Internet Engineering Task Force (IETF) IJ. Wijnands, Ed. Request for Comments: 6826 T. Eckert Category: Standards Track Cisco Systems, Inc. ISSN: 2070-1721 N. Leymann Deutsche Telekom M. Napierala AT&T Labs January 2013
Multipoint LDP In-Band Signaling for Point-to-Multipoint and Multipoint-to-Multipoint Label Switched Paths
Consider an IP multicast tree, constructed by Protocol Independent Multicast (PIM), that needs to pass through an MPLS domain in which Multipoint LDP (mLDP) point-to-multipoint and/or multipoint-to- multipoint Labels Switched Paths (LSPs) can be created. The part of the IP multicast tree that traverses the MPLS domain can be instantiated as a multipoint LSP. When a PIM Join message is received at the border of the MPLS domain, information from that message is encoded into mLDP messages. When the mLDP messages reach the border of the next IP domain, the encoded information is used to generate PIM messages that can be sent through the IP domain. The result is an IP multicast tree consisting of a set of IP multicast sub-trees that are spliced together with a multipoint LSP. This document describes procedures regarding how IP multicast trees are spliced together with multipoint LSPs.
Status of This Memo
This is an Internet Standards Track document.
This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Further information on Internet Standards is available in Section 2 of RFC 5741.
Copyright (c) 2013 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. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
The mLDP (Multipoint LDP) [RFC6388] specification describes mechanisms for creating point-to-multipoint (P2MP) and multipoint-to- multipoint (MP2MP) LSPs (Label Switched Paths). These LSPs are typically used for transporting end-user multicast packets. However, the mLDP specification does not provide any rules for associating particular end-user multicast packets with any particular LSP. Other documents, like [RFC6513], describe applications in which out-of-band signaling protocols, such as PIM and BGP, are used to establish the mapping between an LSP and the multicast packets that need to be forwarded over the LSP.
This document describes an application in which the information needed to establish the mapping between an LSP and the set of multicast packets to be forwarded over it is carried in the "opaque value" field of an mLDP FEC (Forwarding Equivalence Class) element. When an IP multicast tree (either a source-specific tree or a bidirectional tree) enters the MPLS network, the (S,G) or (*,G) information from the IP multicast control-plane state is carried in the opaque value field of the mLDP FEC message. As the tree leaves the MPLS network, this information is extracted from the FEC Element and used to build the IP multicast control plane. PIM messages can be sent outside the MPLS domain. Note that although the PIM control messages are sent periodically, the mLDP messages are not.
Each IP multicast tree is mapped one-to-one to a P2MP or MP2MP LSP in the MPLS network. A network operator should expect to see as many LSPs in the MPLS network as there are IP multicast trees. A network operator should be aware how IP multicast state is created in the network to ensure that it does not exceed the scalability numbers of the protocol, either PIM or mLDP.
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 RFC 2119 [RFC2119].
Nodes A and B are IP-multicast-capable routers and connect to a Source/RP and a Receiver, respectively. Nodes U, C, and D are MPLS Label Switched Routers (LSRs).
LSR D is attached to a network that is capable of MPLS multicast and IP multicast (see figure 1), and D is required to create a IP multicast tree due to a certain IP multicast event, like a PIM Join, MSDP Source Announcement (SA) [RFC3618], BGP Source Active auto- discovery route [SM-MLDP], or Rendezvous Point (RP) discovery. Suppose that D can determine that the IP multicast tree needs to travel through the MPLS network until it reaches LSR U. For instance, when D looks up the route to the Source or RP [RFC4601] of the IP multicast tree, it may discover that the route is a BGP route with U as the BGP next hop. Then D may choose to set up a P2MP or an MP2MP LSP, with U as root, and to make that LSP become part of the IP multicast distribution tree. Note that other methods are possible to determine that an IP multicast tree is to be transported across an MPLS network using P2MP or MP2MP LSPs. However, these methods are outside the scope of this document.
In order to establish a multicast tree via a P2MP or MP2MP LSP using "in-band signaling", LSR D encodes a P2MP or MP2MP FEC Element, with the IP address of LSR U as the "Root Node Address" and with the source and the group encoded into the "opaque value" ([RFC6388], Sections 2.2 and 3.2). Several different opaque value types are defined in this document. LSR D MUST NOT use a particular opaque value type unless it knows (through provisioning or through some other means outside the scope of this document) that LSR U supports the root node procedures for that opaque value type.
The particular type of FEC Element and opaque value used depends on the IP address family being used, and on whether the multicast tree being established is a source-specific or a bidirectional multicast tree.
When an LSR receives a label mapping or withdraw whose FEC Element contains one of the opaque value types defined in this document, and that LSR is not the one identified by the "Root Node Address" field of that FEC Element, the LSR follows the procedures provided in RFC 6388.
When an LSR receives a label mapping or withdraw whose FEC Element contains one of the opaque value types defined in this document, and that LSR is the one identified by the Root Node Address field of that FEC Element, then the following procedure is executed. The multicast source and group are extracted and passed to the multicast code. If a label mapping is being processed, the multicast code will add the downstream LDP neighbor to the olist of the corresponding (S,G) or
Wijnands, et al. Standards Track [Page 5]
RFC 6826 In-Band Signaling with mLDP January 2013
(*,G) state, creating such state if it does not already exist. If a label withdraw is being processed, the multicast code will remove the downstream LDP neighbor from the olist of the corresponding (S,G) or (*,G) state. From this point on, normal PIM processing will occur.
Note that if the LSR identified by the Root Node Address field does not recognize the opaque value type, the MP LSP will be established, but the root node will not send any multicast data packets on it.
Source or RP addresses that are reachable in a VPN context are outside the scope of this document.
Multicast groups that operate in PIM Dense-Mode are outside the scope of this document.
2.1. Transiting Unidirectional IP Multicast Shared Trees
Nothing prevents PIM shared trees, used by PIM-SM in the ASM service model, from being transported across an MPLS core. However, it is not possible to prune individual sources from the shared tree without the use of an additional out-of-band signaling protocol, like PIM or BGP [SM-MLDP]. For this reason, transiting shared trees across a transit LSP is outside the scope of this document.
IP multicast source trees can be created via PIM operating in either SSM mode [RFC4607] or ASM mode [RFC4601]. When PIM-SM is used in ASM mode, the usual means of discovering active sources is to join a sparse-mode shared tree. However, this document does not provide any method of establishing a sparse-mode shared tree across an MPLS network. To apply the technique of this document to PIM-SM in ASM mode, there must be some other means of discovering the active sources. One possible means is the use of MSDP [RFC3618]. Another possible means is to use BGP Source Active auto-discovery routes, as documented in [SM-MLDP]. However, the method of discovering the active sources is outside the scope of this document; as a result, this document does not specify everything that is needed to support the ASM service model using in-band signaling.
The source and group addresses are encoded into the a transit TLV as specified in Sections 3.1 and 3.2.
If a bidirectional IP multicast tree [RFC5015] has to be transported over an MPLS network using in-band signaling, as described in this document, it MUST be transported using an MP2MP LSPs. A bidirectional tree does not have a specific source address; the group address, subnet mask, and RP are relevant for multicast forwarding. This document does not provide procedures to discover RP-to-group mappings dynamically across an MPLS network and assumes the RP is statically defined. Support of dynamic RP mappings in combination with in-band signaling is outside the scope of this document.
The RP for the group is used to select the ingress LSR and the root of the LSP. The group address is encoded according to the rules of Sections 3.3 or 3.4, depending on the IP version. The subnet mask associated with the bidirectional group is encoded in the Transit TLV. There are two types of bidirectional states in IP multicast, the group specific state and the RP state. The first type is typically created when a PIM Join has been received and has a subnet mask of 32 for IPv4 and 128 for IPv6. The RP state is typically created via the static RP mapping and has a variable subnet mask. The RP state is used to build a tree to the RP and is used for sender-only branches. Each state (group specific and RP state) will result in a separate MP2MP LSP. The merging of the two MP2MP LSPs will be done by PIM on the root LSR. No special procedures are necessary for PIM to merge the two LSPs. Each LSP is effectively treated as a PIM-enabled interface. Please see [RFC5015] for more details.
For transporting the packets of a sender-only branch, we create a MP2MP LSP. Other sender-only branches will receive these packets and will not forward them because there are no receivers. These packets will be dropped. If that effect is undesirable, some other means of transport has to be established to forward packets to the root of the tree, for example, a multipoint-to-point LSP for example. A technique to unicast packets to the root of a P2MP or MP2MP LSP is documented in Section 18.104.22.168 of [MVPN-MSPMSI].
[RFC5036] Andersson, L., Ed., Minei, I., Ed., and B. Thomas, Ed., "LDP Specification", RFC 5036, October 2007.
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997.
[RFC6388] Wijnands, IJ., Ed., Minei, I., Ed., Kompella, K., and B. Thomas, "Label Distribution Protocol Extensions for Point- to-Multipoint and Multipoint-to-Multipoint Label Switched Paths", RFC 6388, November 2011.