1199 . 1201 [le_boudec_theory_2018] 1202 J.-Y. Le Boudec, "A Theory of Traffic Regulators for 1203 Deterministic Networks with Application to Interleaved 1204 Regulators", . 1206 [NetCalBook] 1207 Le Boudec, Jean-Yves, and Patrick Thiran, "Network 1208 calculus: a theory of deterministic queuing systems for 1209 the internet", 2001, . 1211 [Specht2016UBS] 1212 J. Specht and S. Samii, "Urgency-Based Scheduler for Time- 1213 Sensitive Switched Ethernet Networks", 1214 . 1216 [TSNwithATS] 1217 E. Mohammadpour, E. Stai, M. Mohiuddin, and J.-Y. Le 1218 Boudec, "End-to-end Latency and Backlog Bounds in Time- 1219 Sensitive Networking with Credit Based Shapers and 1220 Asynchronous Traffic Shaping", 1221 . 1223 Authors' Addresses 1225 Norman Finn 1226 Huawei Technologies Co. Ltd 1227 3101 Rio Way 1228 Spring Valley, California 91977 1229 US 1231 Phone: +1 925 980 6430 1232 Email: nfinn@nfinnconsulting.com 1234 Jean-Yves Le Boudec 1235 EPFL 1236 IC Station 14 1237 Lausanne EPFL 1015 1238 Switzerland 1240 Email: jean-yves.leboudec@epfl.ch 1242 Ehsan Mohammadpour 1243 EPFL 1244 IC Station 14 1245 Lausanne EPFL 1015 1246 Switzerland 1248 Email: ehsan.mohammadpour@epfl.ch 1250 Jiayi Zhang 1251 Huawei Technologies Co. Ltd 1252 Q22, No.156 Beiqing Road 1253 Beijing 100095 1254 China 1256 Email: zhangjiayi11@huawei.com 1257 Balazs Varga 1258 Ericsson 1259 Konyves Kalman krt. 11/B 1260 Budapest 1097 1261 Hungary 1263 Email: balazs.a.varga@ericsson.com 1265 Janos Farkas 1266 Ericsson 1267 Konyves Kalman krt. 11/B 1268 Budapest 1097 1269 Hungary 1271 Email: janos.farkas@ericsson.com