Re: [dhcwg] Call for Adoption: draft-yeh-dhc-dhcpv6-prefix-pool-opt-08

[Ted Lemon <[e-mail]>]

On Sep 5, 2012, at 4:06 AM, Ole Trøan <[e-mail]>
 wrote:
> the DHCP server is not on the data path. unless one follows the model described in 3633 where the DHCP server is on the PE router.

Oh, I see what you mean.   No, the DHCP server isn't on the data path, but it's using the data path.   The fate sharing is with respect to the PE router—the relay—and the CPE router.   Since both are being configured by the same message, there is no possibility of partial configuration.

> I might be missing something here. I don't understand how the aggregate route for a set of delegated prefixes is dynamic. when and how could all subscribers within a aggregate move?

The prefix might expand or contract depending on usage patterns.   I don't know if that's actually a use case that Leaf is considering—it's just one that occurred to me when I read the draft.

> hmm, I didn't think the IPv4 aggregate route was deduced or configured from DHCPv4 either. the deployments I'm aware of, the PE is always configured outside of DHCP with the IPv4 subnets / aggregate routes. what's the analogy?

The analogy is that the PE router's downstream routing configuration is in fact determined from snooping DHCPv4.  This is the entire justification for leasequery—to rebuild the router's DHCP snooping cache after a power cycle.