Independent Submission                                     E. Rosen, Ed.
Request for Comments: 6037                                   Y. Cai, Ed.
Category: Historic                                           I. Wijnands
ISSN: 2070-1721                                      Cisco Systems, Inc.
                                                            October 2010

Cisco Systems' Solution for Multicast in BGP/MPLS IP VPNs

Abstract

   This document describes the MVPN (Multicast in BGP/MPLS IP VPNs)
   solution designed and deployed by Cisco Systems.  The procedures
   specified in this document are largely a subset of the generalized
   MVPN framework recently standardized by the IETF.  However, as the
   deployment of the procedures specified herein predates the
   publication of IETF standards (in some cases by over five years), an
   implementation based on these procedures differs in some respects
   from a fully standards-compliant implementation.  These differences
   are pointed out in the document.

Status of This Memo

   This document is not an Internet Standards Track specification; it is
   published for the historical record.

   This document defines a Historic Document for the Internet community.
   This is a contribution to the RFC Series, independently of any other
   RFC stream.  The RFC Editor has chosen to publish this document at
   its discretion and makes no statement about its value for
   implementation or deployment.  Documents approved for publication by
   the RFC Editor are not a candidate for any level of Internet
   Standard; see Section 2 of RFC 5741.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at
   http://www.rfc-editor.org/info/rfc6037.

Copyright Notice

   Copyright (c) 2010 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (http://trustee.ietf.org/license-info) in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.

   1. Introduction ....................................................3
      1.1. Specification of Requirements ..............................3
      1.2. Scaling Multicast State Information in the Network Core ....3
      1.3. Overview ...................................................4
   2. Multicast VRFs ..................................................6
   3. Multicast Domains ...............................................7
      3.1. Model of Operation .........................................7
   4. Multicast Tunnels ...............................................7
      4.1. Ingress PEs ................................................8
      4.2. Egress PEs .................................................8
      4.3. Tunnel Destination Address(es) .............................8
      4.4. Auto-Discovery .............................................8
           4.4.1. MDT-SAFI ...........................................10
      4.5. Which PIM Variant to Use ..................................10
      4.6. Inter-AS MDT Construction .................................11
           4.6.1. The PIM MVPN Join Attribute ........................11
                  4.6.1.1. Definition ................................11
                  4.6.1.2. Usage .....................................12
      4.7. Encapsulation in GRE ......................................13
      4.8. MTU .......................................................14
      4.9. TTL .......................................................14
      4.10. Differentiated Services ..................................14
      4.11. Avoiding Conflict with Internet Multicast ................14
   5. The PIM C-Instance and the MT ..................................15
      5.1. PIM C-Instance Control Packets ............................15
      5.2. PIM C-Instance RPF Determination ..........................15
           5.2.1. Connector Attribute ................................16
   6. Data MDT: Optimizing Flooding ..................................17
      6.1. Limitation of Multicast Domain ............................17
      6.2. Signaling Data MDTs .......................................17
      6.3. Use of SSM for Data MDTs ..................................19
   7. Packet Formats and Constants ...................................20
      7.1. MDT TLV ...................................................20
      7.2. MDT Join TLV for IPv4 Streams .............................20
      7.3. MDT Join TLV for IPv6 Streams .............................21
      7.4. Multiple MDT Join TLVs per Datagram .......................22
      7.5. Constants .................................................22
   8. IANA Considerations ............................................23
   9. Security Considerations ........................................23
   10. Acknowledgments ...............................................23
   11. References ....................................................24
      11.1. Normative References .....................................24
      11.2. Informative References ...................................24

1. Introduction

   This document describes the MVPN (Multicast in BGP/MPLS IP VPNs)
   solution designed and deployed by Cisco Systems.  This document is
   being made available for the record and as a reference for
   interoperating with deployed implementations.  This document is a
   technical specification and should not be used to infer the current
   or future plans of Cisco Systems.

   The procedures specified in this document are largely a subset of the
   generalized MVPN framework defined in [MVPN].  However, as this
   document specifies an implementation that precedes the
   standardization of [MVPN] by several years, it does differ in a few
   respects from a fully standards-compliant implementation.  These
   differences are pointed out where they occur.

   The base specification for BGP/MPLS IP VPNs [RFC4364] does not
   provide a way for IP multicast data or control traffic to travel from
   one VPN site to another.  This document extends that specification by
   specifying the necessary protocols and procedures for support of IP
   multicast.

   This specification presupposes that:

   1. Protocol Independent Multicast (PIM) [PIM-SM], running over either
      IPv4 or IPv6, is the multicast routing protocol used within the
      VPN,

2. PIM, running over IPv4, is the multicast routing protocol used
within the service-provider (SP) network, and

3. the SP network supports native IPv4 multicast forwarding.

Familiarity with the terminology and procedures of [RFC4364] is
presupposed. Familiarity with [PIM-SM] is also presupposed.

1.1. Specification of Requirements

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
   document are to be interpreted as described in [RFC2119].

1.2. Scaling Multicast State Information in the Network Core

   The BGP/MPLS IP VPN service of [RFC4364] provides a VPN with
   "optimal" unicast routing through the SP backbone, in that a packet
   follows the "shortest path" across the backbone, as determined by the
   backbone's own routing algorithm.  This optimal routing is provided

   without requiring the "P routers" (routers in the provider backbone,
   other than the "provider edge" or "PE" routers) to maintain any
   routing information that is specific to a VPN; indeed, the P routers
   do not maintain any per-VPN state at all.

   Unfortunately, optimal multicast routing cannot be provided without
   requiring the P routers to maintain some VPN-specific state
   information.  Optimal multicast routing would require that one or
   more multicast distribution trees be created in the backbone for each
   multicast group that is in use.  If a particular multicast group from
   within a VPN is using source-based distribution trees, optimal
   routing requires that there be one distribution tree for each
   transmitter of that group.  If shared trees are being used, one tree
   for each group is still required.  Each such tree requires state in
   some set of the P routers, with the amount of state being
   proportional to the number of multicast transmitters.  The reason
   there needs to be at least one distribution tree per multicast group
   is that each group may have a different set of receivers; multicast
   routing algorithms generally go to great lengths to ensure that a
   multicast packet will not be sent to a node that is not on the path
   to a receiver.

   Given that an SP generally supports many VPNs, where each VPN may
   have many multicast groups, and each multicast group may have many
   transmitters, it is not scalable to have one or more distribution
   trees for each multicast group.  The SP has no control whatsoever
   over the number of multicast groups and transmitters that exist in
   the VPNs, and it is difficult to place any bound on these numbers.

   In order to have a scalable multicast solution for BGP/MPLS IP VPNs,
   the amount of state maintained by the P routers needs to be
   proportional to something that IS under the control of the SP.  This
   specification describes such a solution.  In this solution, the
   amount of state maintained in the P routers is proportional only to
   the number of VPNs that run over the backbone; the amount of state in
   the P routers is NOT sensitive to the number of multicast groups or
   to the number of multicast transmitters within the VPNs.  To achieve
   this scalability, the optimality of the multicast routes is reduced.
   A PE that is not on the path to any receiver of a particular
   multicast group may still receive multicast packets for that group,
   and if so, will have to discard them.  The SP does, however, have
   control over the tradeoff between optimal routing and scalability.

1.3. Overview

   An SP determines whether a particular VPN is multicast-enabled.  If
   it is, it corresponds to a "Multicast Domain".  A PE that attaches to
   a particular multicast-enabled VPN is said to belong to the

   corresponding Multicast Domain.  For each Multicast Domain, there is
   a default multicast distribution tree ("MDT") through the backbone,
   connecting ALL of the PEs that belong to that Multicast Domain.  A
   given PE may be in as many Multicast Domains as there are VPNs
   attached to that PE.  However, each Multicast Domain has its own MDT.
   The MDTs are created by running PIM in the backbone, and in general
   an MDT also includes P routers on the paths between the PE routers.

   In a departure from the usual multicast tree distribution procedures,
   the Default MDT for a Multicast Domain is constructed automatically
   as the PEs in the domain come up.  Construction of the Default MDT
   does not depend on the existence of multicast traffic in the domain;
   it will exist before any such multicast traffic is seen.  Default
   MDTs correspond to the Multidirectional Inclusive P-Multicast Service
   Interfaces ("MI-PMSIs") of [MVPN].

   In BGP/MPLS IP VPNs, each CE ("Customer Edge", see [RFC4364]) router
   is a unicast routing adjacency of a PE router, but CE routers at
   different sites do NOT become unicast routing adjacencies of each
   other.  This important characteristic is retained for multicast
   routing -- a CE router becomes a PIM adjacency of a PE router, but CE
   routers at different sites do NOT become PIM adjacencies of each
   other.  Multicast packets from within a VPN are received from a CE
   router by an ingress PE router.  The ingress PE encapsulates the
   multicast packets and (initially) forwards them along the Default MDT
   to all the PE routers connected to sites of the given VPN.  Every PE
   router attached to a site of the given VPN thus receives all
   multicast packets from within that VPN.  If a particular PE router is
   not on the path to any receiver of that multicast group, the PE
   simply discards that packet.

   If a large amount of traffic is being sent to a particular multicast
   group, but that group does not have receivers at all the VPN sites,
   it can be wasteful to forward that group's traffic along the Default
   MDT.  Therefore, we also specify a method for establishing individual
   MDTs for specific multicast groups.  We call these "Data MDTs".  A
   Data MDT delivers VPN data traffic for a particular multicast group
   only to those PE routers that are on the path to receivers of that
   multicast group.  Using a Data MDT has the benefit of reducing the
   amount of multicast traffic on the backbone, as well as reducing the
   load on some of the PEs; it has the disadvantage of increasing the
   amount of state that must be maintained by the P routers.  The SP has
   complete control over this tradeoff.  Data MDTs correspond to the
   Selective PMSI ("S-PMSIs") of [MVPN].

   This solution requires the SP to deploy appropriate protocols and
   procedures, but is transparent to the SP's customers.  An enterprise
   that uses PIM-based multicasting in its network can migrate from a

   private network to a BGP/MPLS IP VPN service, while continuing to use
   whatever multicast router configurations it was previously using; no
   changes need be made to CE routers or to other routers at customer
   sites.  For instance, any dynamic Rendezvous Point ("RP")-discovery
   procedures that are already in use may be left in place.

2.  Multicast VRFs

   The notion of a VPN Routing and Forwarding table ("VRF"), defined in
   [RFC4364], is extended to include multicast routing entries as well
   as unicast routing entries.

   Each VRF has its own multicast routing table.  When a multicast data
   or control packet is received from a particular CE device, multicast
   routing is done in the associated VRF.

   Each PE router runs a number of instances of PIM - Sparse Mode
   (PIM-SM), as many as one per VRF.  In each instance of PIM-SM, the PE
   maintains a PIM adjacency with each of the PIM-capable CE routers
   associated with that VRF.  The multicast routing table created by
   each instance is specific to the corresponding VRF.  We will refer to
   these PIM instances as "VPN-specific PIM instances", or "PIM
   C-instances".

   Each PE router also runs a "provider-wide" instance of PIM-SM (a "PIM
   P-instance"), in which it has a PIM adjacency with each of its IGP
   neighbors (i.e., with P routers), but NOT with any CE routers, and
   not with other PE routers (unless they happen to be adjacent in the
   SP's network).  The P routers also run the P-instance of PIM, but do
   NOT run a C-instance.

   In order to help clarify when we are speaking of the PIM P-instance
   and when we are speaking of a PIM C-instance, we will also apply the
   prefixes "P-" and "C-" respectively to control messages, addresses,
   etc.  Thus, a P-Join would be a PIM Join that is processed by the PIM
   P-instance, and a C-Join would be a PIM Join that is processed by a
   C-instance.  A P-group address would be a group address in the SP's
   address space, and a C-group address would be a group address in a
   VPN's address space.

3. Multicast Domains

3.1. Model of Operation

   A Multicast Domain ("MD") is essentially a set of VRFs associated
   with interfaces that can send multicast traffic to each other.  From
   the standpoint of a PIM C-instance, a Multicast Domain is equivalent
   to a multi-access interface.  The PE routers in a given MD become PIM
   adjacencies of each other in the PIM C-instance.

   Each multicast VRF is assigned to one MD.  Each MD is configured with
   a distinct, multicast P-group address, called the "Default MDT group
   address".  This address is used to build the Default MDT for the MD.

   When a PE router needs to send PIM C-instance control traffic to the
   other PE routers in the MD, it encapsulates the control traffic, with
   its own IPv4 address as the source IP address and the Default MDT
   group address as the destination IP address.  Note that the Default
   MDT is part of the PIM P-instance, whereas the PEs that communicate
   over the Default MDT are PIM adjacencies in a C-instance.  Within the
   C-instance, the Default MDT appears to be a multi-access network to
   which all the PEs are attached.  This is discussed in more detail in
   Section 4.

   The Default MDT does not only carry the PIM control traffic of the
   MD's PIM C-instance.  It also, by default, carries the multicast data
   traffic of the C-instance.  In some cases, though, multicast data
   traffic in a particular MD will be sent on a Data MDT rather than on
   the Default MDT.  The use of Data MDTs is described in Section 6.

   Note that, if an MDT (Default or Data) is set up using the ASM ("Any-
   Source Multicast") Service Model, the MDT (Default or Data) must have
   a P-group address that is "globally unique" (more precisely, unique
   over the set of SP networks carrying the multicast traffic of the
   corresponding MD).  If the MDT is set up using the SSM ("Source-
   Specific Multicast") model, the P-group address of an MDT only needs
   to be unique relative to the source of the MDT (however, see
   Section 4.4).  Nevertheless, some implementations require the same
   SSM group address to be assigned to all the PEs.  Interoperability
   with those implementations requires conformance to this restriction.

4. Multicast Tunnels

   An MD can be thought of as a set of PE routers connected by a
   multicast tunnel ("MT").  From the perspective of a VPN-specific PIM
   instance, an MT is a single multi-access interface.  In the SP
   network, a single MT is realized as a Default MDT combined with zero
   or more Data MDTs.

4.1. Ingress PEs

   An ingress PE is a PE router that is either directly connected to the
   multicast sender in the VPN, or via a CE router.  When the multicast
   sender starts transmitting, and if there are receivers (or a PIM RP)
   behind other PE routers in the common MD, the ingress PE becomes the
   transmitter of either the Default MDT group or a Data MDT group in
   the SP network.

4.2. Egress PEs

   A PE router with a VRF configured in an MD becomes a receiver of the
   Default MDT group for that MD.  A PE router may also join a Data MDT
   group if it has a VPN-specific PIM instance in which it is forwarding
   to one of its attached sites traffic for a particular C-group, and
   that particular C-group has been associated with that particular Data
   MDT.  When a PE router joins any P-group used for encapsulating VPN
   multicast traffic, the PE router becomes one of the endpoints of the
   corresponding MT.

   When a packet is received from an MT, the receiving PE derives the MD
   from the destination address, which is a P-group address, of the
   received packet.  The packet is then passed to the corresponding
   multicast VRF and VPN-specific PIM instance for further processing.

4.3. Tunnel Destination Address(es)

   An MT is an IP tunnel for which the destination address is a P-group
   address.  However, an MT is not limited to using only one P-group
   address for encapsulation.  Based on the payload VPN multicast
   traffic, it can choose to use the Default MDT group address, or one
   of the Data MDT group addresses (as described in Section 6 of this
   document), allowing the MT to reach a different set of PE routers in
   the common MD.

4.4. Auto-Discovery

   Any of the variants of PIM may be used to set up the Default MDT:
   PIM-SM, Bidirectional PIM [BIDIR], or PIM-Source-Specific Multicast
   (PIM-SSM) [SSM].  Except in the case of PIM-SSM, the PEs need only
   know the proper P-group address in order to begin setting up the
   Default MDTs.  The PEs will then discover each others' addresses by
   virtue of receiving PIM control traffic, e.g., PIM Hellos, sourced
   (and encapsulated) by each other.

   However, in the case of PIM-SSM, the necessary MDTs for an MD cannot
   be set up until each PE in the MD knows the source address of each of
   the other PEs in that same MD.  This information needs to be auto-
   discovered.

   A new BGP address family, the MDT-Subsequent Address Family
   Identifier ("MDT-SAFI"), is defined.  The Network Layer Reachability
   Information (NLRI) for this address family consists of a Route
   Distinguisher (RD), an IPv4 unicast address, and a multicast group
   address.  A given PE router in a given MD constructs an NLRI in this
   family from:

   -  Its own IPv4 address.  If it has several, it uses the one that it
      will be placing in the IP Source Address field of multicast
      packets that it will be sending over the MDT.

   -  An RD that has been assigned to the MD.

   -  The P-group address, an IPv4 multicast address that is to be used
      as the IP Destination Address field of multicast packets that will
      be sent over the MDT.

   When a PE distributes this NLRI via BGP, it may include a Route
   Target (RT) Extended Communities attribute.  This RT must be an
   "Import RT" [RFC4364] of each VRF in the MD.  The ordinary BGP
   distribution procedures used by [RFC4364] will then ensure that each
   PE learns the MDT-SAFI "address" of each of the other PEs in the MD,
   and that the learned MDT-SAFI addresses get associated with the right
   VRFs.

   If a PE receives an MDT-SAFI NLRI that does not have an RT attribute,
   the P-group address from the NLRI has to be used to associate the
   NLRI with a particular VRF.  In this case, each Multicast Domain must
   be associated with a unique P-address, even if PIM-SSM is used.
   However, finding a unique P-address for a multi-provider multicast
   group may be difficult.

   In order to facilitate the deployment of multi-provider Multicast
   Domains, this specification REQUIRES the use of the MDT-SAFI NLRI
   (even if PIM-SSM is not used to set up the Default MDT).  This
   specification also REQUIRES that an implementation be capable of
   using PIM-SSM to set up the Default MDT.

   In [MVPN], the MDT-SAFI is replaced by the Intra-Autonomous-System
   Inclusive-PMSI auto-discovery ("Intra-AS I-PMSI A-D") route.  The
   latter is a generalized version of the MDT-SAFI, which allows the
   "Default MDTs" and "Data MDTs" to be implemented as MPLS P2MP LSPs
   ("Point-to-Multipoint Label Switched Paths") or MP2MP LSPs

   ("Multipoint-to-Multipoint Label Switched Paths"), as well as by
   PIM-created multicast distribution trees.  In the latter case, the
   Intra-AS A-D routes carry the same information that the MDT-SAFI
   does, though with a different encoding.

   The Intra-AS A-D routes also carry Route Targets, and so may be
   distributed in the same manner as unicast routes, including being
   distributed inter-AS.  (Despite their name, the inter-AS distribution
   of Intra-AS I-PMSI A-D routes is sometimes necessary in [MVPN].)

   The encoding of the MDT-SAFI is specified in the following
   subsection.

4.4.1. MDT-SAFI

   BGP messages in which AFI=1 and SAFI=66 are "MDT-SAFI" messages.

   The NLRI format is the 8-byte-RD:IPv4-address followed by the MDT
   group address, i.e., the MP_REACH attribute for this SAFI will
   contain one or more tuples of the following form:

          +-------------------------------+
          |                               |
          |  RD:IPv4-address (12 octets)  |
          |                               |
          +-------------------------------+
          |    Group Address (4 octets)   |
          +-------------------------------+

   The IPv4 address identifies the PE that originated this route, and
   the RD identifies a VRF in that PE.  The group address MUST be an
   IPv4 multicast group address and is used to build the P-tunnels.  All
   PEs attached to a given MVPN MUST specify the same group address,
   even if the group is an SSM group.  MDT-SAFI routes do not carry RTs,
   and the group address is used to associate a received MDT-SAFI route
   with a VRF.

4.5. Which PIM Variant to Use

   To minimize the amount of multicast routing state maintained by the P
   routers, the Default MDTs should be realized as shared trees, such as
   PIM bidirectional trees.  However, the operational procedures for
   assigning P-group addresses may be greatly simplified, especially in
   the case of multi-provider MDs, if PIM-SSM is used.

   Data MDTs are best realized as source trees, constructed via PIM-SSM.

4.6. Inter-AS MDT Construction

   Standard PIM techniques for the construction of source trees
   presuppose that every router has a route to the source of the tree.
   However, if the source of the tree is in a different AS than a
   particular P router, it is possible that the P router will not have a
   route to the source.  For example, the remote AS may be using BGP to
   distribute a route to the source, but a particular P router may be
   part of a "BGP-free core", in which the P routers are not aware of
   BGP-distributed routes.

   What is needed in this case is a way for a PE to tell PIM to
   construct the tree through a particular BGP speaker, the "BGP Next
   Hop" for the tree source.  This can be accomplished with a PIM
   extension.

   If the PE has selected the source of the tree from the MDT SAFI
   address family, then it may be desirable to build the tree along the
   route to the MDT SAFI address, rather than along the route to the
   corresponding IPv4 address.  This enables the inter-AS portion of the
   tree to follow a path that is specifically chosen for multicast
   (i.e., it allows the inter-AS multicast topology to be
   "non-congruent" to the inter-AS unicast topology).  This too requires
   a PIM extension.

   The necessary PIM extension is the PIM MVPN Join Attribute described
   in the following subsection.

4.6.1. The PIM MVPN Join Attribute

4.6.1.1. Definition

   In [PIM-ATTRIB], the notion of a "Join Attribute" is defined, and a
   format for included Join Attributes in PIM Join/Prune messages is
   specified.  We now define a new Join Attribute, which we call the
   "MVPN Join Attribute".

   0                   1                   2                   3
   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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |F|E|   Type    | Length        |     Proxy IP address
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                                   |      RD
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-.......

   The 6-bit Type field of the MVPN Join Attribute is set to 1.

   The F-bit is set to 0, indicating that the attribute is
   non-transitive.

   Rules for setting the E-bit are given in [PIM-ATTRIB].

   Two information fields are carried in the MVPN Join Attribute:

   -  Proxy IP address: The IP address of the node towards which the PIM
      Join/Prune message is to be forwarded.  This will either be an
      IPv4 or an IPv6 address, depending on whether the PIM Join/Prune
      message itself is IPv4 or IPv6.

   -  RD: An eight-byte RD.  This immediately follows the proxy IP
      address.

   The PIM message also carries the address of the upstream PE.

   In the case of an intra-AS MVPN, the proxy and the upstream PE are
   the same.  In the case of an inter-AS MVPN, the proxy will be the AS
   Border Router (ASBR) that is the exit point from the local AS on the
   path to the upstream PE.

4.6.1.2. Usage

   When a PE router creates a PIM Join/Prune message in order to set up
   an inter-AS Default MDT, it does so as a result of having received a
   particular MDT-SAFI route.  It includes an MVPN Join Attribute whose
   fields are set as follows:

   -  If the upstream PE is in the same AS as the local PE, then the
      Proxy field contains the address of the upstream PE.  Otherwise,
      it contains the address of the BGP Next Hop on the route to the
      upstream PE.

   -  The RD field contains the RD from the NLRI of the MDT-SAFI route.

   -  The Upstream PE field contains the address of the PE that
      originated the MDT-SAFI route (obtained from the NLRI of that
      route).

   When a PIM router processes a PIM Join/Prune message with an MVPN
   Join Attribute, it first checks to see if the Proxy field contains
   one of its own addresses.

   If not, the router uses the proxy IP address in order to determine
   the Reverse Path Forwarding (RPF) interface and neighbor.  The MVPN
   Join Attribute MUST be passed upstream, unchanged.

   If the proxy address is one of the router's own IP addresses, then
   the router looks in its BGP routing table for an MDT-SAFI route whose
   NLRI consists of the upstream PE address prepended with the RD from
   the Join Attribute.  If there is no match, the PIM message is
   discarded.  If there is a match, the IP address from the BGP Next Hop
   field of the matching route is used in order to determine the RPF
   interface and neighbor.  When the PIM Join/Prune is forwarded
   upstream, the Proxy field is replaced with the address of the BGP
   Next Hop, and the RD and Upstream PE fields are left unchanged.

4.7. Encapsulation in GRE

   Generic Routing Encapsulation (GRE) [GRE1701] is used when sending
   multicast traffic through an MDT.  The following diagram shows the
   progression of the packet as it enters and leaves the service-
   provider network.

   Packets received        Packets in transit      Packets forwarded
   at ingress PE           in the service-         by egress PEs
                           provider network

                           +---------------+
                           |  P-IP Header  |
                           +---------------+
                           |      GRE      |
   ++=============++       ++=============++       ++=============++
   || C-IP Header ||       || C-IP Header ||       || C-IP Header ||
   ++=============++ >>>>> ++=============++ >>>>> ++=============++
   || C-Payload   ||       || C-Payload   ||       || C-Payload   ||
   ++=============++       ++=============++       ++=============++

   The IPv4 Protocol Number field in the P-IP Header MUST be set to 47.
   The Protocol Type field of the GRE Header MUST be set to 0x0800 if
   the C-IP header is an IPv4 header; it MUST be set to 0x86dd if the
   C-IP header is an IPv6 header.

   [GRE2784] specifies an optional GRE checksum, and [GRE2890] specifies
   optional GRE Key and Sequence Number fields.

   The GRE Key field is not needed because the P-group address in the
   delivery IP header already identifies the MD, and thus associates the
   VRF context, for the payload packet to be further processed.

   The GRE Sequence Number field is also not needed because the
   transport layer services for the original application will be
   provided by the C-IP Header.

   The use of the GRE Checksum field MUST follow [GRE2784].

   To facilitate high-speed implementation, this document recommends
   that the ingress PE routers encapsulate VPN packets without setting
   the Checksum, Key, or Sequence Number field.

4.8. MTU

   Because multicast group addresses are used as tunnel destination
   addresses, existing Path MTU discovery mechanisms cannot be used.
   This requires that:

   1. The ingress PE router (one that does the encapsulation) MUST NOT
      set the DF ("Don't Fragment") bit in the outer header, and

   2. If the "DF" bit is cleared in the IP header of the C-Packet,
      fragment the C-Packet before encapsulation if appropriate.  This
      is very important in practice due to the fact that the performance
      of the reassembly function is significantly lower than that of
      decapsulating and forwarding packets on today's router
      implementations.

4.9. TTL

   The ingress PE should not copy the Time to Live (TTL) field from the
   payload IP header received from a CE router to the delivery IP
   header.  Setting the TTL of the delivery IP header is determined by
   the local policy of the ingress PE router.

4.10. Differentiated Services

   By default, setting of the DS ("Differentiated Services") field in
   the delivery IP header should follow the guidelines outlined in
   [DIFF2983].  An SP may also choose to deploy any of the additional
   mechanisms the PE routers support.

4.11. Avoiding Conflict with Internet Multicast

   If the SP is providing Internet multicast, distinct from its VPN
   multicast services, it must ensure that the P-group addresses that
   correspond to its MDs are distinct from any of the group addresses of
   the Internet multicasts it supports.  This is best done by using
   administratively scoped addresses [ADMIN-ADDR].

   The C-group addresses need not be distinct from either the P-group
   addresses or the Internet multicast addresses.

5. The PIM C-Instance and the MT

   If a particular VRF is in a particular MD, the corresponding MT is
   treated by that VRF's VPN-specific PIM instances as a LAN interface.
   As a result, the PEs that are adjacent on the MT will generate and
   process PIM control packets, such as Hello, Join/Prune, and Assert.
   Designated Forwarder election occurs just as it would on an actual
   LAN interface.

5.1. PIM C-Instance Control Packets

   The PIM protocol packets are sent to ALL-PIM-ROUTERS (224.0.0.13 for
   IPv4 or ff02::d for IPv6) in the context of that VRF, but when in
   transit in the provider network, they are encapsulated using the
   Default MDT group configured for that MD.  This allows VPN-specific
   PIM routes to be extended from site to site without appearing in the
   P routers.

   If a PIM C-Instance control packet is an IPv6 packet, its source
   address is the IPv4-mapped IPv6 address corresponding to the IPv4
   address of the PE router sending the packet.

5.2. PIM C-Instance RPF Determination

   Although the MT is treated as a PIM-enabled interface, unicast
   routing is NOT run over it, and there are no unicast routing
   adjacencies over it.  It is therefore necessary to specify special
   procedures for determining when the MT is to be regarded as the "RPF
   Interface" for a particular C-address.

   When a PE needs to determine the RPF interface of a particular
   C-address, it looks up the C-address in the VRF.  If the route
   matching it is not a VPN-IP route learned from MP-BGP as described in
   [RFC4364], or if that route's outgoing interface is one of the
   interfaces associated with the VRF, then ordinary PIM procedures for
   determining the RPF interface apply.

   However, if the route matching the C-address is a VPN-IP route whose
   outgoing interface is not one of the interfaces associated with the
   VRF, then PIM will consider the outgoing interface to be the MT
   associated with the VPN-specific PIM instance.

   Once PIM has determined that the RPF interface for a particular
   C-address is the MT, it is necessary for PIM to determine the RPF
   neighbor for that C-address.  This will be one of the other PEs that
   is a PIM adjacency over the MT.

   The BGP "Connector" Attribute is defined.  Whenever a PE router
   distributes a VPN-IP address from a VRF that is part of an MD, it
   SHOULD distribute a Connector Attribute along with it.  The Connector
   Attribute specifies the MDT address family, and its value is the IP
   address that the PE router is using as its source IP address for the
   multicast packets that are encapsulated and sent over the MT.  When a
   PE has determined that the RPF interface for a particular C-address
   is the MT, it looks up the Connector Attribute that was distributed
   along with the VPN-IP address corresponding to that C-address.  The
   value of this Connector Attribute is considered to be the RPF
   adjacency for the C-address.

   There are older implementations in which the Connector Attribute is
   not present.  In this case, as long as the "BGP Next Hop" for the
   C-address is one of the PEs that is a PIM adjacency, then that PE is
   treated as the RPF adjacency for that C-address.

   However, if the MD spans multiple Autonomous Systems, and an
   "option b" interconnect ([RFC4364], Section 10) is used, the BGP Next
   Hop might not be a PIM adjacency, and the RPF check will not succeed
   unless the Connector Attribute is used.

   In [MVPN], the Connector Attribute is replaced by the "VRF Route
   Import Extended Community" attribute.  The latter is a generalized
   version, but carries the same information as the Connector Attribute
   does; the encoding, however, is different.

   The Connector Attribute is defined in the following subsection.

5.2.1. Connector Attribute

   The Connector Attribute is an optional transitive attribute.  Its
   value field is formatted as follows:

           0                   1
           0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
          +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
          |0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1|
          +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
          |                               |
          |  IPv4 Address of PE           |
          |                               |
          +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

6. Data MDT: Optimizing Flooding

6.1. Limitation of Multicast Domain

   While the procedure specified in the previous section requires the P
   routers to maintain multicast state, the amount of state is bounded
   by the number of supported VPNs.  The P routers do NOT run any VPN-
   specific PIM instances.

   In particular, the use of a single bidirectional tree per VPN scales
   well as the number of transmitters and receivers increases, but not
   so well as the amount of multicast traffic per VPN increases.

   The multicast routing provided by this scheme is not optimal, in that
   a packet of a particular multicast group may be forwarded to PE
   routers that have no downstream receivers for that group, and which
   hence may need to discard the packet.

   In the simplest configuration model, only the Default MDT group is
   configured for each MD.  The result of the configuration is that all
   VPN multicast traffic, whether control or data, will be encapsulated
   and forwarded to all PE routers that are part of the MD.  While this
   limits the number of multicast routing states the provider network
   has to maintain, it also requires PE routers to discard multicast
   C-packets if there are no receivers for those packets in the
   corresponding sites.  In some cases, especially when the content
   involves high bandwidth but only a limited set of receivers, it is
   desirable that certain C-packets only travel to PE routers that do
   have receivers in the VPN to save bandwidth in the network and reduce
   load on the PE routers.

6.2. Signaling Data MDTs

   A simple protocol is proposed to signal additional P-group addresses
   to encapsulate VPN traffic.  These P-group addresses are called Data
   MDT groups.  The ingress PE router advertises a different P-group
   address (as opposed to always using the Default MDT group) to
   encapsulate VPN multicast traffic.  Only the PE routers on the path
   to eventual receivers join the P-group, and therefore form an optimal
   multicast distribution tree in the service-provider network for the
   VPN multicast traffic.  These multicast distribution trees are called
   Data MDTs because they do not carry PIM control packets exchanged by
   PE routers.

   The following text documents the procedures of the initiation and
   teardown of the Data MDTs.  The definition of the constants and
   timers can be found in Section 7.

   -  The PE router connected to the source of the content initially
      uses the Default MDT group when forwarding the content to the MD.

   -  When one or more pre-configured conditions are met, it starts to
      periodically announce the MDT Join TLV at the interval of
      [MDT_INTERVAL].  The MDT Join TLV is forwarded to all the PE
      routers in the MD.

      A commonly used condition is the bandwidth.  When the VPN traffic
      exceeds a certain threshold, it is more desirable to deliver the
      flow to the PE routers connected to receivers in order to optimize
      the performance of PE routers and the resources of the provider
      network.  However, other conditions can also be devised, and they
      are purely implementation specific.

   -  The MDT Join TLV is encapsulated in UDP.

      UDP over IPv4 is used if the multicast stream being assigned to a
      Data MDT is an IPv4 stream.  In this case, the UDP datagram is
      addressed to ALL-PIM-ROUTERS (224.0.0.13).

      UDP over IPv6 is used if the multicast stream being assigned to a
      Data MDT is an IPv6 stream.  In this case, the UDP datagram is
      addressed to ALL-PIM-ROUTERS (ff02::d).

      The destination UDP port is 3232.

      The UDP datagram is sent on the Default MDT.  This allows all PE
      routers to receive the information.  Any MDT Join that is not
      received over a Default MDT MUST be dropped.

   -  Upon receiving an MDT Join TLV, PE routers connected to receivers
      will join the Data MDT group announced by the MDT Join TLV in the
      global table.  When the Data MDT group is in PIM-SM or
      bidirectional PIM mode, the PE routers build a shared tree toward
      the RP.  When the Data MDT group is set up using PIM-SSM, the PE
      routers build a source tree toward the PE router that is
      advertising the MDT Join TLV.  The IP address of that PE router is
      learned from the IP Source Address field of the UDP packet that
      contains the MDT Join TLV.

      PE routers that are not connected to receivers may wish to cache
      the states in order to reduce the delay when a receiver comes up
      in the future.

   -  After [MDT_DATA_DELAY], the PE router connected to the source
      starts encapsulating traffic using the Data MDT group.

   -  When the pre-configured conditions are no longer met, e.g., the
      traffic stops, the PE router connected to the source stops
      announcing the MDT Join TLV.

   -  If the MDT Join TLV is not received for an interval longer than
      [MDT_DATA_TIMEOUT], PE routers connected to the receivers just
      leave the Data MDT group in the global instance.

6.3. Use of SSM for Data MDTs

   The use of Data MDTs requires that a set of multicast P-addresses be
   pre-allocated and dedicated for use as the destination addresses for
   the Data MDTs.

   If SSM is used to set up the Data MDTs, then each MD needs to be
   assigned a set of these multicast P-addresses.  Each VRF in the MD
   needs to be configured with this same set of multicast P-addresses.
   If there are n addresses in this set, then each PE in the MD can be
   the source of n Data MDTs in that MD.

   If SSM is not used for setting up Data MDTs, then each VRF needs to
   be configured with a unique set of multicast P-addresses; two VRFs in
   the same MD cannot be configured with the same set of addresses.
   This requires the pre-allocation of many more multicast P-addresses,
   and the need to configure a different set for each VRF greatly
   complicates the operations and management.  Therefore, the use of SSM
   for Data MDTs is very strongly recommended.

7. Packet Formats and Constants

7.1. MDT TLV

   The MDT TLV has the following format.

       0                   1                   2                   3
       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
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |     Type      |            Length           |     Value       |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                               .                               |
      |                               .                               |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      Type (8 bits):

         the type of the MDT TLV.  In this specification,
         types 1 and 4 are defined.

      Length (16 bits):

         the total number of octets in the TLV for this type,
         including both the Type and Length fields.

      Value (variable length):

         the content of the TLV.

7.2. MDT Join TLV for IPv4 Streams

   The MDT Join TLV for IPv4 streams has the following format.

       0                   1                   2                   3
       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
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |     Type      |           Length            |    Reserved     |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                           C-source                            |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                           C-group                             |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                           P-group                             |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      Type (8 bits):

         Must be set to 1.

      Length (16 bits):

         Must be set to 16.

      Reserved (8 bits):

         for future use.

      C-source (32 bits):

         the IPv4 address of the traffic source in the VPN.

      C-group (32 bits):

         the IPv4 address of the multicast traffic destination address
         in the VPN.

      P-group (32 bits):

         the IPv4 group address that the PE router is going to use to
         encapsulate the flow (C-source, C-group).

7.3. MDT Join TLV for IPv6 Streams

   The MDT Join TLV for IPv6 streams has the following format.

       0                   1                   2                   3
       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
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |     Type      |           Length            |    Reserved     |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                                                               |
      |                           C-source                            |
      |                                                               |
      |                                                               |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                                                               |
      |                           C-group                             |
      |                                                               |
      |                                                               |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                           P-group                             |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      Type (8 bits):

         Must be set to 4.

      Length (16 bits):

         Must be set to 40.

      Reserved (8 bits):

         for future use.

      C-source (128 bits):

         the IPv6 address of the traffic source in the VPN.

      C-group (128 bits):

         the IPv6 address of the multicast traffic destination
         address in the VPN.

      P-group (32 bits):

         the IPv4 group address that the PE router is going to use
         to encapsulate the flow (C-source, C-group).

7.4. Multiple MDT Join TLVs per Datagram

   A single UDP datagram MAY carry multiple MDT Join TLVs, as many as
   can fit entirely within it.  If there are multiple MDT Join TLVs in a
   UDP datagram, they MUST be of the same type.  The end of the last MDT
   Join TLV (as determined by the MDT Join TLV Length field) MUST
   coincide with the end of the UDP datagram, as determined by the UDP
   Length field.  When processing a received UDP datagram that contains
   one or more MDT Join TLVs, a router MUST be able to process all the
   MDT Join TLVs that fit into the datagram.

7.5. Constants

   [MDT_DATA_DELAY]:

      the interval before the PE router connected to the source will
      switch to the Data MDT group.  The default value is 3 seconds.

   [MDT_DATA_TIMEOUT]:

      the interval before which the PE router connected to the receivers
      will time out and leave the Data MDT group if no MDT_JOIN_TLV
      message has been received.  The default value is 3 minutes.  This
      value must be consistent among PE routers.

   [MDT_DATA_HOLDDOWN]:

      the interval before which the PE router will switch back to the
      Default MDT after it started encapsulating packets using the Data
      MDT group.  This is used to avoid oscillation when traffic is
      bursty.  The default value is 1 minute.

   [MDT_INTERVAL]:

      the interval the source PE router uses to periodically send
      MDT_JOIN_TLV messages.  The default value is 60 seconds.

8. IANA Considerations

   The codepoint for the Connector Attribute is defined in IANA's
   registry of BGP attributes.  The reference has been updated to refer
   to this document.  On the IANA web page, the codepoint is denoted as
   "deprecated".  This document does not change that status.  However,
   note that there are a large number of deployments using this
   codepoint, and this is likely to be the case for a number of years.

   The codepoint for MDT-SAFI is defined in IANA's registry of BGP SAFI
   assignments.  The reference has been updated to refer to this
   document.

9. Security Considerations

   [RFC4364] discusses in general the security considerations that
   pertain to when the RFC 4364 type of VPN is deployed.

   [PIM-SM] discusses the security considerations that pertain to the
   use of PIM.

   The security considerations of [RFC4023] and [RFC4797] apply whenever
   VPN traffic is carried through IP or GRE tunnels.

10. Acknowledgments

   Major contributions to this work have been made by Dan Tappan and
   Tony Speakman.

   The authors also wish to thank Arjen Boers, Robert Raszuk, Toerless
   Eckert, and Ted Qian for their help and their ideas.

11. References

11.1. Normative References

   [GRE2784]      Farinacci, D., Li, T., Hanks, S., Meyer, D., and P.
                  Traina, "Generic Routing Encapsulation (GRE)",
                  RFC 2784, March 2000.

   [PIM-SM]       Fenner, B., Handley, M., Holbrook, H., and I.
                  Kouvelas, "Protocol Independent Multicast - Sparse
                  Mode (PIM-SM): Protocol Specification (Revised)",
                  RFC 4601, August 2006.

   [PIM-ATTRIB]   Boers, A., Wijnands, I., and E. Rosen, "The Protocol
                  Independent Multicast (PIM) Join Attribute Format",
                  RFC 5384, November 2008.

   [RFC2119]      Bradner, S., "Key words for use in RFCs to Indicate
                  Requirement Levels", BCP 14, RFC 2119, March 1997.

   [RFC4364]      Rosen, E. and Y. Rekhter, "BGP/MPLS IP Virtual Private
                  Networks (VPNs)", RFC 4364, February 2006.

11.2. Informative References

   [ADMIN-ADDR]   Meyer, D., "Administratively Scoped IP Multicast",
                  BCP 23, RFC 2365, July 1998.

   [BIDIR]        Handley, M., Kouvelas, I., Speakman, T., and L.
                  Vicisano, "Bidirectional Protocol Independent
                  Multicast (BIDIR-PIM)", RFC 5015, October 2007.

   [DIFF2983]     Black, D., "Differentiated Services and Tunnels",
                  RFC 2983, October 2000.

   [GRE1701]      Hanks, S., Li, T., Farinacci, D., and P. Traina,
                  "Generic Routing Encapsulation (GRE)", RFC 1701,
                  October 1994.

   [GRE2890]      Dommety, G., "Key and Sequence Number Extensions to
                  GRE", RFC 2890, September 2000.

   [MVPN]         Rosen, E., Ed., and R. Aggarwal, Ed., "Multicast in
                  MPLS/BGP IP VPNs", Work in Progress, January 2010.

   [SSM]          Holbrook, H. and B. Cain, "Source-Specific Multicast
                  for IP", RFC 4607, August 2006.

   [RFC4023]      Worster, T., Rekhter, Y., and E. Rosen, Ed.,
                  "Encapsulating MPLS in IP or Generic Routing
                  Encapsulation (GRE)", RFC 4023, March 2005.

   [RFC4797]      Rekhter, Y., Bonica, R., and E. Rosen, "Use of
                  Provider Edge to Provider Edge (PE-PE) Generic Routing
                  Encapsulation (GRE) or IP in BGP/MPLS IP Virtual
                  Private Networks", RFC 4797, January 2007.

Authors' Addresses

   Eric C. Rosen (editor)
   Cisco Systems, Inc.
   1414 Massachusetts Avenue
   Boxborough, MA  01719
   EMail: erosen@cisco.com

   Yiqun Cai (editor)
   Cisco Systems, Inc.
   170 Tasman Drive
   San Jose, CA  95134
   EMail: ycai@cisco.com

   IJsbrand Wijnands
   Cisco Systems, Inc.
   170 Tasman Drive
   San Jose, CA  95134
   EMail: ice@cisco.com