BGP Route Reflectors for VPLS
In large networks, it might be necessary to configure BGP route reflectors to reduce the control plane workload for the routers participating in the VPLS network. BGP route reflectors can help to reduce the workload of the network control plane in the following ways.
Making it unnecessary to configure all of the VPLS PE routers in a full mesh.
Limiting the total volume of BGP VPLS messages exchanged within the network by transmitting messages to interested routers only (instead of all of the BGP routers in the network)
Reducing the network signaling load whenever another BGP router is added to or removed from the network
The basic solution to these problems is to deploy a small group of BGP route reflectors that are in a full mesh with one another. Each of the VPLS PE routers is configured to have a BGP session with one or more of the route reflectors, making it unnecessary to maintain a full mesh of BGP sessions between all of the PE routers.
In the VPLS documentation, the word router in terms such as PE router is used to refer to any device that provides routing functions.
This type of configuration only affects the control plane of the VPLS network (how routers signal and tear down pseudowires to one another in the network). The actual data plane state and forwarding paths for the VPLS traffic are not modified by the route reflectors. Effectively, the VPLS pseudowires should take the same paths across the network whether or not you have configured route reflectors. For a description of how VPLS selects the best path to a PE router, see VPLS Path Selection Process for PE Routers.
The MAC addresses themselves are not exchanged or processed in any way by BGP. Each VPLS PE router performs all MAC address learning and aging individually. BGP's only function relative to VPLS is to exchange messages related to automatic discovery of PE routers being added to and removed from the VPLS network and the MPLS label exchange needed to signal a pseudowire from one PE router to another.