idnits 2.17.1 draft-ietf-l3vpn-mvpn-global-table-mcast-00.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 : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (June 24, 2014) is 3594 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(See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Outdated reference: A later version (-15) exists of draft-ietf-idr-add-paths-09 -- Obsolete informational reference (is this intentional?): RFC 4601 (Obsoleted by RFC 7761) == Outdated reference: A later version (-05) exists of draft-ietf-l3vpn-mvpn-extranet-04 == Outdated reference: A later version (-17) exists of draft-ietf-mpls-seamless-mcast-13 Summary: 0 errors (**), 0 flaws (~~), 4 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 L3VPN Working Group Jeffrey Zhang 3 Internet Draft Lenny Giuliano 4 Intended Status: Standards Track Juniper Networks, Inc. 5 Expires: December 24, 2014 6 Eric C. Rosen 7 Karthik Subramanian 8 Cisco Systems, Inc. 10 Dante J. Pacella 11 Verizon 13 Jason Schiller 14 Google 16 June 24, 2014 18 Global Table Multicast with BGP-MVPN Procedures 20 draft-ietf-l3vpn-mvpn-global-table-mcast-00.txt 22 Abstract 24 RFC6513, RFC6514, and other RFCs describe protocols and procedures 25 which a Service Provider (SP) may deploy in order offer Multicast 26 Virtual Private Network (Multicast VPN or MVPN) service to its 27 customers. Some of these procedures use BGP to distribute VPN- 28 specific multicast routing information across a backbone network. 29 With a small number of relatively minor modifications, the very same 30 BGP procedures can also be used to distribute multicast routing 31 information that is not specific to any VPN. Multicast that is 32 outside the context of a VPN is known as "Global Table Multicast", or 33 sometimes simply as "Internet multicast". In this document, we 34 describe the modifications that are needed to use the MVPN BGP 35 procedures for Global Table Multicast. 37 Status of this Memo 39 This Internet-Draft is submitted to IETF in full conformance with the 40 provisions of BCP 78 and BCP 79. 42 Internet-Drafts are working documents of the Internet Engineering 43 Task Force (IETF), its areas, and its working groups. Note that 44 other groups may also distribute working documents as Internet- 45 Drafts. 47 Internet-Drafts are draft documents valid for a maximum of six months 48 and may be updated, replaced, or obsoleted by other documents at any 49 time. It is inappropriate to use Internet-Drafts as reference 50 material or to cite them other than as "work in progress." 52 The list of current Internet-Drafts can be accessed at 53 http://www.ietf.org/ietf/1id-abstracts.txt. 55 The list of Internet-Draft Shadow Directories can be accessed at 56 http://www.ietf.org/shadow.html. 58 Copyright and License Notice 60 Copyright (c) 2014 IETF Trust and the persons identified as the 61 document authors. All rights reserved. 63 This document is subject to BCP 78 and the IETF Trust's Legal 64 Provisions Relating to IETF Documents 65 (http://trustee.ietf.org/license-info) in effect on the date of 66 publication of this document. Please review these documents 67 carefully, as they describe your rights and restrictions with respect 68 to this document. Code Components extracted from this document must 69 include Simplified BSD License text as described in Section 4.e of 70 the Trust Legal Provisions and are provided without warranty as 71 described in the Simplified BSD License. 73 Table of Contents 75 1 Introduction .......................................... 4 76 2 Adapting MVPN Procedures to GTM ....................... 6 77 2.1 Use of Route Distinguishers ........................... 7 78 2.2 Use of Route Targets .................................. 7 79 2.3 UMH-eligible Routes ................................... 9 80 2.3.1 Routes of SAFI 1, 2 or 4 with MVPN ECs ................ 10 81 2.3.2 MVPN ECs on the Route to the Next Hop ................. 11 82 2.3.3 Non-BGP Routes as the UMH-eligible Routes ............. 12 83 2.3.4 Why SFS Does Not Apply to GTM ......................... 13 84 2.4 Inclusive and Selective Tunnels ....................... 14 85 2.5 I-PMSI A-D Routes ..................................... 14 86 2.5.1 Intra-AS I-PMSI A-D Routes ............................ 14 87 2.5.2 Inter-AS I-PMSI A-D Routes ............................ 15 88 2.6 S-PMSI A-D Routes ..................................... 15 89 2.7 Leaf A-D Routes ....................................... 15 90 2.8 Source Active A-D Routes .............................. 15 91 2.8.1 Finding the Originator of an SA A-D Route ............. 15 92 2.8.2 Optional Additional Constraints on Distribution ....... 16 93 2.9 C-multicast Source/Shared Tree Joins .................. 17 94 3 Differences from other MVPN-like GTM Procedures ....... 18 95 4 IANA Considerations ................................... 19 96 5 Security Considerations ............................... 19 97 6 Additional Contributors ............................... 20 98 7 Acknowledgments ....................................... 20 99 8 Authors' Addresses .................................... 21 100 9 References ............................................ 22 101 9.1 Normative References .................................. 22 102 9.2 Informative References ................................ 22 104 1. Introduction 106 [RFC4364] specifies architecture, protocols, and procedures that a 107 Service Provider (SP) can use to provide Virtual Private Network 108 (VPN) service to its customers. In that architecture, one or more 109 Customer Edge (CE) routers attach to a Provider Edge (PE) router. 110 Each CE router belongs to a single VPN, but CE routers from several 111 VPNs may attach to the same PE router. In addition, CEs from the 112 same VPN may attach to different PEs. BGP is used to carry VPN- 113 specific information among the PEs. Each PE router maintains a 114 separate Virtual Routing and Forwarding table (VRF) for each VPN to 115 which it is attached. 117 [RFC6513] and [RFC6514] extend the procedures of [RFC4364] to allow 118 the SP to provide multicast service to its VPN customers. The 119 customer's multicast routing protocol (e.g., PIM) is used to exchange 120 multicast routing information between a CE and a PE. The PE stores a 121 given customer's multicast routing information in the VRF for that 122 customer's VPN. BGP is used to distribute certain multicast-related 123 control information among the PEs that attach to a given VPN, and BGP 124 may also be used to exchange the customer multicast routing 125 information itself among the PEs. 127 While this multicast architecture was originally developed for VPNs, 128 it can also be used (with a small number of modifications to the 129 procedures) to distribute multicast routing information that is not 130 specific to VPNs. The purpose of this document is to specify the way 131 in which BGP MVPN procedures can be adapted to support non-VPN 132 multicast. 134 Multicast routing information that is not specific to VPNs is stored 135 in a router's "global table", rather than in a VRF; hence it is known 136 as "Global Table Multicast" (GTM). GTM is sometimes more simply 137 called "Internet multicast". However, we will avoid that term 138 because it suggests that the multicast data streams are available on 139 the "public" Internet. The procedures for GTM can certainly be used 140 to support multicast on the public Internet, but they can also be 141 used to support multicast streams that are not public, e.g., content 142 distribution streams offered by content providers to paid 143 subscribers. For the purposes of this document, all that matters is 144 that the multicast routing information is maintained in a global 145 table rather than in a VRF. 147 This architecture does assume that the network over which the 148 multicast streams travel can be divided into a "core network" and one 149 or more non-core parts of the network, which we shall call 150 "attachment networks". The multicast routing protocol used in the 151 attachment networks may not be the same as the one used in the core, 152 so we consider there to be a "protocol boundary" between the core 153 network and the attachment networks. We will use the term "Protocol 154 Boundary Router" (PBR) to refer to the core routers that are at the 155 boundary. We will use the term "Attachment Router" (AR) to refer to 156 the routers that are not in the core but that attach to the PBRs. 158 This document does not make any particular set of assumptions about 159 the protocols that the ARs and the PBRs use to exchange unicast and 160 multicast routing information with each other. For instance, 161 multicast routing information could be exchanged between an AR and a 162 PBR via PIM, IGMP, or even BGP. Multicast routing also depends on an 163 exchange of routes that are used for looking up the path to the root 164 of a multicast tree. This routing information could be exchanged 165 between an AR and a PBR via IGP, via EBGP, or via IBGP ([RFC6368]). 166 Note that if IBGP is used, the [RFC6368] "push/pop procedures" are 167 not necessary. 169 The PBRs are not necessarily "edge" routers, in the sense of 170 [RFC4364]. For example, they may be both be Autonomous System Border 171 Routers (ASBR). As another example, an AR may be an "access router" 172 attached to a PBR that is an OSPF Area Border Router (ABR). Many 173 other deployment scenarios are possible. However, the PBRs are 174 always considered to be delimiting a "backbone" or "core" network. A 175 multicast data stream from an AR is tunneled over the core network 176 from an Ingress PBR to one or more Egress PBRs. Multicast routing 177 information that a PBR learns from the ARs attached to it is stored 178 in the PBR's global table. The PBRs use BGP to distribute multicast 179 routing and auto-discovery information among themselves. This is 180 done following the procedures of [RFC6513], [RFC6514], and other MVPN 181 specifications, as modified in this document. 183 In general, PBRs follow the same MVPN/BGP procedures that PE routers 184 follow, except that these procedures are adapted to be applicable to 185 the global table rather than to a VRF. Details are provided in 186 subsequent sections of this document. 188 By supporting GTM using the BGP procedures designed for MVPN, one 189 obtains a single control plane that governs the use of both VPN and 190 non-VPN multicast. Most of the features and characteristics of MVPN 191 carry over automatically to GTM. These include scaling, aggregation, 192 flexible choice of tunnel technology in the SP network, support for 193 both segmented and non-segmented tunnels, ability to use wildcards to 194 identify sets of multicast flows, support for the Any Source 195 Multicast (ASM), Single Source Multicast (SSM), and Bidirectional 196 (bidir) multicast paradigms, support for both IPv4 and IPv6 multicast 197 flows over either an IPv4 or IPv6 SP infrastructure, support for 198 unsolicited flooded data (including support for BSR as RP-to-group 199 mapping protocols), etc. 201 This document not only uses MVPN procedures for GTM, but also, 202 insofar as possible, uses the same protocol elements, encodings, and 203 formats. The BGP Updates for GTM thus use the same Subsequent 204 Address Family Identifier (SAFI), and have the same Network Layer 205 Reachability Information (NLRI) format, as the BGP Updates for MVPN. 207 Details for supporting MVPN (either IPv4 or IPv6 MVPN traffic) over 208 an IPv6 backbone network can be found in [RFC6515]. The procedures 209 and encodings described therein are also applicable to GTM. 211 The document [SEAMLESS-MCAST] extends [RFC6514] by providing 212 procedures that allow tunnels through the core to be "segmented" at 213 ABRs within the core. The ABR segmentation procedures are also 214 applicable to GTM as defined in the current document. In general, 215 the MVPN procedures of [SEAMLESS-MCAST], adapted as specified in the 216 current document, are applicable to GTM. 218 The document [SEAMLESS-MCAST] also defines a set of procedures for 219 GTM. Those procedures are different from the procedures defined in 220 the current document, and the two sets of procedures are not 221 interoperable with each other. The two sets of procedures can co- 222 exist in the same network, as long as they are not applied to the 223 same multicast flows or to the same multicast group addresses. See 224 section 3 for more details. 226 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 227 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 228 document are to be interpreted as described in [RFC2119]. 230 2. Adapting MVPN Procedures to GTM 232 In general, PBRs support Global Table Multicast by using the 233 procedures that PE routers use to support VPN multicast. For GTM, 234 where [RFC6513] and [RFC6514] talk about the "PE-CE interface", one 235 should interpret that to mean the interface between the AR and the 236 PBR. For GTM, where [RFC6513] and [RFC6514] talk about the 237 "backbone" network, one should interpret that to mean the part of the 238 network that is delimited by the PBRs. 240 A few adaptations to the procedures of [RFC6513] and [RFC6514] need 241 to be made. Those adaptations are described in the following sub- 242 sections. 244 2.1. Use of Route Distinguishers 246 The MVPN procedures require the use of BGP routes, defined in 247 [RFC6514], that have a SAFI value of 5 ("MCAST-VPN"). We refer to 248 these simply as "MCAST-VPN routes". [RFC6514] defines the Network 249 Layer Reachability Information (NLRI) format for MCAST-VPN routes. 250 The NLRI field always begins with a "Route Type" octet, and, 251 depending on the route type, may be followed by a "Route 252 Distinguisher" (RD) field. 254 When a PBR originates an MCAST-VPN route in support of GTM, the RD 255 field (for those routes types where it is defined) of that route's 256 NLRI MUST be set to zero (i.e., to 64 bits of zero). Since no VRF 257 may have an RD of zero, this allows "MCAST-VPN" routes that are 258 "about" GTM to be distinguished from MCAST-VPN routes that are about 259 VPNs. 261 2.2. Use of Route Targets 263 The MVPN procedures require all MCAST-VPN routes to carry Route 264 Targets (RTs). When a PE router receives an MCAST-VPN route, it 265 processes the route in the context of a particular VRF if and only if 266 the route is carrying an RT that is configured as one of that VRF's 267 "import RTs". 269 There are two different "kinds" of RT used in MVPN. 271 - One kind of RT is carried only by the following MCAST-VPN route 272 types: C-multicast Shared Tree Joins, C-multicast Source Tree 273 Joins, and Leaf A-D routes. This kind of RT identifies the PE 274 router that has been selected by the route's originator as the 275 "Upstream PE" or as the "Upstream Multicast Hop" (UMH) for a 276 particular (set of) multicast flow(s). Per [RFC6514] and 277 [RFC6515], this RT must be an IPv4-address-specific or 278 IPv6-address-specific Extended Community (EC), whose "Global 279 Administrator" field identifies the Upstream PE or the UMH. If 280 the Global Administrator field identifies the Upstream PE, the 281 "Local Administrator" field identifies a particular VRF in that 282 PE. 284 The GTM procedures of this document require the use of this type 285 of RT, in exactly the same situations where it is used in the 286 MVPN specification. However, one adaptation is necessary: the 287 "Local Administrator" field of this kind of RT MUST always be set 288 to zero, thus implicitly identifying the global table, rather 289 than identifying a VRF. We will refer to this kind of RT as a 290 "PBR-identifying RT". 292 - The other kind of RT is the conventional RT first specified in 293 [RFC4364]. It does not necessarily identify a particular router 294 by address, but is used to constrain the distribution of VPN 295 routes, and to ensure that a given VPN route is processed in the 296 context of a given VRF if and only if the route is carrying an RT 297 that has been configured as one of that VRF's "import RTs". 299 Whereas every VRF must be configured with at least one import RT, 300 there is heretofore no requirement to configure any RTs for the 301 global table of any router. As stated above, this document makes 302 the use of PBR-identifying RTs mandatory for GTM. This document 303 makes the use of non-PBR-identifying RTs OPTIONAL for GTM. 305 The procedures for the use of RTs in GTM are the following: 307 - If the global table of a particular PBR is NOT configured with 308 any import RTs, then a received MCAST-VPN route is processed in 309 the context of the global table only if it is carrying no RTs, or 310 if it is carrying a PBR-identifying RT whose Global Administrator 311 field identifies that PBR. 313 - The global table in each PBR MAY be configured with (a) a set of 314 export RTs to be attached to MCAST-VPN routes that are originated 315 to support GTM, and (b) with a set of import RTs for GTM. 317 If the global table of a given PBR has been so configured, the 318 PBR will process a received MCAST-VPN route in the context of the 319 global table if and only if the route carries an RT that is one 320 of the global table's import RTs, or if the route carries a PBR- 321 identifying RT whose global administrator field identifies the 322 PBR. 324 If the global tables are configured with RTs, care must be taken 325 to ensure that the RTs configured for the global table are 326 distinct from any RTs used in support of MVPN (except in the case 327 where it is actually intended to create an "extranet" 328 [MVPN-extranet] in which some sources are reachable in global 329 table context while others are reachable in VPN context.) 331 The "RT Constraint" procedures of [RFC4684] MAY be used to constrain 332 the distribution of MCAST-VPN routes (or other routes) that carry RTs 333 that have been configured as import RTs for GTM. (This includes the 334 PBR-identifying RTs.) 336 In [RFC6513], the UMH-eligible routes (see section 5.1 of [RFC6513], 337 "Eligible Routes for UMH Selection") are generally routes of SAFI 128 338 (Labeled VPN-IP routes) or 129 (VPN-IP multicast routes), and are 339 required to carry RTs. These RTs determine which VRFs import which 340 such routes. However, for GTM, when the UMH-eligible routes may be 341 routes of SAFI 1, 2, or 4, the routes are not required to carry RTs. 342 This document does NOT specify any new rules for determine whether a 343 SAFI 1, 2, or 4 route is to be imported into the global table of any 344 PBR. 346 2.3. UMH-eligible Routes 348 [RFC6513] section 5.1 defines procedures by which a PE router 349 determines the "C-root", the "Upstream Multicast Hop" (UMH), the 350 "Upstream PE", and the "Upstream RD" of a given multicast flow. (In 351 non-VPN multicast documents, the UMH of a multicast flow at a 352 particular router is generally known as the "RPF neighbor" for that 353 flow.) It also defines procedures for determining the "Source AS" of 354 a particular flow. Note that in GTM, the "Upstream PE" is actually 355 the "Upstream PBR". 357 The definition of the C-root of a flow is the same for GTM as for 358 MVPN. 360 For MVPN, to determine the UMH, Upstream PE, Upstream RD, and Source 361 AS of a flow, one looks up the C-root of the flow in a particular 362 VRF, and finds the "UMH-eligible" routes (see section 5.1.1 of 363 [RFC6513]) that "match" the C-root. From among these, one is chosen 364 as the "selected UMH route". 366 For GTM, the C-root is of course looked up in the global table, 367 rather than in a VRF. For MVPN, the UMH-eligible routes are routes 368 of SAFI 128 or 129. For GTM, the UMH-eligible routes are routes of 369 SAFI 1, SAFI 4, or SAFI 2. If the global table has imported routes 370 of SAFI 2, then these are the UMH-eligible routes. Otherwise, routes 371 of SAFI 1 or SAFI 4 are the UMH-eligible routes. For the purpose of 372 UMH determination, if a SAFI 1 route and a SAFI 4 route contain the 373 same IP prefix in their respective NLRI fields, then the two routes 374 are considered by the BGP bestpath selection process to be 375 comparable. 377 [RFC6513] defines procedures for determining which of the UMH- 378 eligible routes that match a particular C-root is to become the 379 "Selected UMH route". With one exception, these procedures are also 380 applicable to GTM. The one exception is the following. Section 381 9.1.2 of [RFC6513] defines a particular method of choosing the 382 Upstream PE, known as "Single Forwarder Selection" (SFS). This 383 procedure MUST NOT be used for GTM (see section 2.3.4 for an 384 explanation of why the SFS procedure cannot be applied to GTM). 386 In GTM, the "Upstream RD" of a multicast flow is always considered to 387 be zero, and is NOT determined from the Selected UMH route. 389 The MVPN specifications require that when BGP is used for 390 distributing multicast routing information, the UMH-eligible routes 391 MUST carry the VRF Route Import EC and the Source AS EC. To 392 determine the Upstream PE and Source AS for a particular multicast 393 flow, the Upstream PE and Source AS are determined, respectively, 394 from the VRF Route Import EC and the Source AS EC of the Selected UMH 395 route for that flow. These ECs are generally attached to the UMH- 396 eligible routes by the PEs that originate the routes. 398 In GTM, there are certain situations in which it is allowable to omit 399 the VRF Route Import EC and/or the Source AS EC from the UMH-eligible 400 routes. The following sub-sections specify the various options for 401 determining the Upstream PBR and the Source AS in GTM. 403 The procedures in sections 2.3.1 MUST be implemented. The procedures 404 in sections 2.3.2 and 2.3.3 are OPTIONAL to implement. It should be 405 noted that while the optional procedures may be useful in particular 406 deployment scenarios, there is always the potential for 407 interoperability problems when relying on OPTIONAL procedures. 409 2.3.1. Routes of SAFI 1, 2 or 4 with MVPN ECs 411 If the UMH-eligible routes have a SAFI of 1, 2 or 4, then they MAY 412 carry the VRF Route Import EC and/or the Source AS EC. If the 413 selected UMH route is a route of SAFI 1, 2 or 4 that carries the VRF 414 Route Import EC, then the Upstream PBR is determined from that EC. 415 Similarly, if the selected UMH route is a route of SAFI 1, 2, or 4 416 route that carries the Source AS EC, the Source AS is determined from 417 that EC. 419 When the procedure of this section is used, a PBR that distributes a 420 UMH-eligible route to other PBRs is responsible for ensuring that the 421 VRF Route Import and Source AS ECs are attached to it. 423 If the selected UMH-eligible route has a SAFI of 1, 2 or 4, but is 424 not carrying a VRF Route Import EC, then the Upstream PBR is 425 determined as specified in section 2.3.2 or 2.3.3 below. 427 If the selected UMH-eligible route has a SAFI of 1, 2 or 4, but is 428 not carrying a Source AS EC, then the Source AS is considered to be 429 the local AS. 431 2.3.2. MVPN ECs on the Route to the Next Hop 433 Some service providers may consider it to be undesirable to have the 434 PBRs put the VRF Route Import EC on all the UMH-eligible routes. Or 435 there may be deployment scenarios in which the UMH-eligible routes 436 are not advertised by the PBRs at all. The procedures described in 437 this section provide an alternative that can be used under certain 438 circumstances. 440 The procedures of this section are OPTIONAL. 442 In this alternative procedure, each PBR MUST originate a BGP route of 443 SAFI 1, 2 or 4 to itself. This route MUST carry a VRF Route Import 444 EC that identifies the PBR. The address that appears in the Global 445 Administrator field of that EC MUST be the same address that appears 446 in the NLRI and in the Next Hop field of that route. This route MUST 447 also carry a Source AS EC identifying the AS of the PBR. 449 Whenever the PBR distributes a UMH-eligible route for which it sets 450 itself as next hop, it MUST use this same IP address as the Next Hop 451 of the UMH-eligible route that it used in the route discussed in the 452 prior paragraph. 454 When the procedure of his section is used, then when a PBR is 455 determining the Selected UMH Route for a given multicast flow, it may 456 find that the Selected UMH Route has no VRF Route Import EC. In this 457 case, the PBR will look up (in the global table) the route to the 458 Next Hop of the Selected UMH route. If the route to the Next Hop has 459 a VRF Route Import EC, that EC will be used to determine the Upstream 460 PBR, just as if the EC had been attached to the Selected UMH Route. 462 If recursive route resolution is required in order to resolve the 463 next hop, the Upstream PBR will be determined from the first route 464 with a VRF Route Import EC that is encountered during the recursive 465 route resolution process. (The recursive route resolution process 466 itself is not modified by this document.) 468 The same procedure can be applied to find the Source AS, except that 469 the Source AS EC is used instead of the VRF Route Import EC. 471 Note that this procedure is only applicable in scenarios where it is 472 known that the Next Hop of the UMH-eligible routes is not be changed 473 by any router that participates in the distribution of those routes; 474 this procedure MUST NOT be used in any scenario where the next hop 475 may be changed between the time one PBR distributes the route and 476 another PBR receives it. The PBRs have no way of determining 477 dynamically whether the procedure is applicable in a particular 478 deployment; this must be made known to the PBRs by provisioning. 480 Some scenarios in which this procedure can be used are: 482 - all PBRs are in the same AS, or 484 - the UMH-eligible routes are distributed among the PBRs by a Route 485 Reflector (that does not change the next hop), or 487 - the UMH-eligible routes are distributed from one AS to another 488 through ASBRs that do not change the next hop. 490 If the procedures of this section are used in scenarios where they 491 are not applicable, GTM will not function correctly. 493 2.3.3. Non-BGP Routes as the UMH-eligible Routes 495 In particular deployment scenarios, there may be specific procedures 496 that can be used, in those particular scenarios, to determine the 497 Upstream PBR for a given multicast flow. 499 Suppose the PBRs neither put the VRF Route Import EC on the UMH- 500 eligible routes, nor do they distribute BGP routes to themselves. It 501 may still be possible to determine the Upstream PBR for a given 502 multicast flow, using specific knowledge about the deployment. 504 For example, suppose it is known that all the PBRs are in the same 505 OSPF area. It may be possible to determine the Upstream PBR for a 506 given multicast flow by looking at the link state database to see 507 which router is attached to the flow's C-root. 509 As another example, suppose it is known that the set of PBRs is fully 510 meshed via Traffic Engineering (TE) tunnels. When a PBR looks up, in 511 its global table, the C-root of a particular multicast flow, it may 512 find that the next hop interface is a particular TE tunnel. If it 513 can determine the identify of the router at the other end of that TE 514 tunnel, it can deduce that that router is the Upstream PBR for that 515 flow. 517 This is not an exhaustive set of examples. Any procedure that 518 correctly determines the Upstream PBR in a given deployment scenario 519 MAY be used in that scenario. 521 2.3.4. Why SFS Does Not Apply to GTM 523 To see why the SFS procedure cannot be applied to GTM, consider the 524 following example scenario. Suppose some multicast source S is homed 525 to both PBR1 and PBR2, and suppose that both PBRs export a route (of 526 SAFI 1, 2, or 4) whose NLRI is a prefix matching the address of S. 527 These two routes will be considered comparable by the BGP decision 528 process. A route reflector receiving both routes may thus choose to 529 redistribute just one of the routes to S, the one chosen by the 530 bestpath algorithm. Different route reflectors may even choose 531 different routes to redistribute (i.e., one route reflector may 532 choose the route to S via PBR1 as the bestpath, while another chooses 533 the route to S via PBR2 as the bestpath). As a result, some PBRs may 534 receive only the route to S via PBR1 and some may receive only the 535 route to S via PBR2. In that case, it is impossible to ensure that 536 all PBRs will choose the same route to S. 538 The SFS procedure works in VPN context as along the following 539 assumption holds: if S is homed to VRF-x in PE1 and to VRF-y in PE2, 540 then VRF-x and VRF-y have been configured with different RDs. In VPN 541 context, the route to S is of SAFI 128 or 129, and thus has an RD in 542 its NLRI. So the route to S via PE1 will not have the same NLRI as 543 the route to S via PE2. As a result, all PEs will see both routes, 544 and the PEs can implement a procedure that ensures that they all pick 545 the same route to S. 547 That is, the SFS procedure of [RFC6513] relies on the UMH-eligible 548 routes being of SAFI 128 or 129, and relies on certain VRFs being 549 configured with distinct RDs. Thus the procedure cannot be applied 550 to GTM. 552 One might think that the SFS procedure could be applied to GTM as 553 long as the procedures defined in [ADD-PATH] are applied to the UMH- 554 eligible routes. Using the [ADD-PATH] procedures, the BGP speakers 555 could advertise more than one path to a given prefix. Typically 556 [ADD-PATH] is used to report the n best paths, for some small value 557 of n. However, this is not sufficient to support SFS, as can be seen 558 by examining the following scenario. 560 AS-X | AS-Y | AS-Z 561 | | 562 S--PBR1---ASBR1--|--ASBR2--|---ASBR5 563 | \______/ | | 564 | / \ | | 565 |--PBR2---ASBR3--|--ASBR4--|---ASBR6 566 | | 568 In AS-X, PBR1 reports to both ASBR1 and ASBR3 that it has a route to 569 S. Similarly, PBR2 reports to both ASBR1 and ASBR3 that it has a 570 route to S. Using [ADD-PATH], ASBR1 reports both routes to ASBR2, 571 and ASBR3 reports both routes to ASBR4. Now AS-Y sees 4 paths to S. 572 The AS-Z ASBRs will each see eight paths (four via ASBR2 and four via 573 ASBR4). To avoid this explosion in the number of paths, a BGP 574 speaker that uses [ADD-PATH] is usually considered to report only the 575 n best paths. However, there is then no guarantee that the reported 576 set of paths will contain at least one path via PBR1 and at least one 577 path via PBR2. Without such a guarantee, the SFS procedure will not 578 work. 580 2.4. Inclusive and Selective Tunnels 582 The MVPN specifications allow multicast flows to be carried on either 583 Inclusive Tunnels or on Selective Tunnels. When a flow is sent on an 584 Inclusive Tunnel of a particular VPN, it is sent to all PEs in that 585 VPN. When sent on a Selective Tunnel of a particular VPN, it may be 586 sent to only a subset of the PEs in that VPN. 588 This document allows the use of either Inclusive Tunnels or Selective 589 Tunnels for GTM. However, any service provider electing to use 590 Inclusive Tunnels for GTM should carefully consider whether sending a 591 multicast flow to ALL its PBRs would result in problems of scale. 592 There are potentially many more MBRs for GTM than PEs for a 593 particular VPN. If the set of PBRs is large and growing, but most 594 multicast flows do not need to go to all the PBRs, the exclusive use 595 of Selective Tunnels may be a better option. 597 2.5. I-PMSI A-D Routes 599 2.5.1. Intra-AS I-PMSI A-D Routes 601 Per [MVPN-BGP}, there are certain conditions under which is it NOT 602 required for a PE router implementing MVPN to originate one or more 603 Intra-AS I-PMSI A-D routes. These conditions apply as well to PBRs 604 implementing GTM. 606 In addition, a PBR implementing GTM is NOT required to originate an 607 Intra-AS I-PMSI A-D route if both of the following conditions hold: 609 - The PBR is not using Inclusive Tunnels for GTM, and 610 - The distribution of the C-multicast Shared Tree Join and C- 611 multicast Source Tree Join routes is done in such a manner that 612 the next hop of those routes does not change. 614 Please see also the sections on RD and RT usage. 616 2.5.2. Inter-AS I-PMSI A-D Routes 618 There are no GTM-specific procedures for the origination, 619 distribution, and processing of these routes, other than those 620 specified in the sections on RD and RT usage. 622 2.6. S-PMSI A-D Routes 624 There are no GTM-specific procedures for the origination, 625 distribution, and processing of these routes, other than those 626 specified in the sections on RD and RT usage. 628 2.7. Leaf A-D Routes 630 There are no GTM-specific procedures for the origination, 631 distribution, and processing of these routes, other than those 632 specified in the sections on RD and RT usage. 634 2.8. Source Active A-D Routes 636 Please see the sections on RD and RT usage for information applies to 637 the origination and distribution of Source Active A-D routes. 638 Additional procedures governing the use of Source Active A-D routes 639 are given in the sub-sections of this section. 641 2.8.1. Finding the Originator of an SA A-D Route 643 To carry out the procedures specified in [RFC6514] (e.g., in Section 644 13.2 of that document), it is sometimes necessary for an egress PE to 645 determine the ingress PE that originated a given Source Active A-D 646 route. The procedure used in [RFC6514] to find the originator of a 647 Source Active A-D route assumes that no two routes have the same RD 648 unless they have been originated by the same PE. However, this 649 assumption is not valid in GTM, because each Source Active A-D route 650 used for GTM will have an RD of 0, and all the UMH-eligible routes 651 also have an RD of 0. So GTM requires a different procedure for 652 determining the originator of a Source Active A-D route. 654 In GTM, the procedure for determining the originating PE of a Source 655 Active A-D route is the following: 657 - When a Source Active A-D route is originated, the originating PE 658 MAY attach a VRF Route Import Extended Community to the route. 660 - When a Source Active A-D route is distributed by one BGP speaker 661 to another, then 663 * if the Source Active A-D route does not carry the VRF Route 664 Import EC, the BGP speaker distributing the route MUST NOT 665 change the route's next hop field; 667 * if the Source Active A-D route does carry the VRF Route 668 Import EC, the BGP speaker distributing the route MAY change 669 the route's next hop field to itself. 671 - When an egress PE needs to determine the originator of a Source 672 Active A-D route, then 674 * if the Source Active A-D route carries the VRF Route Import 675 EC, the originating PE is the PE identified in the Global 676 Administrator field of that EC; 678 * if the Source Active A-D route does not carry the VRF Route 679 Import EC, the originating PE is the PE identified in the 680 route's next hop field. 682 2.8.2. Optional Additional Constraints on Distribution 684 If some site has receivers for a particular ASM group G, then it is 685 possible (by the procedures of [RFC6514]) that every PBR attached to 686 a site with a source for group G will originate a Source Active A-D 687 route whose NLRI identifies that source and group. These Source 688 Active A-D routes may be distributed to every PBR. If only a 689 relatively small number of PBRs are actually interested in traffic 690 from group G, but there are many sources for group G, this could 691 result in a large number of (S,G) Source Active A-D routes being 692 installed in a large number of PBRs that have no need of them. 694 For GTM, it is possible to constrain the distribution of (S,G) Source 695 Active A-D routes to those PBRs that are interested in GTM traffic to 696 group G. This can be done using the following OPTIONAL procedures: 698 - If a PBR originates a C-multicast Shared Tree Join whose NLRI 699 contains (RD=0,*,G), then it dynamically creates an import RT for 700 its global table, where the Global Administrator field of the RT 701 contains the group address G, and the Local Administrator field 702 contains zero. (Note that an IPv6-address-specific RT would need 703 to be used if the group address is an IPv6 address.) 705 - When a PBR creates such an import RT, it uses "RT Constraint" 706 [RFC4684] procedures to advertise its interest in routes that 707 carry this RT. 709 - When a PBR originates a Source Active A-D route from its global 710 table, it attaches the RT described above. 712 - When the C-multicast Shared Tree Join is withdrawn, so is the 713 corresponding RT constrain route, and the corresponding RT is 714 removed as an import RT of its global table. 716 These procedures enable a PBR to automatically filter all Source 717 Active A-D routes that are about multicast groups in which the PBR 718 has no interest. 720 This procedure does introduce the overhead of distributing additional 721 "RT Constraint" routes, and therefore may not be cost-effective in 722 all scenarios, especially if the number of sources per ASM group is 723 small. This procedure may also result in increased join latency. 725 2.9. C-multicast Source/Shared Tree Joins 727 [RFC6514] section 11.1.3 has the following procedure for determining 728 the IP-address-specific RT that is attached to a C-multicast route: 729 (a) determine the upstream PE, RD, AS, (b) find the proper Inter-AS 730 or Intra-AS I-PMSI A-D route based on (a), (c) find the next hop of 731 that A-D route, (d) base the RT on that next hop. 733 However, for GTM, in environments where it is known a priori that 734 that the next hop of the C-multicast Source/Shared Tree Joins does 735 not change during the distribution of those routes, the proper 736 procedure for creating the IP-address-specific RT is to just put the 737 IP Address of the Upstream PBR in the Global Administrator field of 738 the RT. In other scenarios, the procedure of the previous paragraph 739 (as modified by this document's sections on "RD usage" and "RT 740 usage") is applied by the PBRs. 742 3. Differences from other MVPN-like GTM Procedures 744 The document [SEAMLESS-MCAST] also defines a procedure for GTM that 745 is based on the BGP procedures that were developed for MVPN. 747 However, the GTM procedures of [SEAMLESS-MCAST] are different than 748 and are NOT interoperable with the procedures defined in this 749 document. 751 The two sets of procedures can co-exist in the same network, as long 752 as they are not applied to the same multicast flows or to the same 753 ASM multicast group addresses. 755 Some of the major differences between the two sets of procedures are 756 the following; 758 - The [SEAMLESS-MCAST] procedures for GTM do not use C-multicast 759 Shared Tree Joins or C-multicast Source Tree Joins at all. The 760 procedures of this document use these C-multicast routes for GTM, 761 setting the RD field of the NLRI to zero. 763 - The [SEAMLESS-MCAST] procedures for GTM use Leaf A-D routes 764 instead of C-multicast Shared/Source Tree Join routes. Leaf A-D 765 routes used in that manner can be distinguished from Leaf A-D 766 routes used as specified in [RFC6514] by means of the NLRI 767 format; [SEAMLESS-MCAST] defines a new NLRI format for Leaf A-D 768 routes. Whether a given Leaf A-D route is being used according 769 to the [SEAMLESS-MCAST] procedures or not can be determined from 770 its NLRI. (See [SEAMLESS-MCAST] section "Leaf A-D Route for 771 Global Table Multicast".) 773 - The Leaf A-D routes used by the current document contain an NLRI 774 that is in the format defined in [RFC6514], NOT in the format as 775 defined in [SEAMLESS-MCAST]. The procedures assumed by this 776 document for originating and processing Leaf A-D routes are as 777 specified in [RFC6514], NOT as specified in [SEAMLESS-MCAST]. 779 - The current document uses an RD value of zero in the NLRI in 780 order to indicate that a particular route is "about" a Global 781 Table Multicast, rather than a VPN multicast. No other semantics 782 are inferred from the fact that RD is zero. [SEAMLESS-MCAST] 783 uses two different RD values in its GTM procedures, with semantic 784 differences that depend upon the RD values. 786 - In order for both sets of procedures to co-exist in the same 787 network, the PBRs MUST be provisioned so that for any given IP 788 group address in the global table, all egress PBRs use the same 789 set of procedures for that group address (i.e., for group G, 790 either all egress PBRs use the GTM procedures of this document or 791 all egress PBRs use the GTM procedures of [SEAMLESS-MCAST]. 793 4. IANA Considerations 795 This document has no IANA considerations. 797 5. Security Considerations 799 The security considerations of this document are primarily the 800 security considerations of the base protocols, as discussed in 801 [RFC6514], [RFC4601], and [RFC5294]. 803 This document makes use of a BGP SAFI (MCAST-VPN routes) that was 804 originally designed for use in VPN contexts only. It also makes use 805 of various BGP path attributes and extended communities (VRF Route 806 Import Extended Community, Source AS Extended Community, Route Target 807 Extended Community) that were originally intended for use in VPN 808 contexts. If these routes and/or attributes leak out into "the 809 wild", multicast data flows may be distributed in an unintended 810 and/or unauthorized manner. 812 Internet providers often make extensive use of BGP communities (ie, 813 adding, deleting, modifying communities throughout a network). As 814 such, care should be taken to avoid deleting or modifying the VRF 815 Route Import Extended Community and Source AS Extended Community. 816 Incorrect manipulation of these ECs may result in multicast streams 817 being lost or misrouted. 819 The procedures of this document require certain BGP routes to carry 820 IP multicast group addresses. Generally such group addresses are 821 only valid within a certain scope. If a BGP route containing a group 822 address is distributed outside the boundaries where the group address 823 is meaningful, unauthorized distribution of multicast data flows may 824 occur. 826 6. Additional Contributors 828 Zhenbin Li 829 Huawei Technologies 830 Huawei Bld., No.156 Beiqing Rd. 831 Beijing 100095 832 China 833 Email: lizhenbin@huawei.com 835 Wei Meng 836 ZTE Corporation 837 No.50 Software Avenue, Yuhuatai District 838 Nanjing 839 China 840 Email: meng.wei2@zte.com.cn,vally.meng@gmail.com 842 Cui Wang 843 ZTE Corporation 844 No.50 Software Avenue, Yuhuatai District 845 Nanjing 846 China 847 Email: wang.cui1@zte.com.cn 849 Shunwan Zhuang 850 Huawei Technologies 851 Huawei Bld., No.156 Beiqing Rd. 852 Beijing 100095 853 China 854 Email: zhuangshunwan@huawei.com 856 7. Acknowledgments 858 The authors and contributors would like to thank Rahul Aggarwal, 859 Huajin Jeng, Hui Ni, Yakov Rekhter, and Samir Saad for their 860 contributions to this work. 862 8. Authors' Addresses 864 Lenny Giuliano 865 Juniper Networks 866 2251 Corporate Park Drive 867 Herndon, VA 20171 868 US 869 Email: lenny@juniper.net 871 Dante J. Pacella 872 Verizon 873 Verizon Communications 874 22001 Loudoun County Parkway 875 Ashburn, VA 20147 876 US 877 Email: dante.j.pacella@verizonbusiness.com 879 Eric C. Rosen 880 Cisco Systems, Inc. 881 1414 Massachusetts Avenue 882 Boxborough, MA, 01719 883 US 884 Email: erosen@cisco.com 886 Jason Schiller 887 Google 888 1818 Library Street 889 Suite 400 890 Reston, VA 20190 891 US 892 Email: jschiller@google.com 894 Karthik Subramanian 895 Cisco Systems, Inc. 896 170 Tasman Drive 897 San Jose, CA, 95134 898 US 899 Email: kartsubr@cisco.com 900 Jeffrey Zhang 901 Juniper Networks 902 10 Technology Park Dr. 903 Westford, MA 01886 904 US 905 Email: zzhang@juniper.net 907 9. References 909 9.1. Normative References 911 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 912 Requirement Levels", BCP 14, RFC 2119, March 1997. 914 [RFC4364], Rosen, E. and Y. Rekhter, "BGP/MPLS IP Virtual Private 915 Networks", RFC 4364, February 2006. 917 [RFC6513] Rosen, E. and R. Aggarwal, "Multicast in MPLS/BGP IP VPNs", 918 RFC 6513, February 2012. 920 [RFC6514] Aggarwal, R., Rosen, E., Morin, T., and Y. Rekhter, "BGP 921 Encodings and Procedures for Multicast in MPLS/BGP IP VPNs", RFC 922 6514, February 2012. 924 [RFC6515] Aggarwal, R., and E. Rosen, "IPv4 and IPv6 Infrastructure 925 Addresses in BGP Updates for Multicast VPN", RFC 6515, February 2012. 927 9.2. Informative References 929 [ADD-PATH] "Advertisement of Multiple Paths in BGP", D. Walton, A. 930 Retana, E. Chen, J. Scudder, draft-ietf-idr-add-paths-09.txt, October 931 2013. 933 [RFC6368] Marques, P., Raszuk, R., Patel, K., Kumaki, K., and T. 934 Yamagata, "Internal BGP as the Provider/Customer Edge Protocol for 935 BGP/MPLS IP Virtual Private Networks (VPNs)", RFC 6368, September 936 2011. 938 [RFC4601] Fenner, B., Handley, M., Holbrook, H., and I. Kouvelas, 939 Protocol Independent Multicast - Sparse Mode (PIM-SM): Protocol 940 Specification (Revised)", RFC 4601, August 2006. 942 [RFC4684] P. Marques, et. al., "Constrained Route Distribution for 943 Border Gateway Protocol/MultiProtocol Label Switching (BGP/MPLS) 944 Internet Protocol (IP) Virtual Private Networks (VPNs)", RFC 4684, 945 November 2006. 947 [RFC5294] Savola, P. and J. Lingard, "Host Threats to Protocol 948 Independent Multicast (PIM)", RFC 5294, August 2008. 950 [MVPN-extranet] Rekhter, Y. and E. Rosen (editors), "Extranet 951 Multicast in BGP/IP MPLS VPNs", draft-ietf-l3vpn-mvpn- 952 extranet-04.txt, March 2014 954 [SEAMLESS-MCAST] Rekhter, Y., Aggarwal, R., Morin, T., Grosclaude, 955 I., Leymann, N., and S. Saad,"Inter-Area P2MP Segmented LSPs", draft- 956 ietf-mpls-seamless-mcast-13.txt, June 2014