DHCPv6 for IPv6 routing information, was Re: Mysterious missing DHCPv6 feature

[Shane Kerr]

Steinar,

On Mon, 2010-05-17 at 08:44 +0200, sthaug at nethelp.no wrote:
> > > 	DHCP to hand out DNS servers, NTP servers on request, and do
> > > 	dynamic update for the forward and reverse DNS maps.
> > 
> > And the new feature you were looking for was to hand out a default-gw, 
> > right? That's the "this" feature you were referring to and IETF didn't 
> > want to do?
> 
> I'm glad to see such patches (I'm tempted to say "about time").
> 
> I would of course be even happier to see a *standardized* solution to
> let DHCPv6 hand out a default gateway. The lack of such a feature (and
> the strong religious opposition to it in certain circles), despite clear
> statements from several big operators that they need it, is one of the
> significant factors hampering IPv6 deployment.

I used to be a DHCPv6 guy, although not so much any more. (I was part of
the three-man team that coded ISC's first DHCPv6 release a few years
back.)

As I understand it, there was a persistent and insistent group of people
within the IETF that really, really, REALLY wanted DHCP to not exist on
IPv6 at all. I wasn't involved at all during those times, but I am led
to believe this was largely a "zeroconf" push.

Remember the idea that IPv6 was going to autoconfigure everything, and
no management would be necessary?

                    HAHAHAHAHAHAHAHAHAHAHAHAHAHA!!!

Honestly. Some people. I'm reminded of people who told me in 1990 that
programing would go away because it would be so easy to make software do
what you wanted it to.

Anyway, whatever the reason, there is a group of DHCP-haters. Especially
in the IPv6 universe.



Having said that, the philosophy has been that for each problem we
should have ONE protocol to solve it. This makes a certain amount of
sense. With routing information, this is supposed to be router
advertisements (RA). This is the argument against putting routing
information in DHCPv6 (as I understand it).

Strangely, rather than saying "look we have DNS information in DHCPv6 so
we don't need it in RA", the tactic was taken to create a RA that
includes DNS servers. Which sort of seems like TWO protocols to solve a
particular problem. Oh well, at least it's only EXPERIMENTAL (RFC 5006).

There does seem to be a draft for exactly this feature:

http://tools.ietf.org/html/draft-dec-dhcpv6-route-option-03

So there's hope for the future. The "intended status" is INFORMATIONAL
though, and it hasn't been accepted as a working group item yet, so who
knows... :)

--
Shane