RE: stable addressing

[jnc@mercury.lcs.mit.edu (Noel Chiappa)]

    > From: "Fleischman, Eric" <eric.fleischman@boeing.com>

    > the very large schism between the society in which I operate (Fortune
    > 1000 companies) and the assumptions of the IETF leadership. Back in
    > 1993 I described to the IETF the view from my knothole .. in RFC 1687
    > .. Our 1993 view has not materially changed in the intervening decade

Thanks for reminding us of that. Reading it again was still educational.


    > we will *own* our own IP addresses.

Just out of curiousity, when you move an installation, do you demand to take
your street address with you, too?

Look, I understand that you have real issues and operational requirements
from your networking infrastructure. However, it would be useful if you could
explain what those are at a high level, rather than resort to what some
people (e.g. me) perceive to be inflammatory language (e.g. the above) about
engineering details (which are somewhat in the nature of symptoms, not
causes).

(I should point out that if small businesses - which have the *exact* same
characteristic that you point out in RFC 1687 - that their *business* is
usually something other than networking/computers - took the same attitude as
you display above, we'd be stuck. Fortune 1000 companies may we worthy of
portable addresses, but not every small company can have them *if the network
is going to work*.)

E.g. when you said: "Readdressing 300,000+ devices is not trivial even using
dynamic addressing -- it's just not going to happen for something as
unimportant as switching ISPs" that was more useful - although still not what
I'm really after. What I'm looking for are things like "we don't want to have
to reconfigure individual machines when we switch ISP's, because we have so
many, it would be an O+M nightmare".


So, I imagine that you have issues with security (e.g. you don't want people
to be able to figure out your internal topology), configuration, etc.
But it would be really helpful if we could find out more about what the roots
of your concerns are ("we can't reconigure 300,000+ machines"), rather than
the eventual impact given a particular design ("we own our addresses").

E.g. I'd like to hear more about what you can configure in each machine (can
you do any configuration at all, and if so, how often), what you feel you
need to configure (a machine name, a business unit connection, a physical
network location, whatever). I think people are starting to accept that the
architecture does not have enough name-spaces, but when it comes to figuring
out which ones you should have, and *what characteristics they should have*,
that kind of insight into *what the users need and can accept* is very
important.

Or, to use another example, your concerns ("we want to hide our networking")
rather than what you perceive as what you need to do it ("give us IPv6
NAT's").


Tell us what high-level goal you're trying to meet - not what low-level
engineering feature you're currently using to do it.

	Noel