Re: [dhcwg] WGLC: draft-ietf-dhc-dhcpv6-stateful-issues-00

[Ole Trøan <[e-mail]>]

Ted,

> Not wearing my chair hat, I support what this draft is trying to do.   However, the draft makes a decision to unify the state machines of the IA_NA and IA_PD maintenance process.   We've had some discussions about this with customers, and one question that came up was whether it might make sense to have different DHCP servers answering IA_NA and IA_PD requests.   Obviously this is not possible if the transaction can't be split.
> 
> Have the authors considered this issue?   Any opinions on the topic?
> 
> The reason for the potential separation is that in some cases the business logic involved in delegating prefixes may be different than the logic used to allocate addresses.   Keeping the two functions separate enables significant additional flexibility when provisioning CPE devices.   This draft, if it were to progress in its current state, would rule out any such flexibility.   This is not an absolute deal breaker for me, but I wanted to raise the issue and see if anybody else has thought about this.

yes, we have talked about this issue. this draft clarifies correct protocol behaviour when multiple stateful options are present. I suppose it implicitly recommends one single DHCP session for multiple stateful options, but I don't think the text prohibits using multiple sessions either. that would be a choice the client could make, given the set of ADVERTISE messages received.

could we go forward with this draft that clarifies how multiple stateful options work in a single session, and with the general assumption that a client will choose one DHCP server and that server is authoritative?

then have separate specifications for how clients should behave in a multi-domain / separate session scenario if there is interest for that?

cheers,
Ole