[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]


To: alh-ietf@tndh.net
cc: ngtrans@sunroof.eng.sun.com, namedroppers@ops.ietf.org, ipng@sunroof.eng.sun.com, dnsop@cafax.se
From: Robert Elz <kre@brandenburg.cs.mu.OZ.AU>
Date: Wed, 08 Aug 2001 20:14:50 +0700
In-Reply-To: "Tony Hain"'s message of "Wed, 08 Aug 2001 12:56:27 +0100."
Sender: owner-dnsop@cafax.se
Subject: Re: (ngtrans) Joint DNSEXT & NGTRANS summary


alh-ietf@tndh.net said:
  | Or accept the reality that enforcing PA as the 'only' approach is  in
  | direct conflict with the ultimate goals of the consumer.  

The ultimate goals of the consumer are surely to have a stable internet
connection that works, and allows all available services.

Or for many perhaps at an even higher level, to make lots of money however
it can be made, and care about very little else.

Very few ultimate consumers care at all about renumbering, except to the
extent that it interferes with one of the above real goals.   They care
even less about the format of DNS resource records of course.

If renumbering is forced, and that causes problems, and NAT seems to allow
those problems to be avoided, then NAT is what people will do.  Once NAT
is seen as an inappropriate solution (which it will be once people start
wanting most of their systems to be available as servers, not just clients,
for at least some protocols) then they'll look to find something else that
works.   Geographic based addresses, with their likely increased costs might
be the solution.

Of course, if we can keep on working and get renumbering to work so easily
and cleanly that it ceases to be any kind of real cost, then perhaps
enforcing PA won't be seen as being in direct conflict with anything any
more - as no-one (at the ultimate consumer level) will even notice it
happening.

That's what we should be working towards, what's more, it should be an
attainable target - there's nothing so complex about configuring an IP
address that it needs to be seen as some kind of black art, to be done
once and never repeated.  The only real problems are that with IPv4 we
allowed the IP addresses to be configured everywhere, we assumed they
were a fixture (more permanent that even a domain name, as they have
essentially no vanity value) - and that has made the update process
absurdly difficult.   We just need to make sure that everyone is aware
that the only places an IPv6 address should ever be written are in the
DNS zone files and in router configs for networks (and there, in a form
that router renumbering can update).   Anywhere else you're ever tempted
to enter an IPv6 address we need to find an alternative.

kre


Home | Date list | Subject list