I was going to say host-routes as well but Markus beat me to the punch.

I definitely think we should target a solution without host changes though
as the tactical solution.

I thought LISP had implemented this type of stuff, eg: host address 
mobility for hosts on access-LANs. Can't quite remember how, but it must
be something involving the tracking of a device when it moves from it's
original location, and the first-hop router(s) would then create a host-route
for that host and make sure that the host keeps its address (acting as
DHCP server or the like ?)

Can't be that hard though. Definitely less work than trying to build
an L2 overlay to give stupid L2 APs the ability to do L2 roaming. IN
any case, it should at most be all control-plane, no new forwarding plane.

Of course, host-routes themselves don't solve the possible problem of
link-local multicast after mobility, but i'd very much hope that we
can get rid of apps relying on that anyhow.

Cheers
    Toerless

_______________________________________________
homenet mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/homenet

Reply via email to