> If you have followed the discussion closely, you should have > noticed that ARP is a lot better than ND in a typical > environment where WLANs are used as leaf of the Internet. > > So, as a short term solution, I'd like to suggest to use ARP, > not ND, over WLAN. > > As a long term solution, I already suggested > > 1. periodic beacons from routers > 2. unicast request from hosts to routers > 3. unicast reply from routers > > is the way to go. >
That is one way. In fact there are many ways. For instance, a simple one is to have APs proxy ND for their associated nodes. Basically they forward the multicast ND messages from the associated nodes to the wire only, and in any case answer multicast ND on behalf of their associated nodes. It may be a little more troublesome then it looks; but it shows that there's at least one way to go that's not changing things as dramatically as Ohta-san's proposal. An other aspect is that 802.11 is not THE radio model. Should we make each radio a new link type for IP? Or is there a common link type that we can converge on by providing an adaptation layer (2.5) ? Pascal -------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------