On Oct 12, 2010, at 12:38 PM, David W. Hankins wrote: > As a B4 implementor, I reject the onerus requirement imposed to > contain a DNS recursive stub resolver in B4 equipment, and to > complicate the configuration of the B4 software to be conditional > on the success (through retries) of its resolution.
I'm highly sympathetic. But as you so eloquently argued earlier, having both options is not a good choice. Our initial recommendation was to just use the IP address option, and let the DHCP server load balance or not as its configuration dictates. This was rejected by Mohamed on the basis that DNS is the right way to do load balancing (a point which I find somewhat questionable). I would strongly encourage the working group to develop a consensus *not* to use the FQDN option--it adds little value, much complexity, and is usually a deal-breaker for the DHC working group. The only reason that this didn't get more pushback from the DHC working group is that it was only discussed very briefly there, and we were left with the impression that the FQDN option was going to be removed. The expected formal request for review of this draft that should have come from softwires never did, so we only realized that the draft was broken when it hit the IESG. _______________________________________________ Softwires mailing list Softwires@ietf.org https://www.ietf.org/mailman/listinfo/softwires
