Chris; > Bridging works fine between 802.11 and 802.1--or another 802.11 link. No > routing required. > It works for the IETF meeting networks and pretty much how all Enterprise > gear is setup. > Let's keep it simple and only route when we have to.
Certainly, but that's not the point. As long as you have a natively loopless topology, for which a WiFi ESS is a simple example, you can bridge and proxy/relay/repeat all you want. You'll note that the ESS model is reproduced at L3 when the 6LoWPAN backbone router proxies ND from a backbone link onto a LLN. The problem is that soon enough you'll need more than 1 relay AP between the pool and the lounge. You'll need it to improve throughput through spatial diversity. You'll need it so save power and reduce interferences. You'll need it to improve your delivery ratios. As soon as you have more than one relay and the connectivity is whatever meshing of links, loops can form and you need routing. This is why we came up with spanning tree and its evolutions in L2 switch networks. And all sorts of L2 mesh technologies in wireless, still L2 specific and proprietary wherever you look at. Then again, the IETF addressed the problem at L3, and came up with the RPL protocol that draws one (or more) loopless topology(ies) upon an arbitrary mesh to reach the gateways. We can go (again) through lengthy discussions but in short I agree with other voices (Russ) that indicate that we need a L3 routing solution. I also fully agree with the voices (Robert) that indicated that the backbone distribution network and the meshed fringe at the edge are different bests and might need different treatments. Cheers, Pascal _______________________________________________ homenet mailing list [email protected] https://www.ietf.org/mailman/listinfo/homenet
