Re: IPv6 and Dynamic DNS

[Masataka Ohta <mohta@necom830.hpcl.titech.ac.jp>]

Paul;

> It seems to me that DNS (dynamic or not) can't address what seems to be a
> basic flaw in the basic model of stateless address autoconfiguration.

Right.

> We,
> in the person of Yakov and Susan, designed DNS UPDATE with an eye toward
> DHCP (v4 and v6).  Our assumptions were all based on there being an agent
> on the LAN (meaning the DHCP server) who had permission to update that LAN's
> inverse DNS entries (meaning IN-ADDR.ARPA and IP6.INT).  If stateless auto-
> configuration doesn't work the same way (meaning there's no agent on the LAN
> who has permission to update the inverse DNS entries for that LAN and who
> knows about new address assignments) then stateless autoconfiguration needs
> some work.

Needs some work?

Being an excellent implementer, you tend to specify and implement
anything.

However, if stateless autoconfiguration needs to be modified to
be stateful autoconfiguration, the reasonable approach is to
abandon it and just use DHCP.

Remember that IPng was expected to be simple IP.

> Is namedroppers really the place where this work should occur?

Histrically in IPng WG, the fundamental flaw of stateless
autoconfiguration has survived using other protocols/proposals
as scape goats.

So, namedroppers should be the place to sentence death penalty
to stateless autoconfiguration, unless you want dynamic DNS
be yet another scape goat.

							Masataka Ohta