idnits 2.17.1 draft-ietf-mboned-ieee802-mcast-problems-05.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The document seems to lack a both a reference to RFC 2119 and the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords. RFC 2119 keyword, line 638: '... It is RECOMMENDED that multicast-en...' Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (April 15, 2019) is 1838 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Missing Reference: 'ICMPv6' is mentioned on line 557, but not defined == Outdated reference: A later version (-20) exists of draft-ietf-6lo-backbone-router-11 == Outdated reference: A later version (-30) exists of draft-ietf-6tisch-architecture-20 Summary: 1 error (**), 0 flaws (~~), 4 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Internet Area C. Perkins 3 Internet-Draft M. McBride 4 Intended status: Informational Futurewei 5 Expires: October 17, 2019 D. Stanley 6 HPE 7 W. Kumari 8 Google 9 JC. Zuniga 10 SIGFOX 11 April 15, 2019 13 Multicast Considerations over IEEE 802 Wireless Media 14 draft-ietf-mboned-ieee802-mcast-problems-05 16 Abstract 18 Well-known issues with multicast have prevented the deployment of 19 multicast in 802.11 and other local-area wireless environments. This 20 document offers guidance on known limitations and problems with 21 wireless multicast. Also described are certain multicast enhancement 22 features that have been specified by the IETF and by IEEE 802 for 23 wireless media, as well as some operational choices that can be taken 24 to improve the performace of the network. Finally, some 25 recommendations are provided about the usage and combination of these 26 features and operational choices. 28 Status of This Memo 30 This Internet-Draft is submitted in full conformance with the 31 provisions of BCP 78 and BCP 79. 33 Internet-Drafts are working documents of the Internet Engineering 34 Task Force (IETF). Note that other groups may also distribute 35 working documents as Internet-Drafts. The list of current Internet- 36 Drafts is at https://datatracker.ietf.org/drafts/current/. 38 Internet-Drafts are draft documents valid for a maximum of six months 39 and may be updated, replaced, or obsoleted by other documents at any 40 time. It is inappropriate to use Internet-Drafts as reference 41 material or to cite them other than as "work in progress." 43 This Internet-Draft will expire on October 17, 2019. 45 Copyright Notice 47 Copyright (c) 2019 IETF Trust and the persons identified as the 48 document authors. All rights reserved. 50 This document is subject to BCP 78 and the IETF Trust's Legal 51 Provisions Relating to IETF Documents 52 (https://trustee.ietf.org/license-info) in effect on the date of 53 publication of this document. Please review these documents 54 carefully, as they describe your rights and restrictions with respect 55 to this document. Code Components extracted from this document must 56 include Simplified BSD License text as described in Section 4.e of 57 the Trust Legal Provisions and are provided without warranty as 58 described in the Simplified BSD License. 60 Table of Contents 62 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 63 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 4 64 3. Identified multicast issues . . . . . . . . . . . . . . . . . 5 65 3.1. Issues at Layer 2 and Below . . . . . . . . . . . . . . . 5 66 3.1.1. Multicast reliability . . . . . . . . . . . . . . . . 5 67 3.1.2. Lower and Variable Data Rate . . . . . . . . . . . . 6 68 3.1.3. High Interference . . . . . . . . . . . . . . . . . . 6 69 3.1.4. Power-save Effects on Multicast . . . . . . . . . . . 7 70 3.2. Issues at Layer 3 and Above . . . . . . . . . . . . . . . 7 71 3.2.1. IPv4 issues . . . . . . . . . . . . . . . . . . . . . 8 72 3.2.2. IPv6 issues . . . . . . . . . . . . . . . . . . . . . 8 73 3.2.3. MLD issues . . . . . . . . . . . . . . . . . . . . . 8 74 3.2.4. Spurious Neighbor Discovery . . . . . . . . . . . . . 9 75 4. Multicast protocol optimizations . . . . . . . . . . . . . . 10 76 4.1. Proxy ARP in 802.11-2012 . . . . . . . . . . . . . . . . 10 77 4.2. IPv6 Address Registration and Proxy Neighbor Discovery . 10 78 4.3. Buffering to Improve Battery Life . . . . . . . . . . . . 12 79 4.4. IPv6 support in 802.11-2012 . . . . . . . . . . . . . . . 12 80 4.5. Using Unicast Instead of Multicast . . . . . . . . . . . 13 81 4.5.1. Overview . . . . . . . . . . . . . . . . . . . . . . 13 82 4.5.2. Layer 2 Conversion to Unicast . . . . . . . . . . . . 13 83 4.5.3. Directed Multicast Service (DMS) . . . . . . . . . . 13 84 4.5.4. Automatic Multicast Tunneling (AMT) . . . . . . . . . 14 85 4.6. GroupCast with Retries (GCR) . . . . . . . . . . . . . . 14 86 5. Operational optimizations . . . . . . . . . . . . . . . . . . 15 87 5.1. Mitigating Problems from Spurious Neighbor Discovery . . 15 88 5.2. Mitigating Spurious Service Discovery Messages . . . . . 17 89 6. Multicast Considerations for Other Wireless Media . . . . . . 17 90 7. Recommendations . . . . . . . . . . . . . . . . . . . . . . . 18 91 8. Discussion Items . . . . . . . . . . . . . . . . . . . . . . 18 92 9. Security Considerations . . . . . . . . . . . . . . . . . . . 19 93 10. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 19 94 11. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 19 95 12. Informative References . . . . . . . . . . . . . . . . . . . 19 96 Appendix A. Changes in this draft between revisions 04 versus 05 22 97 Appendix B. Changes in this draft between revisions 03 versus 04 23 98 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 23 100 1. Introduction 102 Well-known issues with multicast have prevented the deployment of 103 multicast in 802.11 [dot11], [mc-props], [mc-prob-stmt], and other 104 local-area wireless environments. Performance issues have been 105 observed when multicast packet transmissions of IETF protocols are 106 used over IEEE 802 wireless media. Even though enhancements for 107 multicast transmissions have been designed at both IETF and IEEE 802, 108 incompatibilities still exist between specifications, implementations 109 and configuration choices. 111 Many IETF protocols depend on multicast/broadcast for delivery of 112 control messages to multiple receivers. Multicast is used for 113 various purposes such as neighbor discovery, network flooding, 114 address resolution, as well minimizing media occupancy for the 115 transmission of data that is intended for multiple receivers. In 116 addition to protocol use of broadcast/multicast for control messages, 117 more applications, such as push to talk in hospitals, or video in 118 enterprises, universities, and homes, are sending multicast IP to end 119 user devices, which are increasingly using wifi for their 120 connectivity. 122 IETF protocols typically rely on network protocol layering in order 123 to reduce or eliminate any dependence of higher level protocols on 124 the specific nature of the MAC layer protocols or the physical media. 125 In the case of multicast transmissions, higher level protocols have 126 traditionally been designed as if transmitting a packet to an IP 127 address had the same cost in interference and network media access, 128 regardless of whether the destination IP address is a unicast address 129 or a multicast or broadcast address. This model was reasonable for 130 networks where the physical medium was wired, like Ethernet. 131 Unfortunately, for many wireless media, the costs to access the 132 medium can be quite different. Multicast over Wi-Fi has often been 133 plagued by such poor performance that it is disallowed. Some 134 enhancements have been designed in IETF protocols that are assumed to 135 work primarily over wireless media. However, these enhancements are 136 usually implemented in limited deployments and not widespread on most 137 wireless networks. 139 IEEE 802 wireless protocols have been designed with certain features 140 to support multicast traffic. For instance, lower modulations are 141 used to transmit multicast frames, so that these can be received by 142 all stations in the cell, regardless of the distance or path 143 attenuation from the base station or access point. However, these 144 lower modulation transmissions occupy the medium longer; they hamper 145 efficient transmission of traffic using higher order modulations to 146 nearby stations. For these and other reasons, IEEE 802 working 147 groups such as 802.11 have designed features to improve the 148 performance of multicast transmissions at Layer 2 [ietf_802-11]. In 149 addition to protocol design features, certain operational and 150 configuration enhancements can ameliorate the network performance 151 issues created by multicast traffic, as described in Section 5. 153 There seems to be general agreement that these problems will not be 154 fixed anytime soon, primarily because it's expensive to do so, and 155 multicast is unreliable. Compared to unicast over Wi-Fi, multicast 156 is often treated as somewhat a second class citizen, even though 157 there are many protocols using multicast. Something needs to be 158 provided in order to make them more reliable. IPv6 neighbor 159 discovery saturating the Wi-Fi link is only part of the problem. Wi- 160 Fi traffic classes may help. This document is intended to help make 161 the determination about what problems should be solved by the IETF 162 and what problems should be solved by the IEEE (see Section 8). 164 This document details various problems caused by multicast 165 transmission over wireless networks, including high packet error 166 rates, no acknowledgements, and low data rate. It also explains some 167 enhancements that have been designed at IETF and IEEE 802 to 168 ameliorate the effects of multicast traffic. Recommendations are 169 also provided to implementors about how to use and combine these 170 enhancements. Some advice about the operational choices that can be 171 taken is also included. It is likely that this document will also be 172 considered relevant to designers of future IEEE wireless 173 specifications. 175 2. Terminology 177 This document uses the following definitions: 179 ACK 180 The 802.11 layer 2 acknowledgement 182 AP 183 IEEE 802.11 Access Point 185 basic rate 186 The slowest rate of all the connected devices, at which multicast 187 and broadcast traffic is generally transmitted 189 DTIM 190 Delivery Traffic Indication Map (DTIM): An information element 191 that advertises whether or not any associated stations have 192 buffered multicast or broadcast frames 194 MCS 195 Modulation and Coding Scheme 197 NOC 198 Network Operations Center 200 PER 201 Packet Error Rate 203 STA 204 802.11 station (e.g. handheld device) 206 TIM 207 Traffic Indication Map (TIM): An information element that 208 advertises whether or not any associated stations have buffered 209 unicast frames 211 3. Identified multicast issues 213 3.1. Issues at Layer 2 and Below 215 In this section we describe some of the issues related to the use of 216 multicast transmissions over IEEE 802 wireless technologies. 218 3.1.1. Multicast reliability 220 Multicast traffic is typically much less reliable than unicast 221 traffic. Since multicast makes point-to-multipoint communications, 222 multiple acknowledgements would be needed to guarantee reception at 223 all recipients. Since typically there are no ACKs for multicast 224 packets, it is not possible for the Access Point (AP) to know whether 225 or not a retransmission is needed. Even in the wired Internet, this 226 characteristic often causes undesirably high error rates. This has 227 contributed to the relatively slow uptake of multicast applications 228 even though the protocols have long been available. The situation 229 for wireless links is much worse, and is quite sensitive to the 230 presence of background traffic. Consequently, there can be a high 231 packet error rate (PER) due to lack of retransmission, and because 232 the sender never backs off. It is not uncommon for there to be a 233 packet loss rate of 5% or more, which is particularly troublesome for 234 video and other environments where high data rates and high 235 reliability are required. 237 3.1.2. Lower and Variable Data Rate 239 Multicast over wired differs from multicast over wireless because 240 transmission over wired links often occurs at a fixed rate. Wi-Fi, 241 on the other hand, has a transmission rate which varies depending 242 upon the STA's proximity to the AP. The throughput of video flows, 243 and the capacity of the broader Wi-Fi network, will change and will 244 impact the ability for QoS solutions to effectively reserve bandwidth 245 and provide admission control. 247 For wireless stations associated with an Access Point, the power 248 necessary for good reception can vary from station to station. For 249 unicast, the goal is to minimize power requirements while maximizing 250 the data rate to the destination. For multicast, the goal is simply 251 to maximize the number of receivers that will correctly receive the 252 multicast packet; generally the Access Point has to use a much lower 253 data rate at a power level high enough for even the farthest station 254 to receive the packet, for example as briefly mentioned in [RFC5757]. 255 Consequently, the data rate of a video stream, for instance, would be 256 constrained by the environmental considerations of the least reliable 257 receiver associated with the Access Point. 259 Because more robust modulation and coding schemes (MCSs) have longer 260 range but also lower data rate, multicast / broadcast traffic is 261 generally transmitted at the slowest rate of all the connected 262 devices, also known as the basic rate. The amount of additional 263 interference depends on the specific wireless technology. In fact 264 backward compatibility and multi-stream implementations mean that the 265 maximum unicast rates are currently up to a few Gb/s, so there can be 266 a more than 3 orders of magnitude difference in the transmission rate 267 between multicast / broadcast versus optimal unicast forwarding. 268 Some techinues employed to increase spectral efficiency, such as 269 spatial multiplexing in mimo systems, are not available with more 270 than one intended reciever; it is not the case that backwards 271 compatibility is the only factor responsible for lower multicast 272 transmission rates. 274 Wired multicast also affects wireless LANs when the AP extends the 275 wired segment; in that case, multicast / broadcast frames on the 276 wired LAN side are copied to WLAN. Since broadcast messages are 277 transmitted at the most robust MCS, many large frames are sent at a 278 slow rate over the air. 280 3.1.3. High Interference 282 Transmissions at a lower rate require longer occupancy of the 283 wireless medium and thus take away from the airtime of other 284 communications and degrade the overall capacity. Furthermore, 285 transmission at higher power, as is required to reach all multicast 286 STAs associated to the AP, proportionately increases the area of 287 interference. 289 3.1.4. Power-save Effects on Multicast 291 One of the characteristics of multicast transmission is that every 292 station has to be configured to wake up to receive the multicast, 293 even though the received packet may ultimately be discarded. This 294 process can have a large effect on the power consumption by the 295 multicast receiver station. 297 Multicast can work poorly with the power-save mechanisms defined in 298 IEEE 802.11e, for the following reasons. 300 o Clients may be unable to stay in sleep mode due to multicast 301 control packets frequently waking them up. 302 o Both unicast and multicast traffic can be delayed by power-saving 303 mechanisms. 304 o A unicast packet is delayed until a STA wakes up and requests it. 305 Unicast traffic may also be delayed to improve power save, 306 efficiency and increase probability of aggregation. 307 o Multicast traffic is delayed in a wireless network if any of the 308 STAs in that network are power savers. All STAs associated to the 309 AP have to be awake at a known time to receive multicast traffic. 310 o Packets can also be discarded due to buffer limitations in the AP 311 and non-AP STA. 313 3.2. Issues at Layer 3 and Above 315 This section identifies some representative IETF protocols, and 316 describes possible negative effects due to performance degradation 317 when using multicast transmissions for control messages. Common uses 318 of multicast include: 320 o Control plane signaling 321 o Neighbor Discovery 322 o Address Resolution 323 o Service discovery 324 o Applications (video delivery, stock data, etc.) 325 o On-demand routing 326 o Backbone construction 327 o Other L3 protocols (non-IP) 329 User Datagram Protocol (UDP) is the most common transport layer 330 protocol for multicast applications. By itself, UDP is not reliable 331 -- messages may be lost or delivered out of order. 333 3.2.1. IPv4 issues 335 The following list contains some representative multicast protocols 336 that are used with IPv4. 338 o ARP 339 o DHCP 340 o mDNS [RFC6762] 341 o uPnP [RFC6970] 343 After initial configuration, ARP and DHCP occur much less commonly, 344 but service discovery can occur at any time. Some widely-deployed 345 service discovery protocols (e.g., for finding a printer) utilize 346 mDNS (i.e., multicast). It's often the first service that operators 347 drop. Even if multicast snooping is utilized, many devices can 348 register at once and cause serious network degradation. 350 3.2.2. IPv6 issues 352 IPv6 makes extensive use of multicast, including the following: 354 o DHCPv6 355 o IPv6 Neighbor Discovery Protocol (NDP) 356 o Duplicate Address Detection (DAD) 357 o Address Resolution 358 o Service Discovery 359 o Route Discovery 360 o Decentralized Address Assignment 361 o Geographic routing 363 IPv6 NDP Neighbor Solicitation (NS) messages used in DAD and Address 364 Lookup make use of Link-Scope multicast. In contrast to IPv4, an 365 IPv6 node will typically use multiple addresses, and may change them 366 often for privacy reasons. This intensifies the impact of multicast 367 messages that are associated to the mobility of a node. Router 368 advertisement (RA) messages are also periodically multicasted over 369 the Link. 371 Neighbors may be considered lost if several consecutive Neighbor 372 Discovery packets fail. 374 3.2.3. MLD issues 376 Multicast Listener Discovery(MLD) [RFC4541] is often used to identify 377 members of a multicast group that are connected to the ports of a 378 switch. Forwarding multicast frames into a Wi-Fi-enabled area can 379 use such switch support for hardware forwarding state information. 380 However, since IPv6 makes heavy use of multicast, each STA with an 381 IPv6 address will require state on the switch for several and 382 possibly many multicast solicited-node addresses. Multicast 383 addresses that do not have forwarding state installed (perhaps due to 384 hardware memory limitations on the switch) cause frames to be flooded 385 on all ports of the switch. 387 3.2.4. Spurious Neighbor Discovery 389 On the Internet there is a "background radiation" of scanning traffic 390 (people scanning for vulnerable machines) and backscatter (responses 391 from spoofed traffic, etc). This means that routers very often 392 receive packets destined for IP addresses regardless of whether they 393 are in use. In the cases where the IP is assigned to a host, the 394 router broadcasts an ARP request, gets back an ARP reply, and caches 395 it; then traffic can be delivered to the host. When the IP address 396 is not in use, the router broadcasts one (or more) ARP requests, and 397 never gets a reply. This means that it does not populate the ARP 398 cache, and the next time there is traffic for that IP address the 399 router will rebroadcast the ARP requests. 401 The rate of these ARP requests is proportional to the size of the 402 subnets, the rate of scanning and backscatter, and how long the 403 router keeps state on non-responding ARPs. As it turns out, this 404 rate is inversely proportional to how occupied the subnet is (valid 405 ARPs end up in a cache, stopping the broadcasting; unused IPs never 406 respond, and so cause more broadcasts). Depending on the address 407 space in use, the time of day, how occupied the subnet is, and other 408 unknown factors, on the order of 2000 broadcasts per second have been 409 observed, for instance at the NOCs during IETF face-to-face meetings. 411 On a wired network, there is not a huge difference between unicast, 412 multicast and broadcast traffic. Due to hardware filtering (see, 413 e.g., [Deri-2010]), inadvertently flooded traffic (or high amounts of 414 ethernet multicast) on wired networks can be quite a bit less costly, 415 compared to wireless cases where sleeping devices have to wake up to 416 process packets. Wired Ethernets tend to be switched networks, 417 further reducing interference from multicast. There is effectively 418 no collision / scheduling problem except at extremely high port 419 utilizations. 421 This is not true in the wireless realm; wireless equipment is often 422 unable to send high volumes of broadcast and multicast traffic. 423 Consequently, on the wireless networks, we observe a significant 424 amount of dropped broadcast and multicast packets. This, in turn, 425 means that when a host connects it is often not able to complete 426 DHCP, and IPv6 RAs get dropped, leading to users being unable to use 427 the network. 429 4. Multicast protocol optimizations 431 This section lists some optimizations that have been specified in 432 IEEE 802 and IETF that are aimed at reducing or eliminating the 433 issues discussed in Section 3. 435 4.1. Proxy ARP in 802.11-2012 437 The AP knows the MAC address and IP address for all associated STAs. 438 In this way, the AP acts as the central "manager" for all the 802.11 439 STAs in its BSS. Proxy ARP is easy to implement at the AP, and 440 offers the following advantages: 442 o Reduced broadcast traffic (transmitted at low MCS) on the wireless 443 medium 444 o STA benefits from extended power save in sleep mode, as ARP 445 requests for STA's IP address are handled instead by the AP. 446 o ARP frames are kept off the wireless medium. 447 o No changes are needed to STA implementation. 449 Here is the specification language as described in clause 10.23.13 of 450 [dot11-proxyarp]: 452 When the AP supports Proxy ARP "[...] the AP shall maintain a 453 Hardware Address to Internet Address mapping for each associated 454 station, and shall update the mapping when the Internet Address of 455 the associated station changes. When the IPv4 address being 456 resolved in the ARP request packet is used by a non-AP STA 457 currently associated to the BSS, the proxy ARP service shall 458 respond on behalf of the non-AP STA". 460 4.2. IPv6 Address Registration and Proxy Neighbor Discovery 462 As used in this section, a Low-Power Wireless Personal Area Network 463 (6LoWPAN) denotes a low power lossy network (LLN) that supports 464 6LoWPAN Header Compression (HC) [RFC6282]. A 6TiSCH network 465 [I-D.ietf-6tisch-architecture] is an example of a 6LowPAN. In order 466 to control the use of IPv6 multicast over 6LoWPANs, the 6LoWPAN 467 Neighbor Discovery (6LoWPAN ND) [RFC6775] standard defines an address 468 registration mechanism that relies on a central registry to assess 469 address uniqueness, as a substitute to the inefficient Duplicate 470 Address Detection (DAD) mechanism found in the mainstream IPv6 471 Neighbor Discovery Protocol (NDP) [RFC4861][RFC4862]. 473 The 6lo Working Group has specified an update [RFC8505] to RFC6775. 474 Wireless devices can register their address to a Backbone Router 475 [I-D.ietf-6lo-backbone-router], which proxies for the registered 476 addresses with the IPv6 NDP running on a high speed aggregating 477 backbone. The update also enables a proxy registration mechanism on 478 behalf of the registered node, e.g. by a 6LoWPAN router to which the 479 mobile node is attached. 481 The general idea behind the backbone router concept is that broadcast 482 and multicast messaging should be tightly controlled in a variety of 483 Wireless Local Area Networks (WLANs) and Wireless Personal Area 484 Networks (WPANs). Connectivity to a particular link that provides 485 the subnet should be left to Layer-3. The model for the Backbone 486 Router operation is represented in Figure 1. 488 | 489 +-----+ 490 | | Gateway (default) router 491 | | 492 +-----+ 493 | 494 | Backbone Link 495 +--------------------+------------------+ 496 | | | 497 +-----+ +-----+ +-----+ 498 | | Backbone | | Backbone | | Backbone 499 | | router 1 | | router 2 | | router 3 500 +-----+ +-----+ +-----+ 501 o o o o o o 502 o o o o o o o o o o o o o o 503 o o o o o o o o o o o o o o o 504 o o o o o o o o o o 505 o o o o o o o 507 LLN 1 LLN 2 LLN 3 509 Figure 1: Backbone Link and Backbone Routers 511 LLN nodes can move freely from an LLN anchored at one IPv6 Backbone 512 Router to an LLN anchored at another Backbone Router on the same 513 backbone, keeping any of the IPv6 addresses they have configured. 514 The Backbone Routers maintain a Binding Table of their Registered 515 Nodes, which serves as a distributed database of all the LLN Nodes. 516 An extension to the Neighbor Discovery Protocol is introduced to 517 exchange Binding Table information across the Backbone Link as needed 518 for the operation of IPv6 Neighbor Discovery. 520 RFC6775 and follow-on work [RFC8505] address the needs of LLNs, and 521 similar techniques are likely to be valuable on any type of link 522 where sleeping devices are attached, or where the use of broadcast 523 and multicast operations should be limited. 525 4.3. Buffering to Improve Battery Life 527 Methods have been developed to help save battery life; for example, a 528 device might not wake up when the AP receives a multicast packet. 529 The AP acts on behalf of STAs in various ways. To enable use of the 530 power-saving feature for STAs in its BSS, the AP buffers frames for 531 delivery to the STA at the time when the STA is scheduled for 532 reception. If an AP, for instance, expresses a DTIM (Delivery 533 Traffic Indication Message) of 3 then the AP will send a multicast 534 packet every 3 packets. In fact, when any single wireless STA 535 associated with an access point has 802.11 power-save mode enabled, 536 the access point buffers all multicast frames and sends them only 537 after the next DTIM beacon. 539 In practice, most AP's will send a multicast every 30 packets. For 540 unicast the AP could send a TIM (Traffic Indication Message), but for 541 multicast the AP sends a broadcast to everyone. DTIM does power 542 management but STAs can choose whether or not to wake up or not and 543 whether or not to drop the packet. Unfortunately, without proper 544 administrative control, such STAs may be unable to determine why 545 their multicast operations do not work. 547 4.4. IPv6 support in 802.11-2012 549 IPv6 uses Neighbor Discovery Protocol (NDP) instead of ARP. Every 550 IPv6 node subscribes to a special multicast address for this purpose. 552 Here is the specification language from clause 10.23.13 of 553 [dot11-proxyarp]: 555 "When an IPv6 address is being resolved, the Proxy Neighbor 556 Discovery service shall respond with a Neighbor Advertisement 557 message [...] on behalf of an associated STA to an [ICMPv6] 558 Neighbor Solicitation message [...]. When MAC address mappings 559 change, the AP may send unsolicited Neighbor Advertisement 560 Messages on behalf of a STA." 562 NDP may be used to request additional information 564 o Maximum Transmission Unit 565 o Router Solicitation 566 o Router Advertisement, etc. 568 NDP messages are sent as group addressed (broadcast) frames in 569 802.11. Using the proxy operation helps to keep NDP messages off the 570 wireless medium. 572 4.5. Using Unicast Instead of Multicast 574 It is often possible to transmit multicast control and data messages 575 by using unicast transmissions to each station individually. 577 4.5.1. Overview 579 In many situations, it's a good choice to use unicast instead of 580 multicast over the Wi-Fi link. This avoids most of the problems 581 specific to multicast over Wi-Fi, since the individual frames are 582 then acknowledged and buffered for power save clients, in the way 583 that unicast traffic normally operates. 585 This approach comes with the tradeoff of sometimes sending the same 586 packet multiple times over the Wi-Fi link. However, in many cases, 587 such as video into a residential home network, this can be a good 588 tradeoff, since the Wi-Fi link may have enough capacity for the 589 unicast traffic to be transmitted to each subscribed STA, even though 590 multicast addressing may have been necessary for the upstream access 591 network. 593 Several technologies exist that can be used to arrange unicast 594 transport over the Wi-Fi link, outlined in the subsections below. 596 4.5.2. Layer 2 Conversion to Unicast 598 It is often possible to transmit multicast control and data messages 599 by using unicast transmissions to each station individually. 601 Although there is not yet a standardized method of conversion, at 602 least one widely available implementation exists in the Linux 603 bridging code [bridge-mc-2-uc]. Other proprietary implementations 604 are available from various vendors. In general, these 605 implementations perform a straightforward mapping for groups or 606 channels, discovered by IGMP or MLD snooping, to the corresponding 607 unicast MAC addresses. 609 4.5.3. Directed Multicast Service (DMS) 611 There are situations where more is needed than simply converting 612 multicast to unicast. For these purposes, DMS enables a STA to 613 request that the AP transmit multicast group addressed frames 614 destined to the requesting STAs as individually addressed frames 615 [i.e., convert multicast to unicast]. Here are some characteristics 616 of DMS: 618 o Requires 802.11n A-MSDUs 619 o Individually addressed frames are acknowledged and are buffered 620 for power save STAs 621 o The requesting STA may specify traffic characteristics for DMS 622 traffic 623 o DMS was defined in IEEE Std 802.11v-2011 624 o DMS requires changes to both AP and STA implementation. 626 DMS is not currently implemented in products. See [Tramarin2017] and 627 [Oliva2013] for more information. 629 4.5.4. Automatic Multicast Tunneling (AMT) 631 AMT[RFC7450] provides a method to tunnel multicast IP packets inside 632 unicast IP packets over network links that only support unicast. 633 When an operating system or application running on a STA has an AMT 634 gateway capability integrated, it's possible to use unicast to 635 traverse the Wi-Fi link by deploying an AMT relay in the non-Wi-Fi 636 portion of the network connected to the AP. 638 It is RECOMMENDED that multicast-enabled networks deploying AMT 639 relays for this purpose make the relays discoverable with both of 640 these methods: 642 o the well-known IP addresses from Section 7 of [RFC7450], and 643 o with DNS-SD [RFC6763] 645 Providing the multiple standard discovery methods makes it more 646 likely that AMT gateway implementations will discover the local 647 multicast-capable network, rather than forming a connection to an AMT 648 relay further upstream. 650 4.6. GroupCast with Retries (GCR) 652 GCR (defined in [dot11aa]) provides greater reliability by using 653 either unsolicited retries or a block acknowledgement mechanism. GCR 654 increases probability of broadcast frame reception success, but still 655 does not guarantee success. 657 For the block acknowledgement mechanism, the AP transmits each group 658 addressed frame as conventional group addressed transmission. 659 Retransmissions are group addressed, but hidden from non-11aa STAs. 660 A directed block acknowledgement scheme is used to harvest reception 661 status from receivers; retransmissions are based upon these 662 responses. 664 GCR is suitable for all group sizes including medium to large groups. 665 As the number of devices in the group increases, GCR can send block 666 acknowledgement requests to only a small subset of the group. GCR 667 does require changes to both AP and STA implementation. 669 GCR may introduce unacceptable latency. After sending a group of 670 data frames to the group, the AP has do the following: 672 o unicast a Block Ack Request (BAR) to a subset of members. 673 o wait for the corresponding Block Ack (BA). 674 o retransmit any missed frames. 675 o resume other operations which may have been delayed. 677 This latency may not be acceptable for some traffic. 679 There are ongoing extensions in 802.11 to improve GCR performance. 681 o BAR is sent using downlink MU-MIMO (note that downlink MU-MIMO is 682 already specified in 802.11-REVmc 4.3). 683 o BA is sent using uplink MU-MIMO (which is a .11ax feature). 684 o Additional 802.11ax extensions are under consideration; see 685 [mc-ack-mux] 686 o Latency may also be reduced by simultaneously receiving BA 687 information from multiple STAs. 689 5. Operational optimizations 691 This section lists some operational optimizations that can be 692 implemented when deploying wireless IEEE 802 networks to mitigate the 693 issues discussed in Section 3. 695 5.1. Mitigating Problems from Spurious Neighbor Discovery 697 ARP Sponges 699 An ARP Sponge sits on a network and learn which IPs addresses 700 are actually in use. It also listen for ARP requests, and, if 701 it sees an ARP for an IP address which it believes is not used, 702 it will reply with its own MAC address. This means that the 703 router now has an IP to MAC mapping, which it caches. If that 704 IP is later assigned to an machine (e.g using DHCP), the ARP 705 sponge will see this, and will stop replying for that address. 706 Gratuitous ARPs (or the machine ARPing for its gateway) will 707 replace the sponged address in the router ARP table. This 708 technique is quite effective; but, unfortunately, the ARP 709 sponge daemons were not really designed for this use (the 710 standard one [arpsponge], was designed to deal with the 711 disappearance of participants from an IXP) and so are not 712 optimized for this purpose. We have to run one daemon per 713 subnet, the tuning is tricky (the scanning rate versus the 714 population rate versus retires, etc.) and sometimes the daemons 715 just seem to stop, requiring a restart of the daemon and 716 causing disruption. 718 Router mitigations 720 Some routers (often those based on Linux) implement a "negative 721 ARP cache" daemon. Simply put, if the router does not see a 722 reply to an ARP it can be configured to cache this information 723 for some interval. Unfortunately, the core routers which we 724 are using do not support this. When a host connects to network 725 and gets an IP address, it will ARP for its default gateway 726 (the router). The router will update its cache with the IP to 727 host MAC mapping learnt from the request (passive ARP 728 learning). 730 Firewall unused space 732 The distribution of users on wireless networks / subnets 733 changes from meeting to meeting (e.g SSIDs are renamed, some 734 SSIDs lose favor, etc). This makes utilization for particular 735 SSIDs difficult to predict ahead of time, but usage can be 736 monitored as attendees use the different networks. Configuring 737 multiple DHCP pools per subnet, and enabling them sequentially, 738 can create a large subnet, from which only addresses in the 739 lower portions are assigned. Therefore input IP access lists 740 can be applied, which deny traffic to the upper, unused 741 portions. Then the router does not attempt to forward packets 742 to the unused portions of the subnets, and so does not ARP for 743 it. This method has proven to be very effective, but is 744 somewhat of a blunt axe, is fairly labor intensive, and 745 requires coordination. 747 Disabling/filtering ARP requests 749 In general, the router does not need to ARP for hosts; when a 750 host connects, the router can learn the IP to MAC mapping from 751 the ARP request sent by that host. This means that we should 752 be able to disable and / or filter ARP requests from the 753 router. Unfortunately, ARP is a very low level / fundamental 754 part of the IP stack, and is often offloaded from the normal 755 control plane. While many routers can filter layer-2 traffic, 756 this is usually implemented as an input filter and / or has 757 limited ability to filter output broadcast traffic. This means 758 that the simple "just disable ARP or filter it outbound" seems 759 like a really simple (and obvious) solution, but 760 implementations / architectural issues make this difficult or 761 awkward in practice. 763 NAT 765 The broadcasts are overwhelmingly being caused by outside 766 scanning / backscatter traffic. This means that, if we were to 767 NAT the entire (or a large portion) of the attendee networks, 768 there would be no NAT translation entries for unused addresses, 769 and so the router would never ARP for them. However, there are 770 many reasons to avoid using NAT in such a blanket fashion. 772 Stateful firewalls 774 Another obvious solution would be to put a stateful firewall 775 between the wireless network and the Internet. This firewall 776 would block incoming traffic not associated with an outbound 777 request. But this conflicts with the need and desire to have 778 the network as open as possible and to honor the end-to-end 779 principle. An attendee on the meeting network should be an 780 Internet host, and should be able to receive unsolicited 781 requests. Unfortunately, keeping the network working and 782 stable is the first priority and a stateful firewall may be 783 required in order to achieve this. 785 5.2. Mitigating Spurious Service Discovery Messages 787 In networks that must support hundreds of STAs, operators have 788 observed network degradation due to many devices simultaneously 789 registering with mDNS. In a network with many clients, it is 790 recommended to ensure that mDNS packets designed to discover 791 services in smaller home networks be constrained to avoid 792 disrupting other traffic. 794 6. Multicast Considerations for Other Wireless Media 796 Many of the causes of performance degradation described in earlier 797 sections are also observable for wireless media other than 802.11. 799 For instance, problems with power save, excess media occupancy, and 800 poor reliability will also affect 802.15.3 and 802.15.4. 801 Unfortunately, 802.15 media specifications do not yet include 802 mechanisms similar to those developed for 802.11. In fact, the 803 design philosophy for 802.15 is oriented towards minimality, with the 804 result that many such functions are relegated to operation within 805 higher layer protocols. This leads to a patchwork of non- 806 interoperable and vendor-specific solutions. See [uli] for some 807 additional discussion, and a proposal for a task group to resolve 808 similar issues, in which the multicast problems might be considered 809 for mitigation. 811 Similar considerations hold for most other wireless media. A brief 812 introduction is provided in [RFC5757] for the following: 814 o 802.16 WIMAX 815 o 3GPP/3GPP2 816 o DVB-H / DVB-IPDC 817 o TV Broadcast and Satellite Networks 819 7. Recommendations 821 This section will provide some recommendations about the usage and 822 combinations of the multicast enhancements described in Section 4 and 823 Section 5. 825 Future protocol documents utilizing multicast signaling should be 826 carefully scrutinized if the protocol is likely to be used over 827 wireless media. 829 Proxy methods should be encouraged to conserve network bandwidth and 830 power utilization by low-power devices. The device can use a unicast 831 message to its proxy, and then the proxy can take care of any needed 832 multicast operations. 834 Multicast signaling for wireless devices should be done in a way 835 compatible with low duty-cycle operation. 837 8. Discussion Items 839 This section suggests two discussion items for further resolution. 841 The IETF should determine guidelines by which it may be decided that 842 multicast packets are to be sent wired. For example, 802.1ak works 843 on ethernet and Wi-Fi. 802.1ak has been pulled into 802.1Q as of 844 802.1Q-2011. 802.1Q-2014 can be found here: 845 http://www.ieee802.org/1/pages/802.1Q-2014.html. If a generic 846 solution is not found, guidelines for multicast over Wi-Fi should be 847 established. 849 Reliable registration to Layer-2 multicast groups and a reliable 850 multicast operation at Layer-2 might provide a generic solution. 851 There is no need to support 2^24 groups to get solicited node 852 multicast working: it is possible to simply select a number of 853 trailing bits that make sense for a given network size to limit the 854 amount of unwanted deliveries to reasonable levels. IEEE 802.1, 855 802.11, and 802.15 should be encouraged to revisit L2 multicast 856 issues. In reality, Wi-Fi provides a broadcast service, not a 857 multicast service. On the physical medium, all frames are broadcast 858 except in very unusual cases in which special beamforming 859 transmitters are used. Unicast offers the advantage of being much 860 faster (2 orders of magnitude) and much more reliable (L2 ARQ). 862 9. Security Considerations 864 This document neither introduces nor modifies any security 865 mechanisms. 867 10. IANA Considerations 869 This document does not request any IANA actions. 871 11. Acknowledgements 873 This document has benefitted from discussions with the following 874 people, in alphabetical order: Mikael Abrahamsson, Stuart Cheshire, 875 Donald Eastlake, Toerless Eckert, Jake Holland, Joel Jaeggli, Jan 876 Komissar, David Lamparter, Pascal Thubert 878 12. Informative References 880 [arpsponge] 881 Vijn, A. and S. Bakker, "Arp Sponge", March 2015, 882 . 885 [bridge-mc-2-uc] 886 Torvalds, L., "bridge: multicast to unicast", Jan 2017, 887 . 890 [Deri-2010] 891 Deri, L. and J. Gasparakis, "10 Gbit Hardware Packet 892 Filtering Using Commodity Network Adapters", RIPE 61, 893 2010, . 896 [dot11] "IEEE 802 Wireless", "802.11-2016 - IEEE Standard for 897 Information technology--Telecommunications and information 898 exchange between systems Local and metropolitan area 899 networks--Specific requirements - Part 11: Wireless LAN 900 Medium Access Control (MAC) and Physical Layer (PHY) 901 Specification", March 2016, 902 . 905 [dot11-proxyarp] 906 "IEEE 802 Wireless P802.11", "IEEE 802 Wireless P802.11", 907 and "IEEE 802 Wireless P802.11", "Proxy ARP in 802.11ax", 908 September 2015, . 911 [dot11aa] "IEEE 802 Wireless", "Part 11: Wireless LAN Medium Access 912 Control (MAC) and Physical Layer (PHY) Specifications 913 Amendment 2: MAC Enhancements for Robust Audio Video 914 Streaming", March 2012, 915 . 918 [I-D.ietf-6lo-backbone-router] 919 Thubert, P., Perkins, C., and E. Levy-Abegnoli, "IPv6 920 Backbone Router", draft-ietf-6lo-backbone-router-11 (work 921 in progress), February 2019. 923 [I-D.ietf-6tisch-architecture] 924 Thubert, P., "An Architecture for IPv6 over the TSCH mode 925 of IEEE 802.15.4", draft-ietf-6tisch-architecture-20 (work 926 in progress), March 2019. 928 [ietf_802-11] 929 Stanley, D., "IEEE 802.11 multicast capabilities", Nov 930 2015, . 934 [mc-ack-mux] 935 Tanaka, Y., Sakai, E., Morioka, Y., Mori, M., Hiertz, G., 936 and S. Coffey, "Multiplexing of Acknowledgements for 937 Multicast Transmission", July 2015, 938 . 942 [mc-prob-stmt] 943 Abrahamsson, M. and A. Stephens, "Multicast on 802.11", 944 March 2015, . 947 [mc-props] 948 Stephens, A., "IEEE 802.11 multicast properties", March 949 2015, . 953 [Oliva2013] 954 de la Oliva, A., Serrano, P., Salvador, P., and A. Banchs, 955 "Performance evaluation of the IEEE 802.11aa multicast 956 mechanisms for video streaming", 2013 IEEE 14th 957 International Symposium on "A World of Wireless, Mobile 958 and Multimedia Networks" (WoWMoM) pp. 1-9, June 2013. 960 [RFC4541] Christensen, M., Kimball, K., and F. Solensky, 961 "Considerations for Internet Group Management Protocol 962 (IGMP) and Multicast Listener Discovery (MLD) Snooping 963 Switches", RFC 4541, DOI 10.17487/RFC4541, May 2006, 964 . 966 [RFC4861] Narten, T., Nordmark, E., Simpson, W., and H. Soliman, 967 "Neighbor Discovery for IP version 6 (IPv6)", RFC 4861, 968 DOI 10.17487/RFC4861, September 2007, 969 . 971 [RFC4862] Thomson, S., Narten, T., and T. Jinmei, "IPv6 Stateless 972 Address Autoconfiguration", RFC 4862, 973 DOI 10.17487/RFC4862, September 2007, 974 . 976 [RFC5757] Schmidt, T., Waehlisch, M., and G. Fairhurst, "Multicast 977 Mobility in Mobile IP Version 6 (MIPv6): Problem Statement 978 and Brief Survey", RFC 5757, DOI 10.17487/RFC5757, 979 February 2010, . 981 [RFC6282] Hui, J., Ed. and P. Thubert, "Compression Format for IPv6 982 Datagrams over IEEE 802.15.4-Based Networks", RFC 6282, 983 DOI 10.17487/RFC6282, September 2011, 984 . 986 [RFC6762] Cheshire, S. and M. Krochmal, "Multicast DNS", RFC 6762, 987 DOI 10.17487/RFC6762, February 2013, 988 . 990 [RFC6763] Cheshire, S. and M. Krochmal, "DNS-Based Service 991 Discovery", RFC 6763, DOI 10.17487/RFC6763, February 2013, 992 . 994 [RFC6775] Shelby, Z., Ed., Chakrabarti, S., Nordmark, E., and C. 995 Bormann, "Neighbor Discovery Optimization for IPv6 over 996 Low-Power Wireless Personal Area Networks (6LoWPANs)", 997 RFC 6775, DOI 10.17487/RFC6775, November 2012, 998 . 1000 [RFC6970] Boucadair, M., Penno, R., and D. Wing, "Universal Plug and 1001 Play (UPnP) Internet Gateway Device - Port Control 1002 Protocol Interworking Function (IGD-PCP IWF)", RFC 6970, 1003 DOI 10.17487/RFC6970, July 2013, 1004 . 1006 [RFC7450] Bumgardner, G., "Automatic Multicast Tunneling", RFC 7450, 1007 DOI 10.17487/RFC7450, February 2015, 1008 . 1010 [RFC8505] Thubert, P., Ed., Nordmark, E., Chakrabarti, S., and C. 1011 Perkins, "Registration Extensions for IPv6 over Low-Power 1012 Wireless Personal Area Network (6LoWPAN) Neighbor 1013 Discovery", RFC 8505, DOI 10.17487/RFC8505, November 2018, 1014 . 1016 [Tramarin2017] 1017 Tramarin, F., Vitturi, S., and M. Luvisotto, "IEEE 802.11n 1018 for Distributed Measurement Systems", 2017 IEEE 1019 International Instrumentation and Measurement Technology 1020 Conference (I2MTC) pp. 1-6, May 2017. 1022 [uli] Kinney, P., "LLC Proposal for 802.15.4", Nov 2015, 1023 . 1026 Appendix A. Changes in this draft between revisions 04 versus 05 1028 This section lists the changes between revisions ...-04.txt and 1029 ...-05.txt of draft-ietf-mboned-ieee802-mcast-problems. 1031 o Incorporated text from Jake Holland for a new section about 1032 conversion of multicast to unicast and included AMT as an existing 1033 solution. 1034 o Included some text about likely future multicast applications that 1035 will emphasize the need for attention to the technical matters 1036 collected in this document. 1037 o Further modified text to be more generic instead of referring 1038 specifically to IETF conference situations. 1039 o Modified text to be more generic instead of referring specifically 1040 to Bonjour. 1041 o Added uPnP as a representative multicast protocol in IP networks. 1042 o Referred to Linux bridging code for multicast to unicast. 1043 o Updated bibliographic citations, included URLs as needed. 1044 o More editorial improvements and grammatical corrections. 1046 Appendix B. Changes in this draft between revisions 03 versus 04 1048 This section lists the changes between revisions ...-03.txt and 1049 ...-04.txt of draft-ietf-mboned-ieee802-mcast-problems. 1051 o Replaced "client" by "STA". 1052 o Used terminology "Wi-Fi" throughout. 1053 o Many editorial improvements and grammatical corrections. 1054 o Modified text to be more generic instead of referring specifically 1055 to IETF conference situations. 1056 o Cited [RFC5757] for introduction to other wireless media. 1057 o Updated bibliographic citations. 1059 Authors' Addresses 1061 Charles E. Perkins 1062 Futurewei Inc. 1063 2330 Central Expressway 1064 Santa Clara, CA 95050 1065 USA 1067 Phone: +1-408-330-4586 1068 Email: charliep@computer.org 1070 Mike McBride 1071 Futurewei Inc. 1072 2330 Central Expressway 1073 Santa Clara, CA 95055 1074 USA 1076 Email: michael.mcbride@huawei.com 1078 Dorothy Stanley 1079 Hewlett Packard Enterprise 1080 2000 North Naperville Rd. 1081 Naperville, IL 60566 1082 USA 1084 Phone: +1 630 979 1572 1085 Email: dstanley@arubanetworks.com 1086 Warren Kumari 1087 Google 1088 1600 Amphitheatre Parkway 1089 Mountain View, CA 94043 1090 USA 1092 Email: warren@kumari.net 1094 Juan Carlos Zuniga 1095 SIGFOX 1096 425 rue Jean Rostand 1097 Labege 31670 1098 France 1100 Email: j.c.zuniga@ieee.org