M-LDP Signaling Through BIER Core

Some operators that are using mLDP P2MP LSPs for their multicast transport would like to deploy BIER technology in some segments of their network. As an example, if a operator is simplifying their core network with IGP underlay and BGP overlay and migrating from LDP to Segment Routing, they might prefer BIER in this segment of network for Multicast technology as BIER uses IGP underlay. The operator might still want to legacy multicast protocols like mLDP on the access of the network and offer end to end multicast services. This draft explains a method to signal mLDP services through a BIER domain, with minimal disruption and operational impact to the mLDP domain.

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 .

Some of the terminology specified in is replicated here and extended by necessary definitions: BIER: Bit Index Explicit Replication, The overall architecture of forwarding multicast using a Bit Position. BFR: Bit Forwarding Router, A router that participates in Bit Index Multipoint Forwarding. A BFR is identified by a unique BFR prefix in a BIER domain. BFIR: Bit-Forwarding Ingress Router, The ingress border router that inserts the Bit Map into the packet. Each BFIR must have a valid BFR-id assigned. BFIR is a term used for data plane packet forwarding. BFER: Bit-Forwarding Egress Router, A router that participates in Bit Index Forwarding as leaf. Each BFER must have a valid BFR-id assigned. BFER is a term used for data plane packet forwarding. BBR: BIER Boundary router. The router between the LDP domain and BIER domain. IBBR: Ingress BIER Boundary Router. The ingress router from a signaling point of view. It maintains mLDP adjacency toward the LDP domain and determines if the Multipoint LDP FEC needs to be signaled across the BIER domain via Targeted LDP. EBBR: Egress BIER Boundary Router. The egress router in a BIER domain from signaling point of view. It terminates the targeted LDP signaling through the BIER domain. It also keeps track of all IBBRs that are part of this P2MP tree BIFT: Bit Index Forwarding Table. BIER sub-domain: A further distinction within a BIER domain identified by its unique sub-domain identifier. A BIER sub-domain can support multiple BitString Lengths. BFR-id: An optional, unique identifier for a BFR within a BIER sub-domain, all BFERs and BFIRs need to be assigned a BFR-id. ILM: MPLS Incoming Label Map.

|--------BIER-------->|-------------> Datapatah (new)]]> As per above figure, point-to-multipoint and multipoint-to-multipoint LSPs established via mLDP can be signaled through a bier domain via Targeted LDP sessions. The procedure of using LDP Multipoint Extensions on Targeted LDP session is explained in . This document provides additional details and defines some needed procedures to use LDP Multipoint Extensions on Targeted LDP as per to connect two or more mLDP domains over a BIER domain.

In above Figure, the Ingress BBR (IBBR) is connected to the mLDP domain on downstream and a BIER domain on the upstream. To connect the LDP domains via the BIER domain, IBBR needs to establish a targeted LDP session with EBBR closest to the root of the P2MP or MP2MP LSP. To do so IBBR will follow procedures in in particular section 6 "Targeted mLDP with Multicast Tunneling". The Target LDP session can be established manually via configuration or via automated mechanism.

tLDP sessions can be signaled automatically from every IBBR to the appropriate EBBR. When mLDP FEC arrives at the IBBR from LDP domain. IBBR can automatically start a tLDP session to the EBBR closest to the root node. Both IBBR and EBBR should be in auto-discovery mode and react to the arriving tLDP signaling packets (i.e. targeted hellos, keep-alives etc...) to establish the session automatically. The root node address in the mLDP FEC can be used to find the EBBR. To identify the EBBR, the same procedures as section 2.1 can be used or the procedures as explained in the appendix A.

If the IBBR finds multiple equal cost EBBRs on the path to the root, it can use a vendor specific algorithm to choose between the EBBRs. These algorithms are beyond the scope of this draft. As an example the IBBR can use the lowest EBBR IP address to establish its mLDP signaling to.

The Egress BBR (EBBR) is connected to the upstream mLDP domain. The EBBR should accept the tLDP session generated form the IBBR. It should assign a unique "upstream assigned label" for each arriving FEC generated by the IBBRs. The EBBR should follow the procedures with the following modifications: The label assigned by the EBBR cannot be Implicit Null. This is to ensure that the identity of each p2mp and/or mp2mp tunnel in the BIER domain is uniquely distinguished. The label can be assigned from a domain-wide Common Block (DCB) The Interface ID TLV, as per should includes a new BIER sub-tlv as tunnel identifier. The EBBR will also generate a new label and FEC toward the root in the LDP domain. The EBBR Should stitch this generated label with the "upstream assigned label" to complete the P2MP or MP2MP LSP. With the same token the EBBR should track all the arriving FECs and the IBBRs that are generating these FECs. The EBBR will use this information to build the BIER header for each set of common FEC arriving from the IBBRs. The EBBR will use the BFR-ID of the IBBRs to build the BIER header.

Upon receiving the "upstream assigned label", the IBBR should create its own stitching instruction between the "upstream assigned label" and the downstream signaled label.

As per when LDP is used for upstream label assignment,the Interface ID TLV is used for signaling the Tunnel Identifier and it carries sub-TLVs. This document defines two new Interface ID sub-TLVs for BIER. Below is the Interface ID BIER sub-TLV for IPv4 BIER prefix:

Below is the Interface ID BIER sub-TLV for IPv6 BIER prefix:

On the BFIR when the MPLS label for P2MP/MP2MP LSP arrives from upstream, a lookup in the ILM table is done and the label is swapped with the tLDP upstream assigned label. The BFIR will note all the BFERs that are interested in specific P2MP/MP2MP LSP (as per section 3.2). The BFIR will put the corresponding BIER header with the bit index set for all the IBBRs interested in this stream. The BFIR builds this bit index from the BFR-ID of the intrested IBBRs. The BFIR will set the BIERHeader.Proto = MPLS and will forward the BIER packet into the BIER domain. In the BIER domain, normal BIER forwarding procedure will be done, as per The BFERs will receive the BIER packet and will look at the protocol field of BIER header, indicating MPLS protocol. The BFER will remove the BIER header and will do a lookup in the ILM table for the upstream assigned label and perform its corresponding action. It should be noted that these procedures are also valid if BFIR is the ILER and/or BFER is the ELER as per

The procedures above are also valid for mLDP recursive FEC. The root used to determine the EBBR is the outer root of the FEC. The entire recursive FEC needs to be preserved when it is forwarded via tLDP and the label request.

IANA maintains a registry of Interface ID Types for use in GMPLS in the registry "Generalized Multi-Protocol Label Switching (GMPLS) Signaling Parameters" and sub-registry "Interface_ID Types" This document defines and requests two new Interface_ID Type for BIER from the Interface_ID Types space, BIER IPv4 prefix TLV (Value TBD) BIER IPv6 prefix TLV (Value TBD)

While in the BIER domain the security considerations of are relevant to this document. The implementation should also take into account the security recommendations of .

Authors would like to acknowledge Jingrong Xie and Nabeel Cocker for his comments and help on this draft.