The traditional layer 2 VPN technologies use MPLS tunnels in order to create point to point circuits, baring most of the layer 1 encapsulations. These circuits can transport traffic over an IP/MPLS network, only between two sites (point to point connection). To accomplish that the provider’s network to behave like an Ethernet environment, switching frames among several devices, is what has encouraged the creation of the Virtual Private LAN Service (VPLS). VPLS is defined as a VPN MPLS service in layer 2, point to multi-point, based on Ethernet.
In the point to point level 2 VPNs scenarios, the association of each circuit in the Provider Edge (PE) Router must be configured manually to remote sites. This requisite can be very demanding and complex, especially in “full-mesh” environments. When a VPLS is implemented, the client sees the instance as a LAN segment. The instance will behave in the same way as a Switch Ethernet, providing MAC learning, flooding and forwarding among all the PE routers involved. In VPLS, the PE routers learn the MAC addresses of the received frames, in the same way that a Ethernet switch would. This information is used to create a forwarding table dynamically and, based on the table, the Ethernet frames of one instance are switched to a local interface or through a LSP of the MPLS network, to another PE device. Thanks to the VPLS Ethernet functions, it is not necessary to associate circuits to remote circuits like in L2 VPN. Moreover, for a particular VPLS, the associated clients, who are connected in different remote sites, see the IP/MPLS network of the provider as a Ethernet switch.
VPLS uses the MPLS structure, over which the service is configured and a signaling session is used to establish the connections between the PE routers. As regards this signaling, there are two standard versions: One version uses BGP as signaling protocol (RFC 4761), and the other one uses LDP (RFC 4762). In the BGP version, the pre-existing MP-BGP sessions among the PE routers can be used and, if necessary, the same scheme of confederations and Route Reflectors can be used. Additionally, BGP offers the possibility of auto-configuration and creation of VPLS instances between different providers (Inter-AS). As an alternative, the LDP protocol can be used as signaling protocol between the PE routers of the topology. It is simpler but less flexible and scalable.