On Aug 21, 2007, at 05:33, Shane Kerr wrote:
[...] having multiple protocols to do one thing is a bad idea, right?
Yes, but having one protocol to do multiple things in cross-cutting problem domains is a bad idea, too.
We need DHCP - in IPv6 as much as in IPv4 - so can we please let this idea that we can do everything in RA die? Please?
I don't want to see RA get bloated out with all the functions currently jammed into DHCP. That would only expand the boundaries of the current trouble, rather than addressing any real problems. I'd prefer to see the functions of DHCP split into profiles of other existing protocols, e.g. EAP, IKE, DNS-SD, etc. Any additions to RA needed to facilitate those operations should be strenuously avoided.
My major philosophical complaints about DHCP are that it (a) wastes multicast network resources doing unnecessary infrastructure discovery, (b) conflates interface address and node authorization identity for accounting purposes, and (c) complicates network configuration by drawing an unnecessary distinction between host configuration and other application data. I realize that others have differing opinions, and I fully respect them. These are mine, and if you don't like them, then I have others.
I don't agree that we *need* DHCP, but I'll agree that we *have* DHCP and it makes practical sense to use it in the interim while we develop a more practical alternative. We need something better than DHCP. It's just *not* a very good hammer-- and, you know what they say: when all you have is a hammer, everything begins to look like a folk singer.
-- james woodyatt <[EMAIL PROTECTED]> member of technical staff, communications engineering -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
