In an effort of figuring out a metro ethernet access network to carry Internet access services, I'm comparing the approaches of several equipment manufacturers insofar as how, they allow a metro service provider to provide point-to-multipoint E-TREE public services, (where the leaves of the multipoint tree are different billing entities from that of the root of the tree), such that it remains possible to distinguish at the root traffic originating from a given service provider demarcation device in the tree. In an effort to solve this problem, I am finding myself diving deep down into the behaviour of Ethernet switches at the 802.1Q 2005 bridge behaviour. I would appreciate feedback from people whom are aware since WHEN and under which conditions, Ethernet switches have begun to implement separate MAC addresses for every client ports. and whether such MAC addresses draw from the global pool of MAC addresses. A cursory understanding of this leads me to believe that as soon as a switch is capable of spanning tree on a downstream client port, it must have separate MAC addresses for every client port. Feedback appreciated -=Francois=-