rfc-7117

Multicast in Virtual Private LAN Service (VPLS)

R. Aggarwal (Editor), Y. Kamite, L. Fang, Y. Rekhter, C. Kodeboniya
date2014-02 streamIETF areartg wgl2vpn statusPROPOSED STANDARD pages50 canonicalhttps://www.rfc-editor.org/rfc/rfc7117 doi10.17487/RFC7117 errataview
RFCs 4761 and 4762 describe a solution for Virtual Private LAN Service (VPLS) multicast that relies on the use of point-to-point or multipoint-to-point unicast Label Switched Paths (LSPs) for carrying multicast traffic. This solution has certain limitations for certain VPLS multicast traffic profiles. For example, it may result in highly non-optimal bandwidth utilization when a large amount of multicast traffic is to be transported. This document describes solutions for overcoming a subset of the limitations of the existing VPLS multicast solution. It describes procedures for VPLS multicast that utilize multicast trees in the service provider (SP) network. The solution described in this document allows sharing of one such multicast tree among multiple VPLS instances. Furthermore, the solution described in this document allows a single multicast tree in the SP network to carry traffic belonging only to a specified set of one or more IP multicast streams from one or more VPLS instances.

Extracted elements (28)

design-rationale §3.5

Aggregation reduces P-router forwarding state so it grows as (AveVPLS / Aggr) × (NPE / AvePTree) rather than linearly with the number of VPLS instances. The Aggregation Factor (Aggr) is the number of VPLS instances sharing a single P-multicast tree.

mpls, multicast

design-rationale §1

Ingress replication for VPLS multicast causes highly non-optimal bandwidth utilization when a large amount of multicast/broadcast traffic is transported, because the ingress PE must replicate each packet for every egress PE in the VPLS instance and send it via unicast tunnels. P2MP trees in the SP network overcome this limitation.

mpls, multicast

design-rationale §8

Switching a high-bandwidth (C-S, C-G) stream to a Selective tree is beneficial when the pseudo-wasted bandwidth—bandwidth multiplied by the number of PEs receiving the stream but having no receivers—exceeds a configured threshold; low-bandwidth streams may not warrant the additional tree state even if sparsely subscribed.

mpls, multicast

interoperability-note §7.3

Non-segmented inter-AS P2MP tunnels using mLDP (without RFC 6512) require that P-routers in one AS have IP reachability to the loopback addresses of PE routers in another AS, meaning those loopback addresses must be present in the IGP of the transit ASes.

mpls, multicast, routing

interoperability-note §3.3

The multicast membership discovery procedures in this document do not apply when PIM is used as the CE-CE multicast routing protocol and PIM join suppression is not disabled on all CEs. In that scenario, PEs cannot reliably determine IP multicast membership, and procedures for this case are outside the scope of this document.

mpls, multicast

normative-requirement §7.2.2.2 MUST

A Leaf A-D route sent via EBGP in the segmented inter-AS model MUST carry the NO_ADVERTISE community to constrain its scope to the immediate EBGP neighbor. Leaf A-D routes for S-PMSI contexts MUST carry the NO_EXPORT community except in inter-AS option (c).

mpls, multicast, routing

normative-requirement §8.1 MUST

After announcing a (C-S, C-G)-to-Selective-tree mapping, the source PE MUST wait for a configurable switchover delay before transmitting on the Selective tree; no (C-S, C-G) packet may be sent on both trees simultaneously.

mpls, multicast

normative-requirement §6.4 MUST

An Aggregate Inclusive or Aggregate Selective P-multicast tree MUST use MPLS encapsulation as specified in RFC 5332.

mpls, multicast

normative-requirement §6.1 MUST NOT

An egress PE MUST NOT advertise IMPLICIT NULL or EXPLICIT NULL for a P2MP LSP that is carrying traffic for one or more VPLS instances, since the non-null label is required for LSP identification and VPLS demultiplexing.

mpls, multicast

normative-requirement §7 MUST

An implementation MUST support inter-AS options (a), (b), and (c), and MAY support option (e). Option (b) MUST be supported using the segmented inter-AS tree model based on BGP VPLS A-D NLRI.

mpls, multicast, routing

normative-requirement §6 MUST

An implementation that supports this document MUST support both RSVP-TE P2MP LSPs [RFC4875] and mLDP P2MP LSPs [RFC6388] as P-multicast tree signaling technologies.

mpls, multicast

normative-requirement §4 MUST

An LDP-VPLS implementation supporting this document MUST support BGP A-D procedures to set up P-multicast trees; it MAY additionally support FEC 129 (Generalized PWid FEC) to automate pseudowire signaling.

mpls, multicast, routing

normative-requirement §7.2.2 MUST

For a given inter-AS tree and a given AS, there MUST be only one ASBR within that AS that accepts traffic flowing on the tree, and MUST be only one ASBR in that AS that sends traffic on that tree to any particular adjacent AS.

mpls, multicast, routing

normative-requirement §4.1 MUST

If an Inclusive P-multicast tree is used to instantiate the provider tunnel, the advertising PE MUST include the PMSI Tunnel attribute in its BGP Intra-AS VPLS A-D route, carrying the tunnel type and the P-multicast tree identity.

mpls, multicast, routing

normative-requirement §7.1 MUST NOT

In inter-AS options (a) and (e), BGP VPLS A-D routes generated by PEs within an AS MUST NOT be propagated outside that AS; only routes originated by ASBRs are permitted to cross AS boundaries.

mpls, multicast, routing

normative-requirement §5.1 MUST

Penultimate Hop Popping (PHP) MUST be disabled on any P2MP LSP (RSVP-TE or mLDP) carrying VPLS traffic. Egress PEs rely on the label advertised to the upstream neighbor to identify the P2MP LSP on which a C-multicast data packet arrives.

mpls, multicast

normative-requirement §4.1 MUST

When aggregating multiple VPLS instances onto one P-multicast tree, the PMSI Tunnel attribute of each re-advertised or newly advertised Intra-AS A-D route MUST carry the tree identity and an MPLS upstream-assigned label that is distinct within the scope of the advertising PE.

mpls, multicast, routing

normative-requirement §3.5 MUST

When multiple VPLS instances are mapped onto the same aggregate P-multicast tree, upstream-assigned labels [RFC5331] MUST be assigned and distributed by the root of the tree to enable egress PEs to demultiplex per-VPLS traffic.

mpls, multicast

normative-requirement §8.1 MUST NOT

When the P2MP LSP is instantiated by RSVP-TE, an egress PE MUST NOT switch to the Selective tree until both the S-PMSI A-D binding announcement has been received and the RSVP-TE P2MP LSP has been established to that PE.

mpls, multicast

protocol-element §8.3

A Leaf A-D route is originated by an egress PE in response to receiving an S-PMSI A-D route with the Leaf Information Required flag set to 1. It carries a PMSI Tunnel attribute with Tunnel Type set to Ingress Replication and a downstream-assigned MPLS label used to demultiplex VPLS traffic received over the unicast tunnel.

mpls, multicast, routing

protocol-element §3.1

A Selective P-multicast tree carries IP multicast traffic for one or more specific (C-S, C-G) streams only to PEs that have receivers for those streams, avoiding flooding uninvolved PEs. Selective trees MUST NOT be used for non-IP multicast traffic.

mpls, multicast

protocol-element §3.1

An Aggregate Selective tree carries traffic belonging to a specified set of IP multicast streams from more than one VPLS instance on a given PE. Aggregation binds multiple VPLS instances to the same P-multicast tree and requires upstream-assigned labels for demultiplexing.

mpls, multicast

protocol-element §3.1

An Inclusive P-multicast tree is a P2MP tree in the SP network that carries all multicast traffic from a specified set of VPLS sites on a given PE, regardless of whether traffic is IP-encapsulated. An Aggregate Inclusive tree carries traffic from more than one VPLS instance on the same tree.

mpls, multicast

protocol-element §8.2

An S-PMSI A-D route is originated by an ingress PE in the MCAST-VPLS address family to announce the binding of specific (C-S, C-G) streams to a Selective P-multicast tree. It carries an RD, multicast source and group fields (with wildcard encoded as address 0), the Originating Router's IP address, and a PMSI Tunnel attribute.

mpls, multicast, routing

protocol-element §9.1

The PMSI (Provider Multicast Service Interface) Tunnel attribute [RFC6514], carried in BGP VPLS A-D routes, encodes the tunnel type (LDP P2MP LSP or RSVP-TE P2MP LSP), the tunnel identifier, an optional upstream-assigned MPLS label, and a Leaf Information Required flag for explicit tracking.

mpls, multicast, routing

registry §14

IANA assigned a new Subsequent Address Family Identifier (SAFI) value for the MCAST-VPLS address family, used to carry the S-PMSI A-D routes and Leaf A-D routes defined in this document.

mpls, multicast, registry

security-consideration §13

This document inherits the security considerations of RFC 4761 (BGP-VPLS) and RFC 4762 (LDP-VPLS). Unauthorized injection or modification of BGP VPLS A-D routes or PMSI Tunnel attributes could redirect or disrupt VPLS multicast traffic; BGP security mechanisms (e.g., MD5 authentication, BGPsec) should be applied to protect A-D route exchange.

mpls, multicast, security, routing

wire-format §8.2

The MCAST-VPLS NLRI used in S-PMSI A-D routes contains: Route Distinguisher (8 octets), Multicast Source Length (1 octet), Multicast Source (variable), Multicast Group Length (1 octet), Multicast Group (variable), and Originating Router's IP Address. A wildcard source or group is encoded as address value 0.

mpls, multicast, routing