idnits 2.17.1 draft-ietf-bess-evpn-igmp-mld-proxy-06.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 : ---------------------------------------------------------------------------- ** There is 1 instance of too long lines in the document, the longest one being 7 characters in excess of 72. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == Using lowercase 'not' together with uppercase 'MUST', 'SHALL', 'SHOULD', or 'RECOMMENDED' is not an accepted usage according to RFC 2119. Please use uppercase 'NOT' together with RFC 2119 keywords (if that is what you mean). Found 'MUST not' in this paragraph: 3. When the first hop PE receives an IGMP version-X Join first for (*,G) and then later it receives an IGMP version-Y Join for the same (*,G), then it MUST re-advertise the same EVPN SMET route with flag for version-Y set in addition to any previously-set version flag(s). In other words, the first hop PE MUST not withdraw the EVPN route before sending the new route because the flag field is not part of BGP route key processing. == Using lowercase 'not' together with uppercase 'MUST', 'SHALL', 'SHOULD', or 'RECOMMENDED' is not an accepted usage according to RFC 2119. Please use uppercase 'NOT' together with RFC 2119 keywords (if that is what you mean). Found 'MUST not' in this paragraph: 5. When a PE receives an EVPN SMET route with more than one version flag set, it will generate the corresponding IGMP report for (*,G) for each version specified in the flags field. With multiple version flags set, there MUST not be source IP address in the receive EVPN route. If there is, then an error SHOULD be logged . If the v3 flag is set (in addition to v2), then the include/exclude flag MUST indicate "exclude". If not, then an error SHOULD be logged. The PE MUST generate an IGMP membership report (Join) for that (*,G) and each IGMP version in the version flag. == Using lowercase 'not' together with uppercase 'MUST', 'SHALL', 'SHOULD', or 'RECOMMENDED' is not an accepted usage according to RFC 2119. Please use uppercase 'NOT' together with RFC 2119 keywords (if that is what you mean). Found 'SHOULD not' in this paragraph: A PE which has received an IGMP Join, would have synced the IGMP Join by the procedure defined in section 6.1. If a PE with local join state goes down or the PE to CE link goes down, it would lead to a mass withdraw of multicast routes. Remote PEs (PEs where these routes were remote IGMP Joins) SHOULD not remove the state immediately; instead General Query SHOULD be generated to refresh the states. There are several ways to Some of the way to detect failure at a peer, e.g. using IGP next hop tracking or ES route withdraw. == Using lowercase 'not' together with uppercase 'MUST', 'SHALL', 'SHOULD', or 'RECOMMENDED' is not an accepted usage according to RFC 2119. Please use uppercase 'NOT' together with RFC 2119 keywords (if that is what you mean). Found 'MUST not' in this paragraph: This section describes the procedures used to construct the IGMP Leave Synch route. Support for this route type is optional. If a PE does not support this route, then it MUST not indicate that it supports 'IGMP proxy' in Multicast Flag extended community for the EVIs corresponding to its multi-homed Ethernet Segments. == Using lowercase 'not' together with uppercase 'MUST', 'SHALL', 'SHOULD', or 'RECOMMENDED' is not an accepted usage according to RFC 2119. Please use uppercase 'NOT' together with RFC 2119 keywords (if that is what you mean). Found 'MUST not' in this paragraph: If a router does not support this specification, it MUST not add Multicast Flags Extended Community in BGP route. A router receiving BGP update , if M and I both flag are zero (0), the router MUST treat this Update as malformed . Receiver of such update MUST ignore the extended community. -- The document date (January 22, 2021) is 1190 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Missing Reference: 'ES' is mentioned on line 606, but not defined == Missing Reference: 'BD' is mentioned on line 606, but not defined Summary: 1 error (**), 0 flaws (~~), 8 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 BESS WorkGroup A. Sajassi 3 Internet-Draft S. Thoria 4 Intended status: Standards Track Cisco Systems 5 Expires: July 26, 2021 K. Patel 6 Arrcus 7 J. Drake 8 W. Lin 9 Juniper Networks 10 January 22, 2021 12 IGMP and MLD Proxy for EVPN 13 draft-ietf-bess-evpn-igmp-mld-proxy-06 15 Abstract 17 Ethernet Virtual Private Network (EVPN) solution is becoming 18 pervasive in data center (DC) applications for Network Virtualization 19 Overlay (NVO) and DC interconnect (DCI) services, and in service 20 provider (SP) applications for next generation virtual private LAN 21 services. 23 This draft describes how to support efficiently endpoints running 24 IGMP for the above services over an EVPN network by incorporating 25 IGMP proxy procedures on EVPN PEs. 27 Status of This Memo 29 This Internet-Draft is submitted in full conformance with the 30 provisions of BCP 78 and BCP 79. 32 Internet-Drafts are working documents of the Internet Engineering 33 Task Force (IETF). Note that other groups may also distribute 34 working documents as Internet-Drafts. The list of current Internet- 35 Drafts is at https://datatracker.ietf.org/drafts/current/. 37 Internet-Drafts are draft documents valid for a maximum of six months 38 and may be updated, replaced, or obsoleted by other documents at any 39 time. It is inappropriate to use Internet-Drafts as reference 40 material or to cite them other than as "work in progress." 42 This Internet-Draft will expire on July 26, 2021. 44 Copyright Notice 46 Copyright (c) 2020 IETF Trust and the persons identified as the 47 document authors. All rights reserved. 49 This document is subject to BCP 78 and the IETF Trust's Legal 50 Provisions Relating to IETF Documents 51 (https://trustee.ietf.org/license-info) in effect on the date of 52 publication of this document. Please review these documents 53 carefully, as they describe your rights and restrictions with respect 54 to this document. Code Components extracted from this document must 55 include Simplified BSD License text as described in Section 4.e of 56 the Trust Legal Provisions and are provided without warranty as 57 described in the Simplified BSD License. 59 Table of Contents 61 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 62 2. Specification of Requirements . . . . . . . . . . . . . . . . 4 63 3. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 4 64 4. IGMP/MLD Proxy . . . . . . . . . . . . . . . . . . . . . . . 5 65 4.1. Proxy Reporting . . . . . . . . . . . . . . . . . . . . . 6 66 4.1.1. IGMP/MLD Membership Report Advertisement in BGP . . . 6 67 4.1.2. IGMP/MLD Leave Group Advertisement in BGP . . . . . . 8 68 4.2. Proxy Querier . . . . . . . . . . . . . . . . . . . . . . 8 69 5. Operation . . . . . . . . . . . . . . . . . . . . . . . . . . 9 70 5.1. PE with only attached hosts/VMs for a given subnet . . . 10 71 5.2. PE with a mix of attached hosts/VMs and multicast source 11 72 5.3. PE with a mix of attached hosts/VMs, a multicast source 73 and a router . . . . . . . . . . . . . . . . . . . . . . 11 74 6. All-Active Multi-Homing . . . . . . . . . . . . . . . . . . . 11 75 6.1. Local IGMP/MLD Join Synchronization . . . . . . . . . . . 11 76 6.2. Local IGMP/MLD Leave Group Synchronization . . . . . . . 12 77 6.2.1. Remote Leave Group Synchronization . . . . . . . . . 13 78 6.2.2. Common Leave Group Synchronization . . . . . . . . . 13 79 6.3. Mass Withdraw of Multicast join Sync route in case of 80 failure . . . . . . . . . . . . . . . . . . . . . . . . . 14 81 7. Single-Active Multi-Homing . . . . . . . . . . . . . . . . . 14 82 8. Selective Multicast Procedures for IR tunnels . . . . . . . . 14 83 9. BGP Encoding . . . . . . . . . . . . . . . . . . . . . . . . 15 84 9.1. Selective Multicast Ethernet Tag Route . . . . . . . . . 15 85 9.1.1. Constructing the Selective Multicast Ethernet Tag 86 route . . . . . . . . . . . . . . . . . . . . . . . . 17 87 9.1.2. Default Selective Multicast Route . . . . . . . . . . 18 88 9.2. Multicast Join Synch Route . . . . . . . . . . . . . . . 19 89 9.2.1. Constructing the Multicast Join Synch Route . . . . . 20 90 9.3. Multicast Leave Synch Route . . . . . . . . . . . . . . . 22 91 9.3.1. Constructing the Multicast Leave Synch Route . . . . 23 92 9.4. Multicast Flags Extended Community . . . . . . . . . . . 25 93 9.5. EVI-RT Extended Community . . . . . . . . . . . . . . . . 26 94 9.6. Rewriting of RT ECs and EVI-RT ECs by ASBRs . . . . . . . 28 95 9.7. BGP Error Handling . . . . . . . . . . . . . . . . . . . 29 96 10. IGMP/MLD Immediate Leave . . . . . . . . . . . . . . . . . . 29 97 11. IGMP Version 1 Membership Request . . . . . . . . . . . . . . 29 98 12. Security Considerations . . . . . . . . . . . . . . . . . . . 29 99 13. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 29 100 14. Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . 30 101 15. Contributors . . . . . . . . . . . . . . . . . . . . . . . . 30 102 16. References . . . . . . . . . . . . . . . . . . . . . . . . . 31 103 16.1. Normative References . . . . . . . . . . . . . . . . . . 31 104 16.2. Informative References . . . . . . . . . . . . . . . . . 32 105 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 32 107 1. Introduction 109 Ethernet Virtual Private Network (EVPN) solution [RFC7432] is 110 becoming pervasive in data center (DC) applications for Network 111 Virtualization Overlay (NVO) and DC interconnect (DCI) services, and 112 in service provider (SP) applications for next generation virtual 113 private LAN services. 115 In DC applications, a point of delivery (POD) can consist of a 116 collection of servers supported by several top of rack (TOR) and 117 Spine switches. This collection of servers and switches are self 118 contained and may have their own control protocol for intra-POD 119 communication and orchestration. However, EVPN is used as standard 120 way of inter-POD communication for both intra-DC and inter-DC. A 121 subnet can span across multiple PODs and DCs. EVPN provides robust 122 multi-tenant solution with extensive multi-homing capabilities to 123 stretch a subnet (VLAN) across multiple PODs and DCs. There can be 124 many hosts/VMs ( several hundreds) attached to a subnet that is 125 stretched across several PODs and DCs. 127 These hosts/VMs express their interests in multicast groups on a 128 given subnet/VLAN by sending IGMP membership reports (Joins) for 129 their interested multicast group(s). Furthermore, an IGMP router 130 periodically sends membership queries to find out if there are hosts 131 on that subnet that are still interested in receiving multicast 132 traffic for that group. The IGMP/MLD Proxy solution described in 133 this draft accomplishes has three objectives: 135 1. Reduce flooding of IGMP messages: just like the ARP/ND 136 suppression mechanism in EVPN to reduce the flooding of ARP 137 messages over EVPN, it is also desired to have a mechanism to 138 reduce the flooding of IGMP messages (both Queries and Reports) 139 in EVPN. 141 2. Distributed anycast multicast proxy: it is desirable for the EVPN 142 network to act as a distributed anycast multicast router with 143 respect to IGMP/MLD proxy function for all the hosts attached to 144 that subnet. 146 3. Selective Multicast: to forward multicast traffic over EVPN 147 network such that it only gets forwarded to the PEs that have 148 interest in the multicast group(s), multicast traffic will not be 149 forwarded to the PEs that have no receivers attached to them for 150 that multicast group. This draft shows how this objective may be 151 achieved when Ingress Replication is used to distribute the 152 multicast traffic among the PEs. Procedures for supporting 153 selective multicast using P2MP tunnels can be found in [bum- 154 procedure-updates] 156 The first two objectives are achieved by using IGMP/MLD proxy on the 157 PE and the third objective is achieved by setting up a multicast 158 tunnel (e.g., ingress replication) only among the PEs that have 159 interest in that multicast group(s) based on the trigger from IGMP/ 160 MLD proxy processes. The proposed solutions for each of these 161 objectives are discussed in the following sections. 163 2. Specification of Requirements 165 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 166 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 167 "OPTIONAL" in this document are to be interpreted as described in BCP 168 14 [RFC2119] [RFC8174] when, and only when, they appear in all 169 capitals, as shown here. 171 3. Terminology 173 o POD: Point of Delivery 175 o ToR: Top of Rack 177 o NV: Network Virtualization 179 o NVO: Network Virtualization Overlay 181 o EVPN: Ethernet Virtual Private Network 183 o IGMP: Internet Group Management Protocol 185 o MLD: Multicast Listener Discovery 187 o EVI: An EVPN instance spanning the Provider Edge (PE) devices 188 participating in that EVPN 190 o MAC-VRF: A Virtual Routing and Forwarding table for Media Access 191 Control (MAC) addresses on a PE 193 o IR: Ingress Replication 194 o Ethernet Segment (ES): When a customer site (device or network) is 195 connected to one or more PEs via a set of Ethernet links, then 196 that set of links is referred to as an 'Ethernet Segment'. 198 o Ethernet Segment Identifier (ESI): A unique non-zero identifier 199 that identifies an Ethernet Segment is called an 'Ethernet Segment 200 Identifier'. 202 o PE: Provider Edge. 204 o BD: Broadcast Domain. As per [RFC7432], an EVI consists of a 205 single or multiple BDs. In case of VLAN-bundle and VLAN-aware 206 bundle service model, an EVI contains multiple BDs. Also, in this 207 document, BD and subnet are equivalent terms. 209 o Ethernet Tag: An Ethernet tag identifies a particular broadcast 210 domain, e.g., a VLAN. An EVPN instance consists of one or more 211 broadcast domains. 213 o Single-Active Redundancy Mode: When only a single PE, among all 214 the PEs attached to an Ethernet segment, is allowed to forward 215 traffic to/from that Ethernet segment for a given VLAN, then the 216 Ethernet segment is defined to be operating in Single-Active 217 redundancy mode. 219 o All-Active Redundancy Mode: When all PEs attached to an Ethernet 220 segment are allowed to forward known unicast traffic to/from that 221 Ethernet segment for a given VLAN, then the Ethernet segment is 222 defined to be operating in All-Active redundancy mode. 224 This document also assumes familiarity with the terminology of 225 [RFC7432]. Though most of the place this document uses term IGMP 226 membership request (Joins), the text applies equally for MLD 227 membership request too. Similarly, text for IGMPv2 applies to MLDv1 228 and text for IGMPv3 applies to MLDv2. IGMP / MLD version encoding in 229 BGP update is stated in Section 9 231 4. IGMP/MLD Proxy 233 The IGMP Proxy mechanism is used to reduce the flooding of IGMP 234 messages over an EVPN network similar to ARP proxy used in reducing 235 the flooding of ARP messages over EVPN. It also provides a 236 triggering mechanism for the PEs to setup their underlay multicast 237 tunnels. The IGMP Proxy mechanism consists of two components: 239 1. Proxy for IGMP Reports. 241 2. Proxy for IGMP Queries. 243 4.1. Proxy Reporting 245 When IGMP protocol is used between hosts/VMs and their first hop EVPN 246 router (EVPN PE), Proxy-reporting is used by the EVPN PE to summarize 247 (when possible) reports received from downstream hosts and propagate 248 them in BGP to other PEs that are interested in the information. 249 This is done by terminating the IGMP Reports in the first hop PE, and 250 translating and exchanging the relevant information among EVPN BGP 251 speakers. The information is again translated back to IGMP message 252 at the recipient EVPN speaker. Thus it helps create an IGMP overlay 253 subnet using BGP. In order to facilitate such an overlay, this 254 document also defines a new EVPN route type NLRI, the EVPN Selective 255 Multicast Ethernet Tag route, along with its procedures to help 256 exchange and register IGMP multicast groups Section 9. 258 4.1.1. IGMP/MLD Membership Report Advertisement in BGP 260 When a PE wants to advertise an IGMP membership report (Join) using 261 the BGP EVPN route, it follows the following rules (BGP encoding 262 stated in Section 9): 264 1. When the first hop PE receives several IGMP membership reports 265 (Joins), belonging to the same IGMP version, from different 266 attached hosts/VMs for the same (*,G) or (S,G), it only SHOULD 267 send a single BGP message corresponding to the very first IGMP 268 Join (BGP update as soon as possible) for that (*,G) or (S,G). 269 This is because BGP is a stateful protocol and no further 270 transmission of the same report is needed. If the IGMP Join is 271 for (*,G), then multicast group address MUST be sent along with 272 the corresponding version flag (v2 or v3) set. In case of 273 IGMPv3, the exclude flag MUST also needs to be set to indicate 274 that no source IP address to be excluded (include all 275 sources"*"). If the IGMP Join is for (S,G), then besides setting 276 multicast group address along with the version flag v3, the 277 source IP address and the include/exclude flag MUST be set. It 278 should be noted that when advertising the EVPN route for (S,G), 279 the only valid version flag is v3 (v2 flags MUST be set to zero). 281 2. When the first hop PE receives an IGMPv3 Join for (S,G) on a 282 given BD, it SHOULD advertise the corresponding EVPN Selective 283 Multicast Ethernet Tag (SMET) route regardless of whether the 284 source (S) is attached to itself or not in order to facilitate 285 the source move in the future. 287 3. When the first hop PE receives an IGMP version-X Join first for 288 (*,G) and then later it receives an IGMP version-Y Join for the 289 same (*,G), then it MUST re-advertise the same EVPN SMET route 290 with flag for version-Y set in addition to any previously-set 291 version flag(s). In other words, the first hop PE MUST not 292 withdraw the EVPN route before sending the new route because the 293 flag field is not part of BGP route key processing. 295 4. When the first hop PE receives an IGMP version-X Join first for 296 (*,G) and then later it receives an IGMPv3 Join for the same 297 multicast group address but for a specific source address S, then 298 the PE MUST advertise a new EVPN SMET route with v3 flag set (and 299 v2 reset). The include/exclude flag also need to be set 300 accordingly. Since source IP address is used as part of BGP 301 route key processing it is considered as a new BGP route 302 advertisement. 304 5. When a PE receives an EVPN SMET route with more than one version 305 flag set, it will generate the corresponding IGMP report for 306 (*,G) for each version specified in the flags field. With 307 multiple version flags set, there MUST not be source IP address 308 in the receive EVPN route. If there is, then an error SHOULD be 309 logged . If the v3 flag is set (in addition to v2), then the 310 include/exclude flag MUST indicate "exclude". If not, then an 311 error SHOULD be logged. The PE MUST generate an IGMP membership 312 report (Join) for that (*,G) and each IGMP version in the version 313 flag. 315 6. When a PE receives a list of EVPN SMET NLRIs in its BGP update 316 message, each with a different source IP address and the same 317 multicast group address, and the version flag is set to v3, then 318 the PE generates an IGMPv3 membership report with a record 319 corresponding to the list of source IP addresses and the group 320 address along with the proper indication of inclusion/exclusion. 322 7. Upon receiving EVPN SMET route(s) and before generating the 323 corresponding IGMP Join(s), the PE checks to see whether it has 324 any CE multicast router for that BD on any of its ES's . The PE 325 provides such a check by listening for PIM Hello messages on that 326 AC (i.e, ES,BD). If the PE does have the router's ACs, then the 327 generated IGMP Join(s) are sent to those ACs. If it doesn't have 328 any of the router's AC, then no IGMP Join(s) needs to be 329 generated. This is because sending IGMP Joins to other hosts can 330 result in unintentionally preventing a host from joining a 331 specific multicast group using IGMPv2 - i.e., if the PE does not 332 receive a join from the host it will not forward multicast data 333 to it. Per [RFC4541] , when an IGMPv2 host receives a membership 334 report for a group address that it intends to join, the host will 335 suppress its own membership report for the same group, and if the 336 PE does not receive an IGMP Join from host it will not forward 337 multicast data to it. In other words, an IGMPv2 Join MUST NOT be 338 sent on an AC that does not lead to a CE multicast router. This 339 message suppression is a requirement for IGMPv2 hosts. This is 340 not a problem for hosts running IGMPv3 because there is no 341 suppression of IGMP Membership reports. 343 4.1.2. IGMP/MLD Leave Group Advertisement in BGP 345 When a PE wants to withdraw an EVPN SMET route corresponding to an 346 IGMPv2 Leave Group (Leave) or IGMPv3 "Leave" equivalent message, it 347 follows the following rules: 349 1. When a PE receives an IGMPv2 Leave Group or its "Leave" 350 equivalent message for IGMPv3 from its attached host, it checks 351 to see if this host is the last host that is interested in this 352 multicast group by sending a query for the multicast group. If 353 the host was indeed the last one (i.e. no responses are received 354 for the query), then the PE MUST re-advertises EVPN SMET 355 Multicast route with the corresponding version flag reset. If 356 this is the last version flag to be reset, then instead of re- 357 advertising the EVPN route with all version flags reset, the PE 358 MUST withdraws the EVPN route for that (*,G). 360 2. When a PE receives an EVPN SMET route for a given (*,G), it 361 compares the received version flags from the route with its per- 362 PE stored version flags. If the PE finds that a version flag 363 associated with the (*,G) for the remote PE is reset, then the PE 364 MUST generate IGMP Leave for that (*,G) toward its local 365 interface (if any) attached to the multicast router for that 366 multicast group. It should be noted that the received EVPN route 367 SHOULD at least have one version flag set. If all version flags 368 are reset, it is an error because the PE should have received an 369 EVPN route withdraw for the last version flag. Error MUST be 370 considered as BGP error and the PE MUST apply the "treat-as- 371 withdraw" procedure of [RFC7606]. 373 3. When a PE receives an EVPN SMET route withdraw, it removes the 374 remote PE from its OIF list for that multicast group and if there 375 are no more OIF entries for that multicast group (either locally 376 or remotely), then the PE MUST stop responding to queries from 377 the locally attached router (if any). If there is a source for 378 that multicast group, the PE stops sending multicast traffic for 379 that source. 381 4.2. Proxy Querier 383 As mentioned in the previous sections, each PE MUST have proxy 384 querier functionality for the following reasons: 386 1. To enable the collection of EVPN PEs providing L2VPN service to 387 act as distributed multicast router with Anycast IP address for 388 all attached hosts/VMs in that subnet. 390 2. To enable suppression of IGMP membership reports and queries over 391 MPLS/IP core. 393 5. Operation 395 Consider the EVPN network of Figure-1, where there is an EVPN 396 instance configured across the PEs shown in this figure (namely PE1, 397 PE2, and PE3). Let's consider that this EVPN instance consists of a 398 single bridge domain (single subnet) with all the hosts, sources, and 399 the multicast router connected to this subnet. PE1 only has hosts 400 connected to it. PE2 has a mix of hosts and a multicast source. PE3 401 has a mix of hosts, a multicast source, and a multicast router. 402 Furthermore, let's consider that for (S1,G1), R1 is used as the 403 multicast router. The following subsections describe the IGMP proxy 404 operation in different PEs with regard to whether the locally 405 attached devices for that subnet are: 407 o only hosts/VMs 409 o mix of hosts/VMs and multicast source 411 o mix of hosts/VMs, multicast source, and multicast router 412 +--------------+ 413 | | 414 | | 415 +----+ | | +----+ 416 H1:(*,G1)v2 ---| | | | | |---- H6(*,G1)v2 417 H2:(*,G1)v2 ---| PE1| | IP/MPLS | | PE2|---- H7(S2,G2)v3 418 H3:(*,G1)v3 ---| | | Network | | |---- S2 419 H4:(S2,G2)v3 --| | | | | | 420 +----+ | | +----+ 421 | | 422 +----+ | | 423 H5:(S1,G1)v3 --| | | | 424 S1 ---| PE3| | | 425 R1 ---| | | | 426 +----+ | | 427 | | 428 +--------------+ 430 Figure 1: EVPN network 432 5.1. PE with only attached hosts/VMs for a given subnet 434 When PE1 receives an IGMPv2 Join Report from H1, it does not forward 435 this join to any of its other ports (for this subnet) because all 436 these local ports are associated with the hosts/VMs. PE1 sends an 437 EVPN Multicast Group route corresponding to this join for (*,G1) and 438 setting v2 flag. This EVPN route is received by PE2 and PE3 that are 439 the members of the same BD (i.e., same EVI in case of VLAN-based 440 service or EVI,VLAN in case of VLAN-aware bundle service). PE3 441 reconstructs the IGMPv2 Join Report from this EVPN BGP route and only 442 sends it to the port(s) with multicast routers attached to it (for 443 that subnet). In this example, PE3 sends the reconstructed IGMPv2 444 Join Report for (*,G1) only to R1. Furthermore, even though PE2 445 receives the EVPN BGP route, it does not send it to any of its ports 446 for that subnet; viz, ports associated with H6 and H7. 448 When PE1 receives the second IGMPv2 Join from H2 for the same 449 multicast group (*,G1), it only adds that port to its OIF list but it 450 doesn't send any EVPN BGP route because there is no change in 451 information. However, when it receives the IGMPv3 Join from H3 for 452 the same (*,G1). Besides adding the corresponding port to its OIF 453 list, it re-advertises the previously sent EVPN SMET route with the 454 v3 and exclude flag set. 456 Finally when PE1 receives the IMGMPv3 Join from H4 for (S2,G2), it 457 advertises a new EVPN SMET route corresponding to it. 459 5.2. PE with a mix of attached hosts/VMs and multicast source 461 The main difference in this case is that when PE2 receives the IGMPv3 462 Join from H7 for (S2,G2), it does advertise it in BGP to support 463 source move even though PE2 knows that S2 is attached to its local 464 AC. PE2 adds the port associated with H7 to its OIF list for 465 (S2,G2). The processing for IGMPv2 received from H6 is the same as 466 the IGMPv2 Join described in previous section. 468 5.3. PE with a mix of attached hosts/VMs, a multicast source and a 469 router 471 The main difference in this case relative to the previous two 472 sections is that IGMP v2/v3 Join messages received locally needs to 473 be sent to the port associated with router R1. Furthermore, the 474 Joins received via BGP (SMET) need to be passed to the R1 port but 475 filtered for all other ports. 477 6. All-Active Multi-Homing 479 Because the LAG flow hashing algorithm used by the CE is unknown at 480 the PE, in an All-Active redundancy mode it must be assumed that the 481 CE can send a given IGMP message to any one of the multi-homed PEs, 482 either DF or non-DF; i.e., different IGMP Join messages can arrive at 483 different PEs in the redundancy group and furthermore their 484 corresponding Leave messages can arrive at PEs that are different 485 from the ones that received the Join messages. Therefore, all PEs 486 attached to a given ES must coordinate IGMP Join and Leave Group 487 (x,G) state, where x may be either '*' or a particular source S, for 488 each BD on that ES. This allows the DF for that [ES,BD] to correctly 489 advertise or withdraw a Selective Multicast Ethernet Tag (SMET) route 490 for that (x,G) group in that BD when needed. All-Active multihoming 491 PEs for a given ES MUST support IGMP synchronization procedures 492 described in this section if they need to perform IGMP proxy for 493 hosts connected to that ES. 495 6.1. Local IGMP/MLD Join Synchronization 497 When a PE, either DF or non-DF, receives on a given multihomed ES 498 operating in All-Active redundancy mode, an IGMP Membership Report 499 for (x,G), it determines the BD to which the IGMP Membership Report 500 belongs. If the PE doesn't already have local IGMP Join (x,G) state 501 for that BD on that ES, it MUST instantiate local IGMP Join (x,G) 502 state and MUST advertise a BGP IGMP Join Synch route for that 503 [ES,BD]. Local IGMP Join (x, G) state refers to IGMP Join (x,G) 504 state that is created as a result of processing an IGMP Membership 505 Report for (x,G). 507 The IGMP Join Synch route MUST carry the ES-Import RT for the ES on 508 which the IGMP Membership Report was received. Thus it MUST only be 509 imported by the PEs attached to that ES and not any other PEs. 511 When a PE, either DF or non-DF, receives an IGMP Join Synch route it 512 installs that route and if it doesn't already have IGMP Join (x,G) 513 state for that [ES,BD], it MUST instantiate that IGMP Join (x,G) 514 state - i.e., IGMP Join (x,G) state is the union of the local IGMP 515 Join (x,G) state and the installed IGMP Join Synch route. If the DF 516 did not already advertise (originate) a SMET route for that (x,G) 517 group in that BD, it MUST do so now. 519 When a PE, either DF or non-DF, deletes its local IGMP Join (x, G) 520 state for that [ES,BD], it MUST withdraw its BGP IGMP Join Synch 521 route for that [ES,BD]. 523 When a PE, either DF or non-DF, receives the withdrawal of an IGMP 524 Join Synch route from another PE it MUST remove that route. When a 525 PE has no local IGMP Join (x,G) state and it has no installed IGMP 526 Join Synch routes, it MUST remove IGMP Join (x,G) state for that 527 [ES,BD]. If the DF no longer has IGMP Join (x,G) state for that BD 528 on any ES for which it is DF, it MUST withdraw its SMET route for 529 that (x,G) group in that BD. 531 In other words, a PE advertises an SMET route for that (x,G) group in 532 that BD when it has IGMP Join (x,G) state in that BD on at least one 533 ES for which it is DF and it withdraws that SMET route when it does 534 not have IGMP Join (x,G) state in that BD on any ES for which it is 535 DF. 537 6.2. Local IGMP/MLD Leave Group Synchronization 539 When a PE, either DF or non-DF, receives, on a given multihomed ES 540 operating in All-Active redundancy mode, an IGMP Leave Group message 541 for (x,G) from the attached CE, it determines the BD to which the 542 IGMPv2 Leave Group belongs. Regardless of whether it has IGMP Join 543 (x,G) state for that [ES,BD], it initiates the (x,G) leave group 544 synchronization procedure, which consists of the following steps: 546 1. It computes the Maximum Response Time, which is the duration of 547 (x,G) leave group synchronization procedure. This is the product 548 of two locally configured values, Last Member Query Count and 549 Last Member Query Interval (described in Section 3 of [RFC2236]), 550 plus a delta corresponding to the time it takes for a BGP 551 advertisement to propagate between the PEs attached to the 552 multihomed ES (delta is a consistently configured value on all 553 PEs attached to the multihomed ES). 555 2. It starts the Maximum Response Time timer. Note that the receipt 556 of subsequent IGMP Leave Group messages or BGP Leave Synch routes 557 for (x,G) do not change the value of a currently running Maximum 558 Response Time timer and are ignored by the PE. 560 3. It initiates the Last Member Query procedure described in 561 Section 3 of [RFC2236]; viz, it sends a number of Group-Specific 562 Query (x,G) messages (Last Member Query Count) at a fixed 563 interval (Last Member Query Interval) to the attached CE. 565 4. It advertises an IGMP Leave Synch route for that that [ES,BD]. 566 This route notifies the other multihomed PEs attached to the 567 given multihomed ES that it has initiated an (x,G) leave group 568 synchronization procedure; i.e., it carries the ES-Import RT for 569 the ES on which the IGMP Leave Group was received. It also 570 contains the Maximum Response Time. 572 5. When the Maximum Response Timer expires, the PE that has 573 advertised the IGMP Leave Synch route withdraws it. 575 6.2.1. Remote Leave Group Synchronization 577 When a PE, either DF or non-DF, receives an IGMP Leave Synch route it 578 installs that route and it starts a timer for (x,G) on the specified 579 [ES,BD] whose value is set to the Maximum Response Time in the 580 received IGMP Leave Synch route. Note that the receipt of subsequent 581 IGMPv2 Leave Group messages or BGP Leave Synch routes for (x,G) do 582 not change the value of a currently running Maximum Response Time 583 timer and are ignored by the PE. 585 6.2.2. Common Leave Group Synchronization 587 If a PE attached to the multihomed ES receives an IGMP Membership 588 Report for (x,G) before the Maximum Response Time timer expires, it 589 advertises a BGP IGMP Join Synch route for that [ES,BD]. If it 590 doesn't already have local IGMP Join (x, G) state for that [ES, BD], 591 it instantiates local IGMP Join (x,G) state. If the DF is not 592 currently advertising (originating) a SMET route for that (x,G) group 593 in that BD, it does so now. 595 If a PE attached to the multihomed ES receives an IGMP Join Synch 596 route for (x,G) before the Maximum Response Time timer expires, it 597 installs that route and if it doesn't already have IGMP Join (x,G) 598 state for that BD on that ES, it instantiates that IGMP Join (x,G) 599 state. If the DF has not already advertised (originated) a SMET 600 route for that (x,G) group in that BD, it does so now. 602 When the Maximum Response Timer expires a PE that has advertised an 603 IGMP Leave Synch route, withdraws it. Any PE attached to the 604 multihomed ES, that started the Maximum Response Time and has no 605 local IGMP Join (x,G) state and no installed IGMP Join Synch routes, 606 it removes IGMP Join (x,G) state for that [ES,BD]. If the DF no 607 longer has IGMP Join (x,G) state for that BD on any ES for which it 608 is DF, it withdraws its SMET route for that (x,G) group in that BD. 610 6.3. Mass Withdraw of Multicast join Sync route in case of failure 612 A PE which has received an IGMP Join, would have synced the IGMP Join 613 by the procedure defined in section 6.1. If a PE with local join 614 state goes down or the PE to CE link goes down, it would lead to a 615 mass withdraw of multicast routes. Remote PEs (PEs where these 616 routes were remote IGMP Joins) SHOULD not remove the state 617 immediately; instead General Query SHOULD be generated to refresh the 618 states. There are several ways to Some of the way to detect failure 619 at a peer, e.g. using IGP next hop tracking or ES route withdraw. 621 7. Single-Active Multi-Homing 623 Note that to facilitate state synchronization after failover, the PEs 624 attached to a mutihomed ES operating in Single-Active redundancy mode 625 SHOULD also coordinate IGMP Join (x,G) state. In this case all IGMP 626 Join messages are received by the DF and distributed to the non-DF 627 PEs using the procedures described above. 629 8. Selective Multicast Procedures for IR tunnels 631 If an ingress PE uses ingress replication, then for a given (x,G) 632 group in a given BD: 634 1. It sends (x,G) traffic to the set of PEs not supporting IGMP 635 Proxy. This set consists of any PE that has advertised an 636 Inclusive Multicast Tag route for the BD without the "IGMP Proxy 637 Support" flag. 639 2. It sends (x,G) traffic to the set of PEs supporting IGMP Proxy 640 and having listeners for that (x,G) group in that BD. This set 641 consists of any PE that has advertised an Inclusive Multicast Tag 642 route for the BD with the "IGMP Proxy Support" flag and that has 643 advertised a SMET route for that (x,G) group in that BD. 645 If an ingress PE's Selective P-Tunnel for a given BD uses P2MP and 646 all of the PEs in the BD support that tunnel type and IGMP proxy, 647 then for a given (x,G) group in a given BD it sends (x,G) traffic 648 using the Selective P-Tunnel for that (x,G) group in that BD. This 649 tunnel includes those PEs that have advertised a SMET route for that 650 (x,G) group on that BD (for Selective P-tunnel) but it may include 651 other PEs as well (for Aggregate Selective P-tunnel). 653 9. BGP Encoding 655 This document defines three new BGP EVPN routes to carry IGMP 656 membership reports. The route type is known as: 658 + 6 - Selective Multicast Ethernet Tag Route 660 + 7 - Multicast Join Synch Route 662 + 8 - Multicast Leave Synch Route 664 The detailed encoding and procedures for this route type are 665 described in subsequent sections. 667 9.1. Selective Multicast Ethernet Tag Route 669 A Selective Multicast Ethernet Tag route type specific EVPN NLRI 670 consists of the following: 672 +---------------------------------------+ 673 | RD (8 octets) | 674 +---------------------------------------+ 675 | Ethernet Tag ID (4 octets) | 676 +---------------------------------------+ 677 | Multicast Source Length (1 octet) | 678 +---------------------------------------+ 679 | Multicast Source Address (variable) | 680 +---------------------------------------+ 681 | Multicast Group Length (1 octet) | 682 +---------------------------------------+ 683 | Multicast Group Address (Variable) | 684 +---------------------------------------+ 685 | Originator Router Length (1 octet) | 686 +---------------------------------------+ 687 | Originator Router Address (variable) | 688 +---------------------------------------+ 689 | Flags (1 octet) | 690 +---------------------------------------+ 692 For the purpose of BGP route key processing, all the fields are 693 considered to be part of the prefix in the NLRI except for the one- 694 octet flag field. The Flags fields are defined as follows: 696 0 1 2 3 4 5 6 7 697 +--+--+--+--+--+--+--+--+ 698 | reserved |IE|v3|v2|v1| 699 +--+--+--+--+--+--+--+--+ 701 o The least significant bit, bit 7 indicates support for IGMP 702 version 1. Since IGMP V1 is being deprecated , sender MUST set it 703 as 0 for IGMP and receiver MUST ignore it. 705 o The second least significant bit, bit 6 indicates support for IGMP 706 version 2. 708 o The third least significant bit, bit 5 indicates support for IGMP 709 version 3. 711 o The fourth least significant bit, bit 4 indicates whether the 712 (S,G) information carried within the route-type is of an Include 713 Group type (bit value 0) or an Exclude Group type (bit value 1). 714 The Exclude Group type bit MUST be ignored if bit 5 is not set. 716 o This EVPN route type is used to carry tenant IGMP multicast group 717 information. The flag field assists in distributing IGMP 718 membership interest of a given host/VM for a given multicast 719 route. The version bits help associate IGMP version of receivers 720 participating within the EVPN domain. 722 o The include/exclude bit helps in creating filters for a given 723 multicast route. 725 o If route is used for IPv6 (MLD) then bit 7 indicates support for 726 MLD version 1. The second least significant bit, bit 6 indicates 727 support for MLD version 2. Since there is no MLD version 3, in 728 case of IPv6 route third least significant bit MUST be 0. In case 729 of IPv6 routes, the fourth least significant bit MUST be ignored 730 if bit 6 is not set. 732 o Reserve bit SHOULD be set to 0 by sender. And receiver SHOULD 733 ignore the reserve bit. 735 9.1.1. Constructing the Selective Multicast Ethernet Tag route 737 This section describes the procedures used to construct the Selective 738 Multicast Ethernet Tag (SMET) route. 740 The Route Distinguisher (RD) SHOULD be a Type 1 RD [RFC4364] . The 741 value field comprises an IP address of the PE (typically, the 742 loopback address) followed by a number unique to the PE. 744 The Ethernet Tag ID MUST be set as follows: 746 o EVI is VLAN-Based or VLAN Bundle service - set to 0 748 o EVI is VLAN-Aware Bundle service without translation - set to the 749 customer VID for that BD 751 o EVI is VLAN-Aware Bundle service with translation - set to the 752 normalized Ethernet Tag ID - e.g., normalized VID 754 The Multicast Source Length MUST be set to length of the multicast 755 Source address in bits. If the Multicast Source Address field 756 contains an IPv4 address, then the value of the Multicast Source 757 Length field is 32. If the Multicast Source Address field contains 758 an IPv6 address, then the value of the Multicast Source Length field 759 is 128. In case of a (*, G) Join, the Multicast Source Length is set 760 to 0. 762 The Multicast Source Address is the source IP address from the IGMP 763 membership report. In case of a (*, G), this field is not used. 765 The Multicast Group Length MUST be set to length of multicast group 766 address in bits. If the Multicast Group Address field contains an 767 IPv4 address, then the value of the Multicast Group Length field is 768 32. If the Multicast Group Address field contains an IPv6 address, 769 then the value of the Multicast Group Length field is 128. 771 The Multicast Group Address is the Group address from the IGMP or MLD 772 membership report. 774 The Originator Router Length is the length of the Originator Router 775 Address in bits. 777 The Originator Router Address is the IP address of router originating 778 this route. The SMET Originator Router IP address MUST match that of 779 the IMET (or SPMSI AD) route originated for the same EVI by the same 780 downstream PE. 782 The Flags field indicates the version of IGMP protocol from which the 783 membership report was received. It also indicates whether the 784 multicast group had the INCLUDE or EXCLUDE bit set. 786 Reserve bit MUST be set to 0. They can be defined in future by other 787 document. 789 IGMP is used to receive group membership information from hosts/VMs 790 by TORs. Upon receiving the hosts/VMs expression of interest of a 791 particular group membership, this information is then forwarded using 792 SMET route. The NLRI also keeps track of receiver's IGMP protocol 793 version and any source filtering for a given group membership. All 794 EVPN SMET routes are announced with per- EVI Route Target extended 795 communities. 797 9.1.2. Default Selective Multicast Route 799 If there is multicast router connected behind the EVPN domain, the PE 800 MAY originate a default SMET (*,*) to get all multicast traffic in 801 domain. 803 +--------------+ 804 | | 805 | | 806 | | +----+ 807 | | | |---- H1(*,G1)v2 808 | IP/MPLS | | PE1|---- H2(S2,G2)v3 809 | Network | | |---- S2 810 | | | | 811 | | +----+ 812 | | 813 +----+ | | 814 +----+ | | | | 815 | | S1 ---| PE2| | | 816 |PIM |----R1 ---| | | | 817 |ASM | +----+ | | 818 | | | | 819 +----+ +--------------+ 821 Figure 2: Multicast Router behind EVPN domain 823 Consider the EVPN network of Figure-2, where there is an EVPN 824 instance configured across the PEs. Lets consider PE2 is connected 825 to multicast router R1 and there is a network running PIM ASM behind 826 R1. If there are receivers behind the PIM ASM network, the PIM Join 827 would be forwarded to the PIM RP (Rendezvous Point). If receivers 828 behind PIM ASM network are interested in a multicast flow originated 829 by multicast source S2 (behind PE1), it is necessary for PE2 to 830 receive multicast traffic. In this case PE2 MUST originate a (*,*) 831 SMET route to receive all of the multicast traffic in the EVPN 832 domain. 834 9.2. Multicast Join Synch Route 836 This EVPN route type is used to coordinate IGMP Join (x,G) state for 837 a given BD between the PEs attached to a given ES operating in All- 838 Active (or Single-Active) redundancy mode and it consists of 839 following: 841 +--------------------------------------------------+ 842 | RD (8 octets) | 843 +--------------------------------------------------+ 844 | Ethernet Segment Identifier (10 octets) | 845 +--------------------------------------------------+ 846 | Ethernet Tag ID (4 octets) | 847 +--------------------------------------------------+ 848 | Multicast Source Length (1 octet) | 849 +--------------------------------------------------+ 850 | Multicast Source Address (variable) | 851 +--------------------------------------------------+ 852 | Multicast Group Length (1 octet) | 853 +--------------------------------------------------+ 854 | Multicast Group Address (Variable) | 855 +--------------------------------------------------+ 856 | Originator Router Length (1 octet) | 857 +--------------------------------------------------+ 858 | Originator Router Address (variable) | 859 +--------------------------------------------------+ 860 | Flags (1 octet) | 861 +--------------------------------------------------+ 863 For the purpose of BGP route key processing, all the fields are 864 considered to be part of the prefix in the NLRI except for the one- 865 octet Flags field, whose fields are defined as follows: 867 0 1 2 3 4 5 6 7 868 +--+--+--+--+--+--+--+--+ 869 | reserved |IE|v3|v2|v1| 870 +--+--+--+--+--+--+--+--+ 872 o The least significant bit, bit 7 indicates support for IGMP 873 version 1. 875 o The second least significant bit, bit 6 indicates support for IGMP 876 version 2. 878 o The third least significant bit, bit 5 indicates support for IGMP 879 version 3. 881 o The fourth least significant bit, bit 4 indicates whether the (S, 882 G) information carried within the route-type is of Include Group 883 type (bit value 0) or an Exclude Group type (bit value 1). The 884 Exclude Group type bit MUST be ignored if bit 5 is not set. 886 o Reserve bit MUST be set to 0. They can be defined in future by 887 other document. 889 The Flags field assists in distributing IGMP membership interest of a 890 given host/VM for a given multicast route. The version bits help 891 associate IGMP version of receivers participating within the EVPN 892 domain. The include/exclude bit helps in creating filters for a 893 given multicast route. 895 If route is being prepared for IPv6 (MLD) then bit 7 indicates 896 support for MLD version 1. The second least significant bit, bit 6 897 indicates support for MLD version 2. Since there is no MLD version 898 3, in case of IPv6 route third least significant bit MUST be 0. In 899 case of IPv6 route, the fourth least significant bit MUST be ignored 900 if bit 6 is not set. 902 9.2.1. Constructing the Multicast Join Synch Route 904 This section describes the procedures used to construct the IGMP Join 905 Synch route. Support for this route type is optional. If a PE does 906 not support this route, then it MUST NOT indicate that it supports 907 'IGMP proxy' in the Multicast Flag extended community for the EVIs 908 corresponding to its multi-homed Ethernet Segments (ESs). 910 An IGMP Join Synch route MUST carry exactly one ES-Import Route 911 Target extended community, the one that corresponds to the ES on 912 which the IGMP Join was received. It MUST also carry exactly one 913 EVI-RT EC, the one that corresponds to the EVI on which the IGMP Join 914 was received. See Section 9.5 for details on how to encode and 915 construct the EVI-RT EC. 917 The Route Distinguisher (RD) SHOULD be a Type 1 RD [RFC4364] . The 918 value field comprises an IP address of the PE (typically, the 919 loopback address) followed by a number unique to the PE. 921 The Ethernet Segment Identifier (ESI) MUST be set to the 10-octet 922 value defined for the ES. 924 The Ethernet Tag ID MUST be set as follows: 926 o EVI is VLAN-Based or VLAN Bundle service - set to 0 928 o EVI is VLAN-Aware Bundle service without translation - set to the 929 customer VID for the BD 931 o EVI is VLAN-Aware Bundle service with translation - set to the 932 normalized Ethernet Tag ID - e.g., normalized VID 934 The Multicast Source length MUST be set to length of Multicast Source 935 address in bits. If the Multicast Source field contains an IPv4 936 address, then the value of the Multicast Source Length field is 32. 937 If the Multicast Source field contains an IPv6 address, then the 938 value of the Multicast Source Length field is 128. In case of a 939 (*,G) Join, the Multicast Source Length is set to 0. 941 The Multicast Source is the Source IP address of the IGMP membership 942 report. In case of a (*, G) Join, this field does not exist. 944 The Multicast Group length MUST be set to length of multicast group 945 address in bits. If the Multicast Group field contains an IPv4 946 address, then the value of the Multicast Group Length field is 32. 947 If the Multicast Group field contains an IPv6 address, then the value 948 of the Multicast Group Length field is 128. 950 The Multicast Group is the Group address of the IGMP membership 951 report. 953 The Originator Router Length is the length of the Originator Router 954 address in bits. 956 The Originator Router Address is the IP address of Router Originating 957 the prefix. 959 The Flags field indicates the version of IGMP protocol from which the 960 membership report was received. It also indicates whether the 961 multicast group had INCLUDE or EXCLUDE bit set. 963 Reserve bit MUST be set to 0. They can be defined in future by other 964 document. 966 9.3. Multicast Leave Synch Route 968 This EVPN route type is used to coordinate IGMP Leave Group (x,G) 969 state for a given BD between the PEs attached to a given ES operating 970 in All-Active (or Single-Active) redundancy mode and it consists of 971 following: 973 +--------------------------------------------------+ 974 | RD (8 octets) | 975 +--------------------------------------------------+ 976 | Ethernet Segment Identifier (10 octets) | 977 +--------------------------------------------------+ 978 | Ethernet Tag ID (4 octets) | 979 +--------------------------------------------------+ 980 | Multicast Source Length (1 octet) | 981 +--------------------------------------------------+ 982 | Multicast Source Address (variable) | 983 +--------------------------------------------------+ 984 | Multicast Group Length (1 octet) | 985 +--------------------------------------------------+ 986 | Multicast Group Address (Variable) | 987 +--------------------------------------------------+ 988 | Originator Router Length (1 octet) | 989 +--------------------------------------------------+ 990 | Originator Router Address (variable) | 991 +--------------------------------------------------+ 992 | Reserved (4 octet) | 993 +--------------------------------------------------+ 994 | Maximum Response Time (1 octet) | 995 +--------------------------------------------------+ 996 | Flags (1 octet) | 997 +--------------------------------------------------+ 999 For the purpose of BGP route key processing, all the fields are 1000 considered to be part of the prefix in the NLRI except for the 1001 Reserved, Maximum Response Time and the one-octet Flags field, whose 1002 fields are defined as follows: 1004 0 1 2 3 4 5 6 7 1005 +--+--+--+--+--+--+--+--+ 1006 | reserved |IE|v3|v2|v1| 1007 +--+--+--+--+--+--+--+--+ 1009 o The least significant bit, bit 7 indicates support for IGMP 1010 version 1. 1012 o The second least significant bit, bit 6 indicates support for IGMP 1013 version 2. 1015 o The third least significant bit, bit 5 indicates support for IGMP 1016 version 3. 1018 o The fourth least significant bit, bit 4 indicates whether the (S, 1019 G) information carried within the route-type is of Include Group 1020 type (bit value 0) or an Exclude Group type (bit value 1). The 1021 Exclude Group type bit MUST be ignored if bit 5 is not set. 1023 o Reserve bit MUST be set to 0. They can be defined in future by 1024 other document. 1026 The Flags field assists in distributing IGMP membership interest of a 1027 given host/VM for a given multicast route. The version bits help 1028 associate IGMP version of receivers participating within the EVPN 1029 domain. The include/exclude bit helps in creating filters for a 1030 given multicast route. 1032 If route is being prepared for IPv6 (MLD) then bit 7 indicates 1033 support for MLD version 1. The second least significant bit, bit 6 1034 indicates support for MLD version 2. Since there is no MLD version 1035 3, in case of IPv6 route third least significant bit MUST be 0. In 1036 case of IPv6 route, the fourth least significant bit MUST be ignored 1037 if bit 6 is not set. 1039 Reserve bit in flag MUST be set to 0. They can be defined in future 1040 by other document. 1042 9.3.1. Constructing the Multicast Leave Synch Route 1044 This section describes the procedures used to construct the IGMP 1045 Leave Synch route. Support for this route type is optional. If a PE 1046 does not support this route, then it MUST not indicate that it 1047 supports 'IGMP proxy' in Multicast Flag extended community for the 1048 EVIs corresponding to its multi-homed Ethernet Segments. 1050 An IGMP Leave Synch route MUST carry exactly one ES-Import Route 1051 Target extended community, the one that corresponds to the ES on 1052 which the IGMP Leave was received. It MUST also carry exactly one 1053 EVI-RT EC, the one that corresponds to the EVI on which the IGMP 1054 Leave was received. See Section 9.5 for details on how to form the 1055 EVI-RT EC. 1057 The Route Distinguisher (RD) SHOULD be a Type 1 RD [RFC4364]. The 1058 value field comprises an IP address of the PE (typically, the 1059 loopback address) followed by a number unique to the PE. 1061 The Ethernet Segment Identifier (ESI) MUST be set to the 10-octet 1062 value defined for the ES. 1064 The Ethernet Tag ID MUST be set as follows: 1066 o EVI is VLAN-Based or VLAN Bundle service - set to 0 1068 o EVI is VLAN-Aware Bundle service without translation - set to the 1069 customer VID for the BD 1071 o EVI is VLAN-Aware Bundle service with translation - set to the 1072 normalized Ethernet Tag ID - e.g., normalized VID 1074 The Multicast Source length MUST be set to length of multicast source 1075 address in bits. If the Multicast Source field contains an IPv4 1076 address, then the value of the Multicast Source Length field is 32. 1077 If the Multicast Source field contains an IPv6 address, then the 1078 value of the Multicast Source Length field is 128. In case of a (*, 1079 G) Join, the Multicast Source Length is set to 0. 1081 The Multicast Source is the Source IP address of the IGMP membership 1082 report. In case of a (*, G) Join, this field does not exist. 1084 The Multicast Group length MUST be set to length of multicast group 1085 address in bits. If the Multicast Group field contains an IPv4 1086 address, then the value of the Multicast Group Length field is 32. 1087 If the Multicast Group field contains an IPv6 address, then the value 1088 of the Multicast Group Length field is 128. 1090 The Multicast Group is the Group address of the IGMP membership 1091 report. 1093 The Originator Router Length is the length of the Originator Router 1094 address in bits. 1096 The Originator Router Address is the IP address of Router Originating 1097 the prefix. 1099 Reserved field is not part of the route key. The originator MUST set 1100 the reserved field to Zero , the receiver SHOULD ignore it and if it 1101 needs to be propagated, it MUST propagate it unchanged 1103 Maximum Response Time is value to be used while sending query as 1104 defined in [RFC2236] 1106 The Flags field indicates the version of IGMP protocol from which the 1107 membership report was received. It also indicates whether the 1108 multicast group had INCLUDE or EXCLUDE bit set. 1110 9.4. Multicast Flags Extended Community 1112 The 'Multicast Flags' extended community is a new EVPN extended 1113 community. EVPN extended communities are transitive extended 1114 communities with a Type field value of 6. IANA will assign a Sub- 1115 Type from the 'EVPN Extended Community Sub-Types' registry. 1117 A PE that supports IGMP proxy on a given BD MUST attach this extended 1118 community to the Inclusive Multicast Ethernet Tag (IMET) route it 1119 advertises for that BD and it MUST set the IGMP Proxy Support flag to 1120 1. Note that an [RFC7432] compliant PE will not advertise this 1121 extended community so its absence indicates that the advertising PE 1122 does not support IGMP Proxy. 1124 The advertisement of this extended community enables more efficient 1125 multicast tunnel setup from the source PE specially for ingress 1126 replication - i.e., if an egress PE supports IGMP proxy but doesn't 1127 have any interest in a given (x,G), it advertises its IGMP proxy 1128 capability using this extended community but it does not advertise 1129 any SMET route for that (x,G). When the source PE (ingress PE) 1130 receives such advertisements from the egress PE, it does not 1131 replicate the multicast traffic to that egress PE; however, it does 1132 replicate the multicast traffic to the egress PEs that don't 1133 advertise such capability even if they don't have any interests in 1134 that (x,G). 1136 A Multicast Flags extended community is encoded as an 8-octet value, 1137 as follows: 1139 1 2 3 1140 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 1141 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1142 | Type=0x06 | Sub-Type=0x09| Flags (2 Octets) |M|I| 1143 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1144 | Reserved=0 | 1145 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1147 The low-order (lease significant) two bits are defined as the "IGMP 1148 Proxy Support and MLD Proxy Support" bit. The absence of this 1149 extended community also means that the PE does not support IGMP 1150 proxy. where: 1152 o Type is 0x06 as registered with IANA for EVPN Extended 1153 Communities. 1155 o Sub-Type : 0x09 1157 o Flags are two Octets value. 1159 * Bit 15 (shown as I) defines IGMP Proxy Support. Value of 1 for 1160 bit 15 means that PE supports IGMP Proxy. Value of 0 for bit 1161 15 means that PE does not supports IGMP Proxy. 1163 * Bit 14 (shown as M) defines MLD Proxy Support. Value of 1 for 1164 bit 14 means that PE supports MLD Proxy. Value of 0 for bit 14 1165 means that PE does not support MLD proxy. 1167 * Bit 0 to 13 are reserved for future. Sender MUST set it 0 and 1168 receiver MUST ignore it. 1170 o Reserved bits are set to 0. Sender MUST set it to 0 and receiver 1171 MUST ignore it. 1173 If a router does not support this specification, it MUST not add 1174 Multicast Flags Extended Community in BGP route. A router receiving 1175 BGP update , if M and I both flag are zero (0), the router MUST treat 1176 this Update as malformed . Receiver of such update MUST ignore the 1177 extended community. 1179 9.5. EVI-RT Extended Community 1181 In EVPN, every EVI is associated with one or more Route Targets 1182 (RTs). These Route Targets serve two functions: 1184 1. Distribution control: RTs control the distribution of the routes. 1185 If a route carries the RT associated with a particular EVI, it 1186 will be distributed to all the PEs on which that EVI exists. 1188 2. EVI identification: Once a route has been received by a 1189 particular PE, the RT is used to identify the EVI to which it 1190 applies. 1192 An IGMP Join Synch or IGMP Leave Synch route is associated with a 1193 particular combination of ES and EVI. These routes need to be 1194 distributed only to PEs that are attached to the associated ES. 1195 Therefore these routes carry the ES-Import RT for that ES. 1197 Since an IGMP Join Synch or IGMP Leave Synch route does not need to 1198 be distributed to all the PEs on which the associated EVI exists, 1199 these routes cannot carry the RT associated with that EVI. 1200 Therefore, when such a route arrives at a particular PE, the route's 1201 RTs cannot be used to identify the EVI to which the route applies. 1202 Some other means of associating the route with an EVI must be used. 1204 This document specifies four new Extended Communities (EC) that can 1205 be used to identify the EVI with which a route is associated, but 1206 which do not have any effect on the distribution of the route. These 1207 new ECs are known as the "Type 0 EVI-RT EC", the "Type 1 EVI-RT EC", 1208 the "Type 2 EVI-RT EC", and the "Type 3 EVI-RT EC". 1210 1. A Type 0 EVI-RT EC is an EVPN EC (type 6) of sub-type 0xA. 1212 2. A Type 1 EVI-RT EC is an EVPN EC (type 6) of sub-type 0xB. 1214 3. A Type 2 EVI-RT EC is an EVPN EC (type 6) of sub-type 0xC. 1216 4. A Type 3 EVI-RT EC is an EVPN EC (type 6) of sub-type 0xD 1218 Each IGMP Join Synch or IGMP Leave Synch route MUST carry exactly one 1219 EVI-RT EC. The EVI-RT EC carried by a particular route is 1220 constructed as follows. Each such route is the result of having 1221 received an IGMP Join or an IGMP Leave message from a particular BD. 1222 The route is said to be associated associated with that BD. For each 1223 BD, there is a corresponding RT that is used to ensure that routes 1224 "about" that BD are distributed to all PEs attached to that BD. So 1225 suppose a given IGMP Join Synch or Leave Synch route is associated 1226 with a given BD, say BD1, and suppose that the corresponding RT for 1227 BD1 is RT1. Then: 1229 o 0. If RT1 is a Transitive Two-Octet AS-specific EC, then the EVI- 1230 RT EC carried by the route is a Type 0 EVI-RT EC. The value field 1231 of the Type 0 EVI-RT EC is identical to the value field of RT1. 1233 o 1. If RT1 is a Transitive IPv4-Address-specific EC, then the EVI- 1234 RT EC carried by the route is a Type 1 EVI-RT EC. The value field 1235 of the Type 1 EVI-RT EC is identical to the value field of RT1. 1237 o 2. If RT1 is a Transitive Four-Octet-specific EC, then the EVI-RT 1238 EC carried by the route is a Type 2 EVI-RT EC. The value field of 1239 the Type 2 EVI-RT EC is identical to the value field of RT1. 1241 o 3. If RT1 is a Transitive IPv6-Address-specific EC, then the EVI- 1242 RT EC carried by the route is a Type 3 EVI-RT EC. The value field 1243 of the Type 3 EVI-RT EC is identical to the value field of RT1. 1245 An IGMP Join Synch or Leave Synch route MUST carry exactly one EVI-RT 1246 EC. 1248 Suppose a PE receives a particular IGMP Join Synch or IGMP Leave 1249 Synch route, say R1, and suppose that R1 carries an ES-Import RT that 1250 is one of the PE's Import RTs. If R1 has no EVI-RT EC, or has more 1251 than one EVI-RT EC, the PE MUST apply the "treat-as-withdraw" 1252 procedure of [RFC7606]. 1254 Note that an EVI-RT EC is not a Route Target Extended Community, is 1255 not visible to the RT Constrain mechanism [RFC4684] , and is not 1256 intended to influence the propagation of routes by BGP. 1258 1 2 3 1259 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 1260 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1261 | Type=0x06 | Sub-Type=n | RT associated with EVI | 1262 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1263 | RT associated with the EVI (cont.) | 1264 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1266 Where the value of 'n' is 0x0A, 0x0B, 0x0C, or 0x0D corresponding to 1267 EVI-RT type 0, 1, 2, or 3 respectively. 1269 9.6. Rewriting of RT ECs and EVI-RT ECs by ASBRs 1271 There are certain situations in which an ES is attached to a set of 1272 PEs that are not all in the same AS, or not all operated by the same 1273 provider. In some such situations, the RT that corresponds to a 1274 particular EVI may be different in each AS. If a route is propagated 1275 from AS1 to AS2, an ASBR at the AS1/AS2 border may be provisioned 1276 with a policy that removes the RTs that are meaningful in AS1 and 1277 replaces them with the corresponding (i.e., RTs corresponding to the 1278 same EVIs) RTs that are meaningful in AS2. This is known as RT- 1279 rewriting. 1281 Note that if a given route's RTs are rewritten, and the route carries 1282 an EVI-RT EC, the EVI-RT EC needs to be rewritten as well. 1284 9.7. BGP Error Handling 1286 If a received BGP update contains Flags not in accordance with IGMP 1287 version-X expectation, the PE MUST apply the "treat-as-withdraw" 1288 procedure as per [RFC7606] 1290 If a received BGP update is malformed such that BGP route keys cannot 1291 be extracted, then BGP update MUST be considered as invalid. 1292 Receiving PE MUST apply the "Session reset" procedure of [RFC7606]. 1294 10. IGMP/MLD Immediate Leave 1296 IGMP MAY be configured with immediate leave option. This allows the 1297 device to remove the group entry from the multicast routing table 1298 immediately upon receiving a IGMP leave message for (x,G). In case 1299 of all active multi-homing while synchronizing the IGMP Leave state 1300 to redundancy peers, Maximum Response Time MAY be filled in as Zero. 1301 Implementations SHOULD have identical configuration across multi- 1302 homed peers. In case IGMP Leave Synch route is received with Maximum 1303 Response Time Zero, irrespective of local IGMP configuration it MAY 1304 be processed as an immediate leave. 1306 11. IGMP Version 1 Membership Request 1308 This document does not provide any detail about IGMPv1 processing. 1309 Multicast working group are in process of deprecating uses of IGMPv1. 1310 Implementations MUST only use IGMPv2 and above for IPv4 and MLDv1 and 1311 above for IPv6. IGMP V1 routes MUST be considered as invalid and the 1312 PE MUST apply the "treat-as-withdraw" procedure as per [RFC7606] 1314 12. Security Considerations 1316 Same security considerations as [RFC7432] ,[RFC2236] ,[RFC3376] , 1317 [RFC2710], [RFC3810]. 1319 13. IANA Considerations 1321 IANA has allocated the following codepoints from the EVPN Extended 1322 Community sub-types registry. 1324 0x09 Multicast Flags Extended Community [this document] 1325 0x0A EVI-RT Type 0 [this document] 1326 0x0B EVI-RT Type 1 [this document] 1327 0x0C EVI-RT Type 2 [this document] 1329 IANA is requested to allocate a new codepoint from the EVPN Extended 1330 Community sub-types registry for the following. 1332 0x0D EVI-RT Type 3 [this document] 1334 IANA has allocated the following EVPN route types from the EVPN Route 1335 Type registry. 1337 6 - Selective Multicast Ethernet Tag Route 1338 7 - Multicast Join Synch Route 1339 8 - Multicast Leave Synch Route 1341 The Multicast Flags Extended Community contains a 16-bit Flags field. 1342 The bits are numbered 0-15, from high-order to low-order. 1344 The registry should be initialized as follows: 1345 Bit Name Reference 1346 ---- -------------- ------------- 1347 0 - 13 Unassigned 1348 14 MLD Proxy Support This document 1349 15 IGMP Proxy Support This document 1351 The registration policy should be "First Come First Served". 1353 14. Acknowledgement 1355 The authors would like to thank Stephane Litkowski, Jorge Rabadan, 1356 Anoop Ghanwani, Jeffrey Haas, Krishna Muddenahally Ananthamurthy for 1357 reviewing and providing valuable comment. 1359 15. Contributors 1361 Mankamana Mishra 1363 Cisco systems 1365 Email: mankamis@cisco.com 1366 Derek Yeung 1368 Arrcus 1370 Email: derek@arrcus.com 1372 16. References 1374 16.1. Normative References 1376 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1377 Requirement Levels", BCP 14, RFC 2119, 1378 DOI 10.17487/RFC2119, March 1997, 1379 . 1381 [RFC2236] Fenner, W., "Internet Group Management Protocol, Version 1382 2", RFC 2236, DOI 10.17487/RFC2236, November 1997, 1383 . 1385 [RFC2710] Deering, S., Fenner, W., and B. Haberman, "Multicast 1386 Listener Discovery (MLD) for IPv6", RFC 2710, 1387 DOI 10.17487/RFC2710, October 1999, 1388 . 1390 [RFC3376] Cain, B., Deering, S., Kouvelas, I., Fenner, B., and A. 1391 Thyagarajan, "Internet Group Management Protocol, Version 1392 3", RFC 3376, DOI 10.17487/RFC3376, October 2002, 1393 . 1395 [RFC3810] Vida, R., Ed. and L. Costa, Ed., "Multicast Listener 1396 Discovery Version 2 (MLDv2) for IPv6", RFC 3810, 1397 DOI 10.17487/RFC3810, June 2004, 1398 . 1400 [RFC4364] Rosen, E. and Y. Rekhter, "BGP/MPLS IP Virtual Private 1401 Networks (VPNs)", RFC 4364, DOI 10.17487/RFC4364, February 1402 2006, . 1404 [RFC4684] Marques, P., Bonica, R., Fang, L., Martini, L., Raszuk, 1405 R., Patel, K., and J. Guichard, "Constrained Route 1406 Distribution for Border Gateway Protocol/MultiProtocol 1407 Label Switching (BGP/MPLS) Internet Protocol (IP) Virtual 1408 Private Networks (VPNs)", RFC 4684, DOI 10.17487/RFC4684, 1409 November 2006, . 1411 [RFC7432] Sajassi, A., Ed., Aggarwal, R., Bitar, N., Isaac, A., 1412 Uttaro, J., Drake, J., and W. Henderickx, "BGP MPLS-Based 1413 Ethernet VPN", RFC 7432, DOI 10.17487/RFC7432, February 1414 2015, . 1416 [RFC7606] Chen, E., Ed., Scudder, J., Ed., Mohapatra, P., and K. 1417 Patel, "Revised Error Handling for BGP UPDATE Messages", 1418 RFC 7606, DOI 10.17487/RFC7606, August 2015, 1419 . 1421 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 1422 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 1423 May 2017, . 1425 16.2. Informative References 1427 [RFC4541] Christensen, M., Kimball, K., and F. Solensky, 1428 "Considerations for Internet Group Management Protocol 1429 (IGMP) and Multicast Listener Discovery (MLD) Snooping 1430 Switches", RFC 4541, DOI 10.17487/RFC4541, May 2006, 1431 . 1433 Authors' Addresses 1435 Ali Sajassi 1436 Cisco Systems 1437 821 Alder Drive, 1438 MILPITAS, CALIFORNIA 95035 1439 UNITED STATES 1441 Email: sajassi@cisco.com 1443 Samir Thoria 1444 Cisco Systems 1445 821 Alder Drive, 1446 MILPITAS, CALIFORNIA 95035 1447 UNITED STATES 1449 Email: sthoria@cisco.com 1451 Keyur PAtel 1452 Arrcus 1453 UNITED STATES 1455 Email: keyur@arrcus.com 1456 John Drake 1457 Juniper Networks 1459 Email: jdrake@juniper.net 1461 Wen Lin 1462 Juniper Networks 1464 Email: wlin@juniper.net