On 13/11/12 20:48, Victor Kuarsingh wrote:
On Tue, Nov 13, 2012 at 3:19 PM, Simon Kelley<si...@thekelleys.org.uk>wrote:
Given that hosts are going to want to talk RA or DHCPv6, at least
initially, one option down this route has the flood include the unicast
address of a single, centralised DHCPv6 server, and routers run DHCP-relay
agents which forward to that address. That gives you DHCPv6 functionality
without recursive-PD complications. It also eliminates the problems of
distributing the available prefixes as they traverse PD chains. The
one-and-only DHCP server knows about all the prefixes delegated from the
ISP and the relays know which particular prefix has been given to the local
router by the routing protocol or AHCP.
I assume in your model, hosts on the various subnets will then have access
to a centralized DHCPv6 server and other hosts can still use SLACC (which
is connected to ::/64s on the local attached router's interface). Would
the internal routers also get the ::/64 prefixes from the central DHCPv6
server (assume so)? If yes, would they grab a PD (with hint) for a range
large enough to cover all of its' interfaces?
This part of the scheme is still wooly in the extreme, but my thinking
is that DHCP-PD would _not_ be involved. The centralized server would be
a DHCP-PD _client_ and get a set of prefixes from the ISP(s), but the
distribution of those prefixes over the routers would be a side-effect
of whatever configuration system generates the routing tables and
spanning-tree for the homenet. I think I understood the OSPF guys at
IETF as saying that OSPF could do this, and I think I understand Dave as
saying that AHCP can do it. More details of either or both schemes would
be very interesting to hear.
Cheers,
Simon.
_______________________________________________
homenet mailing list
homenet@ietf.org
https://www.ietf.org/mailman/listinfo/homenet
