Re: [v6ops] DHCPv6/SLAAC Make Hosts Confusing-//RE: new draft: draft-liu-bonica-v6ops-dhcpv6-slaac-problem

[james woodyatt <[e-mail]>]

On Oct 23, 2013, at 23:41 , Lorenzo Colitti <[e-mail]> wrote:

> I see two possible ways out of this:
> 
> 1. Decide a clear split between what goes into RA and what goes into DHCPv6, and then stick to it. For example, we could say that configuration information that is required to obtain basic connectivity on a network must be available in RA so it can be dynamically updated and shares fate with routing, and that everything that's not strictly required for the host's connectivity can go into DHCPv6. I don't know if it will be possible to draw a line here. If operators insist that DHCPv6 by itself must be sufficient, then we have no choice but to duplicate information.

I prefer this way out.  I've said this before, but now seems like a good time to say it again: stateless DHCPv6 has a lifetime of information problem, and I think RFC 3736 should be sent to pasture.

p1. If there are information objects that must be individually and separately configured at each host by the network, then stateful DHCPv6 [RFC 3315] is the protocol for doing that.

p2. The only information objects that deserve to be in RA messages are those that facilitate discovering routers and domain name resolving servers. (If pressed, I can come up with a couple of very minor exceptions, but I'm saving that for another discussion.)

p3. If there are information objects that do not strictly need to be individually and separately configured at each host by the network, and they are also not strictly required to facilitate routing and domain service discovery, then there are two viable options available: A) use stateful DHCPv6 despite not requiring state, or B) use DNS-SD, which is stateless.  If both are available on the same network, then DHCPv6 information should override DNS-SD information, because— well— sometimes the network decides that hosts need state. There is no good reason to lard up RA messages with this stuff too.

If operators insist that DHCPv6 by itself MUST be sufficient, then the only information that requires duplication is the minimal set that is currently defined in RA messages. Everything else

--james woodyatt <[e-mail]>