> -----Original Message-----
> From: Thomas Narten [mailto:[EMAIL PROTECTED]
> Sent: Tuesday, April 18, 2006 9:34 AM
> To: Durand, Alain
> Cc: Ralph Droms; Erik Nordmark; IPv6 Mailing List
> Subject: Re: Proposed M&O bits text for RFC2461bis
>
> "Durand, Alain" <[EMAIL PROTECTED]> writes:
>
> > BTW, M&A bot set *is* a valid configuration, where on the
> same subnet
> > one would like to use both servers that want to use nothing
> but DHCP
> > and laptops who are perfectly happy with stateless autoconf
> (or small
> > devices who do not have implemented anything else)
>
> This is the crux of the fundamental underlying disagreement
> we are having on what we are trying to solve here.
>
> In my view, it is nuts to have devices "that choose to
> implement only stateless addrconf" because they are "simple"
> or "perfectly happy" or something. If we go down this route,
> it is not an operational choice whether to run DHC, because
> implementations have already made that choice.
I agree with you 100% that implementing only stateless
because it is simpler is "nuts". If you look at the number
of packets that need to be exchanged, DHCP is really not that
hard.
However, one need to deal with today's reality.
It seems to still be the case in 2006 that most implementations only
offer stateless autoconf. There is some kind of myth that this
is "enough".
I'm designing a network where this is clearly not "enough" and
we have made a decision (for many reasons) to push for DHCPv6
address provisioning. I can get this on some critical devices
but I have to be realistic and deal with the fact that DHCPv6
code is still hard to get and there will be many years where
DHCP-enabled and non-DHCP-enabled devices may have to co-exist
on the same subnets.
- Alain.
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
