[dhcwg] Re: draft-kinnear-dhc-relay-agent-flags-00.txt

On Thu, Jun 08, 2006 at 03:53:01PM -0400, Kim Kinnear wrote:
> We have put together a draft which specifies a flags sub-option
> for the relay agent info option, and includes a flag for unicast
> or broadcast reception by the Relay Agent:
> 
> http://www.ietf.org/internet-drafts/draft-kinnear-dhc-relay-agent-flags-00.txt

This is perfect.

In the event an option content longer than 1 byte is received by the
server, do we need stronger language there to make sure servers evaluate
the first octet?  It's pretty plain to me since it describes itself as
an extensible flags field, but I'm also going into this knowing that's
the goal.

I also think the flags field gives us a general hammer to use when (if?)
someone undertakes the work for Ted-Complete (sorry Ted) failover support,
wherein the servers no longer receive copies and the relay agent provides
the load balancing...the agent advertising the capability can be carried
this way for example without additional option overhead.

I think I see (hazily) how failover works in this interim world order
with just this flag.  I will endeavour to see this more crisply before
the submission deadline.

> In any case, whatever we decide to do here (and it doesn't make
> *that* much difference), we thought that reviewing the ideas and
> words might be easier using a separate draft.  Once we get
> general agreement that we have the sub-option we want, we can
> integrate it into the server-id-override draft or continue to
> push it as a separate draft.

I said it in person in Dallas, I'll encode it here for posterity as
well: I only ask that the WG accept Kim's draft as a WG item.  That
solves my objection.  In my opinion, with that done, the other draft
may proceed in parallel on the basis that despite even my own concerns
for "protocol purity".  It solves the more serious problem where
the servers and clients can't directly unicast if only people are
a little careful about how they deploy the technology.

> If we can get feedback soon on this draft, then we can fix it in
> time for the submission deadline for the next IETF and integrate
> it if that is the sense of the list.

Given.

> Enjoy -- Kim

Thank you for your effort on this, Kim, I appreciate it.

-- 
David W. Hankins		"If you don't do it right the first time,
Software Engineer			you'll just have to do it again."
Internet Systems Consortium, Inc.		-- Jack T. Hankins