On Wed, Nov 07, 2012 at 06:29:54PM +0000, Simon Kelley wrote: > On 07/11/12 18:21, David Lamparter wrote: > > On Wed, Nov 07, 2012 at 06:03:52PM +0000, Simon Kelley wrote: > >> On 07/11/12 15:46, Ted Lemon wrote: > >> > >>> I think the disconnect here is that people are thinking the routers > >>> to which prefixes are delegated need to themselves be delegating > >>> routers, but this is incorrect. What they need to do is _relay_ > >>> prefix delegation requests to the delegating router from which they > >>> got their own delegation. > >>> > >> > >> ... and once the delegating routers are relaying DHCP-PD requests, they > >> can also relay DHCP address-lease requests to the DHCP server in the CPE > >> router. This puts all the information about DHCP leases and associated > >> names in one place, and greatly simplifies the naming problem. > > > > This really falls apart when I'm using 2 ISPs with 2 exit routers. > > a-k-a, "Where's up?" > > and be put back together by noting that 2 exit routers doesn't have to > mean 2 DHCPv6 servers. A single DHCP server can take PDs from 2 or more > upstream ISPs. If the problem is that exit routers come with DHCPv6 > servers, then adding the ability to disable all but one (automatically > or manually) would be worth considering.
Yes, and now we need some protocol to decide which of the routers serves as DHCPv6/PD server, that needs to work while crossing intermediate routers possibly... (assuming we want to do it automatically) So that's heading for another information propagation/flooding/election protocol, at which point I don't see why we shouldn't just use OSPFv3. -David _______________________________________________ homenet mailing list [email protected] https://www.ietf.org/mailman/listinfo/homenet
