VPLS.US

So, I was reviewing OTV and MACVPN and Ethernet VPN and VPLS, and had an idea to reduce some of the overhead that's associated with L2 boundaries. Today, Layer 3 boundaries represent the boundary between a Layer 2 domain and a Layer 3 domain. Why couldn't one of today's switches do this? They are looking into the packet for TOS / DSCP information which means that a switch is already looking way into the layer 3 header of the IP frame. So why couldn't every Source MAC coming from a particular location be hidden behind the MAC address of PE?

This would reduce the MAC-address-table of the PE and reduce the load and overhead of the whole system.

Anyway that's an idea. It might have some issues with resiliency ]]>