Margaret,
>
> Not to pick on you... You are making some excellent points, and I
> want to make sure I understand them.
>
> >And, I have serious doubts that
> >they will ever scale. But, if that is the case and we are going to be
> >forbidden from seeking other solutions that involve site-local addressing and
> >renumbering then we have one hell of a pickle here.
>
> Can you explain how site-local addresses and renumbering will actually
> solve the route scaling issues in IPv6 and simultaneously avoid IPv6 NAT?
> If so, I certainly wouldn't want to forbid it...
>
Trying to develop a painless renumbering mechanism, whether using site-local
addresses or some other form of private addresses, won't solve route scaling.
By developing such a mechanism we are explicitly admitting that we can't
solve it. By developing it, we are admitting that PA addresses are the best
we can do and consequently, if we want to avoid NAT, we need to address the
problems that motivate enterprises and users to use NAT. My experience tells
me that the major drivers for NAT deployment in the IPv4 world are; address
scarcity (real or manufactured) and address stability concerns. If we can
provide stable addresses for internal communication and allow for easy re-
numbering then we can provide some help in these areas. Internal communication
can continue and internal configuration can be preserved across a renumbering
event thus addressing the stability concerns. If in fact the IPv6 address
space is large enough to remove the possibility of real scarcity then we are
left with the battle against manufactured scarcity. What drives manufactured
scarcity is the desire of ISPs to charge for address space as a proxy for band-
width usage or in some cases just because they can. If I, as small business
owner or home user, can trivially move to another provider when the cost of
global address space gets too high, I won't be tempted to insert a NAT box
between me and my provider to protect myself against the equally awful choices
of arbitrary increased cost or manual renumbering of my network. Similarly, the
ISP will have less incentive to manufacture scarcity since it will do no good.
Their customers will simply move to another provider and renumber. The other
thing which needs to happen to prevent NAT is that compelling applications that
don't work well with NAT need to be developed and deployed. We have very few
sticks to use against NAT so we need to work really hard on the carrots.
Admittedly, solving renumbering for the general case is really hard. The
hardcoding of IP addresses in all sorts of configuration files is a long and
storied tradition. It will take some serious work to tease out all these
usages and replace them with more appropriate mechanisms. We started down the
path with RFC 2894 and the A6 work. But that was really just the beginning of
the work. It may in fact not be possible to nail this problem or it could take
a long time to do it. It is also possible that market forces won't work as
described above. Dan may well be correct that oligopoly will prevent real
competition from existing and consequently ISPs will be able to manufacture
scarcity regardless how easy it is for customers to change providers. However,
if all we have and all we can hope to have are PA addresses then, if we are
committed to not reliving the IPv4 NAT experience, we better get a move on.
The alternative of rearchitecting the routing system to allow for truely PI
addressing is obviously more attractive than solving renumbering. Un-
fortunately, the problem may be no easier than solving renumbering and ISPs
that profit from the current architecture may be resistent to the solution
even if it is viable. I can't speak to the probability if success in
addressing the technological and political portions of this problem.
There may well be other alternatives to the above two paths, but what is clear
to me is that unless we get a move on with something serious, IPv6 NAT is a
certainty. It may well happen no matter what we do simply because of built up
technological inertia, but it is an absolute certainty if the status quo of
doing nothing continues.
Tim Hartrick
Mentat Inc.
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
