MPLS Allows Multisite LANs via VPLS
In Figure 1, the PE devices are termed edge nodes. Typically, edge nodes are more powerful than the core P nodes because the PEs aggregate many incoming links and services. In addition, edge nodes may operate at several protocol layers; for example, layers 1, 2, and 3. Edge devices are busy, technology-rich, and complexin other words, big business for vendors!
An increasingly important service is that of VPLS, in which the PE devices receive (and send) Ethernet frames that originate (and terminate) in adjacent customer LANs. The service provider wants these frames to be transmitted through its core network and on to another site associated with the same customer LAN. In other words, the service provider core acts just like a big Ethernet switch.
Each PE device must host a virtual forwarding instance (VFI) that examines the incoming Ethernet frames and pushes them into the core in the direction of the intended recipient. The core contains MPLS connections or LSPs that connect the PE nodes and provide the core transport mechanism. In Figure 1, LSP A and LSP B are two examples of core MPLS connections.
Suppose a frame arrives at PE1 (from the HQ CE), destined for the remote office site. Let's say that this frame has a destination address contained in the remote office site. The destination MAC address is looked up and the frame is MPLS-encapsulated and pushed into LSP A. The MPLS-encapsulated frame arrives at PE3, the MPLS details are stripped off, and the frame is forwarded across Link C to the remote office site. If the destination MAC address is unknown, or is a broadcast/multicast address, the frame is flooded to all ports associated with the ingress VFI. In this case, the service provider functions as a learning bridge.