Re: DNS discovery

[Tim Chown <tjc@ecs.soton.ac.uk>]

On Wed, Nov 12, 2003 at 09:49:55AM +0900, itojun@iijlab.net wrote:
> 
> 	given that, my proposal is to pick dhcpv6-lite as the default mechanism
> 	to be used.  with reading the previous paragraph, could those who
> 	favors RA-based approach hum for dhcpv6-lite?

I think we had a 3:1 or better consensus that a single method should be 
agreed upon now.  But it was a lot less than 3:1 that that should be
DHCPv6-lite.

I feel the (multicast) RA-style method may have usage cases.  But we
don't have the operational experience yet to know what such cases may be
(in preference to DHCPv6-lite), and while RAs are well-defined, the
method to carry DNS info in the RA is not yet so (compared to DHCPv6-lite).

It seems the best way forward is to agree that DHCPv6-lite is the method
that implementors should proceed with now (some already have) and we
suck it and see.

That doesn't preclude:
a) people continuing to define RA method and implement/test it
b) future modification of dhcpv6-lite to multicast messages/responses
c) future studies of how dhcpv6-lite and RA method can co-exist
d) a future decision that RA method is the best (unlikely, but...)

Lets get the client support and dhcpv6 lite server and relay agents
implemented (Itojun said 13K lines of code for all?) and some experience
to base future work on.

Tim
#----------------------------------------------------------------------
# To unsubscribe, send a message to <dnsop-request@cafax.se>.