In Multicast VPN (MVPN) using Multipoint to Multipoint LDP (MLDP),
multicast traffic received by an ingress provider edge (PE) router is
replicated/copied and passes through the core to egress PE routers. State is
maintained in the core and this may lead to some dynamic fluctuation of state
in the service provider core routers.
In PE to PE IR, multicast traffic from an ingress PE router is
replicated/copied and tunneled over multiple LSPs directly from the ingress PE
to egress PE. State is not maintained as the traffic passes through the core.
The primary advantage of using PE to PE Ingress Replication (IR) is not having to maintain state in the core as mentioned above. A potential disadvantage to using PE-PE IR, is that the bandwidth utilization in the core increases in some parts of the core; For example, at the point in the network shown in the figure below, at callout 3, there are multiple copies of traffic going from the Ingress PE in a path towards the egress PEs.
BGP MVPN AD is the control plane that is used between PEs for PE-PE IR.
Figure 1. PE-PE Ingress Replication Topology
Red-dashed line, which shows the route taken by one copy or replication of traffic from an Ingress PE to an Egress PE.
Cloud showing the network core.
On this route, a multicast packet is replicated four times by PE-PE Ingress Replication.
In the figure above, when the Ingress PE router receives multicast traffic, it sends a copy of the traffic to each of the egress PE routers: Egress PE 1, Egress PE 2, Egress PE 3 and Egress PE 4. The egress PE routers inform the ingress PE routers of their interest in receiving multicast traffic by originating BGP Leaf Auto-Discovery routes.
Each replication is sent over a MPLS Label Switched Path (LSP) to the
specific destination PE router. MVPN sites that are connected to a particular
egress PE are allowed to send multicast traffic to other sites in the same
LSPs are assumed to exist before PE-PE IR is configured. LSPs can be
setup using a variety of types of LSP. e.g BGP labeled unicast SAFI LSPs can be
used with PE-PE IR but it is better to use P2P (TE) LDP LSPs, which are
verified and tested.
Ingress Replication MVPN Profiles
For Ingress Replication, the MVPN Profile may be one of the following two profiles. Data MDT is supported within either of the following two profiles:
Partitioned MDT Profile
Default MDT Profile
Partitioned MDT Profile
In a partitioned MDT profile, only those egress PE routers that receive traffic requests from a particular ingress PE are associated with that ingress PE. This results in a low number of ingress PE routers in the network and a limited number of trees in the core.
When a Partitioned MDT profile is used in PE-PE IR, egress PEs may need identify traffic from ingress PEs, which uses an internal RPF-ID. A strict RPF check is done using the RPF-ID.
Interface-based strict RPF check is not supported. If the PE-PE IR MVPN is configured to include both IPv4 and IPv6 address-families of a VRF, the default or partitioned MDT profile must be same in both the IPv4 and the IPv6 address-families.
Default MDT Profile
In a default MDT Profile, the MDT carries control plane and data
traffic. All PE routers in the MVPN need to join the default MDT tree. The
number of MDT trees that are set up for every PE equals the number of PEs in
If the PE-PE IR MVPN is configured to include both IPv4 and IPv6 address-families of a VRF, the default or partitioned MDT profile must be same in both the IPv4 and the IPv6 address-families.
PE-PE IR profile must be same in IPv4. mLDP and PE-PE IR cannot exist
Similar to the Partitioned MDT profile, PE-PE IR cannot exist
alongside an mLDP MVPN profile under the same address family of the same VRF.
Also, a PE-PE IR cannot exist with mLDP inband signaling in the same address
family of a VRF.
By default, the number of data MDTs that can be used in either a
partitioned or default MDT profile for PE-PE IR is unlimited. However, each
data MDT consumes system resources; therefore, optionally, the number of data
MDTs can be limited between 1 and 20000.
An ACL can be optionally configured, via the list keyword, to control
the multicast flows that can use a data MDT. When the data MDT threshold is 0,
a PIM-SSM (multicast) state is assigned to a data MDT as soon as it is
instantiated. Using the immediate-switch keyword, the first packet of the
(multicast) flow switches onto the data MDT. Without this immediate-switch
keyword, a three second delay occurs before flow packets transition to the data
BGP auto-discovery (AD) control plane states are reduced by not
announcing certain BGP AD routes. Note that receiver-only sites are not
specific to PE-PE IR but also apply to mLDP. In a PE-PE IR configuration, each
egress PE must be connected to a receiver-only site. The PE-PE IR C-multicast
protocol is BGP C-routes.