Re: IPv6-only devices?

["Sathya Rao" <rao@telscom.ch>]

I liked particularly the recommendation of Brian how IETF can make a mileage
for early deployment of IPv6.

> > I think this means the IETF should restrict new
> > work to IPv6-only. In other words: either move to IPv6 or abondon
> > the effort.
>
> It's a bit soon for that. But we should clearly not accept any
> WG that does stuff that isn't IPv6-ready. So IPv6 people need to
> colonise every WG that exists, and read every draft for
> IPv6ness.
>


Sathya

----- Original Message -----
From: "Brian E Carpenter" <brian@hursley.ibm.com>
To: "Ronald van der Pol" <Ronald.vanderPol@rvdp.org>
Cc: <v6ops@ops.ietf.org>
Sent: Friday, February 07, 2003 1:44 PM
Subject: Re: IPv6-only devices?


> Ronald van der Pol wrote:
> >
> > On Thu, Feb 06, 2003 at 16:16:09 -0500, Keith Moore wrote:
> ...
> > If you read my initial email again you would see that I also say that
> > it is perfectly OK to have IPv6-only appliances that do not need to
> > communicate to IPv4-only nodes.
>
> And in a SOHO context or an on-board network on some kind of vehicle,
> it might be very natural for those IPv6-only devices to communicate
> with a local server that proxies for them in some way to the
> outside world. E.g. a "building services" server that doesn't
> expose individual IPv6 thermostats to the outside world, but does
> expose room temperatures. That server can be dual stack.
>
> This is to me the most natural model for really low-end devices, which
> will be very price sensitive (i.e. a few dollars difference in the cost
> of ASIC or memory really matters). Unless you want to build a NAT-PT
> with a Thermostat ALG for each band of thermostat.
>
> >
> > I think dual stack is the best migration path. I also think that the
> > transition period should be short. Not a flag day, but also not a
> > transition period of decades. I agree with you that running dual
> > stack networks is costly *). I think the transition should be done
> > in a few years.
>
> Er, how to say this politely? "In your dreams." The only realistic
> thing to plan for is an indefinite period of coexistence. Once IPv6
> is seriously out there, I can't see the overlap being less than ten
> years, and even that is very optimistic.
>
> Of course, the day should come when IPv4 is an overpriced
> legacy service only supported by a couple of dinosaur telcos, but
> I don't expect that to happen until I'm happily retired.
>
> > I think this means the IETF should restrict new
> > work to IPv6-only. In other words: either move to IPv6 or abondon
> > the effort.
>
> It's a bit soon for that. But we should clearly not accept any
> WG that does stuff that isn't IPv6-ready. So IPv6 people need to
> colonise every WG that exists, and read every draft for
> IPv6ness.
>
>     Brian
>
>