VPLS Provides for Core Discovery and Signaling
Let's take the dime tour of the network core in Figure 1. When services have been deployed in any network, the configuration and operational details are, in a sense, corporate assets of the service provider. In Figure 1, the VFIs, LSPs, protocols, and so on combine to form a network service. Access to this service is then sold to the customers, who use it to send and receive their Ethernet traffic. The provider wants to do two things:
Know as much as possible about the deployed service
Be able to deploy the service easily and quickly
This is where discovery and signaling come in. An important element of network management is the (preferably) automatic discovery of the network nodes, interfaces, links, protocols, and services. If the network provides this information automatically, it's likely to be both more accurate and more rapidly acquired than if a human user attempts to infer it.
If the core can be set up so that it's self-configuring, the network manager's workload is substantially reduced. In other words, if the core MPLS LSPs are set up automatically, you may get to see your favorite network manager at the next Friday afternoon beer fest.
As it happens, there are two VPLS draft standards, often called (after their respective authors) draft Kompella and draft Laserre-Vkompella. Let's call these DK and DLVK, respectively. The details can be found at the IETF web site, but here's a quick summary of the differences:
DK specifies auto-discovery using the BGP protocol.
DLVK doesn't specify auto-discovery.
DK advocates the use of BGP for signaling.
DLVK advocates the use of LDP for signaling.
There are issues and tradeoffs with each of the above.