Re: Another driver for v6?

["David W. Hankins" <David_Hankins () isc org>]

On Thu, Oct 30, 2008 at 08:10:26AM +0100, Mikael Abrahamsson wrote:
> On Wed, 29 Oct 2008, David W. Hankins wrote:
> > It is almost lunacy to deploy IPv6 in a customer-facing sense (note
> > for example Google's choice to put its AAAA on a separate FQDN).  At
>
> Could you please elaborate on this point? My data presented
> <http://www.ops.ietf.org/lists/v6ops/v6ops.2008/msg01582.html> indicates 
> that there are very very few (the longer I collected the data, the better 
> the ratio got) who cannot properly fetch a resource that has A/AAAA.

I'm sorry I led you down the wrong ferret hole.  The issue isn't
directly the involvement of a A/AAAA mixed RRsets.  The issue is that
such dual placement complicates debugging and operations.

If someone can't reach www.google.com now, you know that it is either
a DNS or IPv4 issue.  It is very straightforward.

If someone can't reach the hypothetical A/AAAA www.google.com RRset,
you've just increased your support costs.  "My network is slow."
"Are you using IPv4 or IPv6?" "Netscape."

This costs you something, but doesn't gain anything.

Nevermind that IPv6 often breaks at some networks, and no one at
the remote network seems to care to fix it.  It is someone's
experiment.

I'm recommending a variety of caution which is to go ahead and deploy
your initial/experimental IPv6 on separate FQDN's, so that you can
easily migrate your AAAA's onto "production names" when there is an
advantage to doing so, and in the meantime you aren't breaking
anything for the rest of the planet.

> > will migrate to IPv6.  Which seems like a pretty bad time to still be
> > trying to figure that out, but ohwell.
>
> 6to4 and Teredo traffic is increasing very rapidly, so that seems to be one 
> path taken right now:
>
> <http://ipv6.tele2.net/mrtg/total.html>

I don't know how to ask this question without sounding mean, but did
the graph spike out of zero, or did you start collecting two months
ago?


Both Teredo and 6to4 strike me as a kind of network operations
"terrorism."  That is, the clients that engage in this automatically.

It proposes precisely the same support-cost-increase problem for the
ISP ("My network is slow." "Are you using IPv4 or IPv6?" "Netscape."),
and it's not clear to me how an ISP can "opt out".

So it's kind of like these OS manufacturers are sending ISP's a little
message; spend your support costs, or we'll spend them for you.

I'm not sure that's productive overall.

> > IPv6: It's kind of like storing dry food in preparation for the
> >      apocalypse.
>
> If you actually KNOW the apocalypse is coming (but not when), this is 
> correct.

I think everyone knows the IPv4 shortfall is coming.  I do not think
the world's view of the consequences is consistent yet.

So I don't think it matters; it's prudent to get a defensible position
today.

-- 
Ash bugud-gul durbatuluk agh burzum-ishi krimpatul.
Why settle for the lesser evil?  https://secure.isc.org/store/t-shirt/
-- 
David W. Hankins        "If you don't do it right the first time,
Software Engineer                    you'll just have to do it again."
Internet Systems Consortium, Inc.               -- Jack T. Hankins

Attachment: _bin
Description: