Date:        Fri, 01 Sep 2000 17:00:39 -0400
    From:        Ken Hornstein <kenh@cmf.nrl.navy.mil>
    Message-ID:  <200009012100.e81L0eU17536@ginger.cmf.nrl.navy.mil>

  | I can't even BELIEVE I'm answering this .... buuuutttt ...

Me too...

  | Even the mobile IP people are not suggesting that applications and
  | protocols be designed to handle renumbering "on the fly".

Actually, it would be nice - but no-one rational is expecting that to
happen any time soon.   Whatever the pain of upgrading to IPv6 might be
for the net, the pain of redefining a whole bunch of application protocol=
s
and then distributing new versions of those around (and getting people
to upgrade) is much much worse.

And that's even assuming it can be done - the "stick a DNS name in instea=
d
of an address" is na=EFve at best, the cans of worms that opens has to be=

seen to be believed (like how it relates to systems like dial ups, which
get an address for the 20 minutes they're connected, and expect to live
in a DNS world where the negative cache interval is an hour or so).

Sean's arguments can be amusing at times, even to the point of actually
requiring some thought - this one was a weak effort.  Supposedly NAT is
the solution to the address problem, so IPv6 isn't needed.   But NAT has
all of these problems, so we carefully define NAT to be exactly what
doesn't have the worst of the problems (and require the rest of the world=

to adjust itself to fit the model then presented) - while carefully forge=
tting
that the NAT now defined doesn't actually solve the addressing problems
any more (providing no more than a trivial, perhaps 2* actual address
space increase - and probably less than that).

It is truly amazing watching anyone attempt to claim that NAT is more tha=
n
a temporary quick hack fix "solution" to anything at all.  The one thing =
it
seems to make easier is external address renumbering (renumbering the
externally visible address of the systems at a site) until you notice tha=
t
the truly hard part of any renumbering is handling all the places where t=
he
number is known off site (other people's filter lists, other people's DNS=

configurations, ...) for which NAT gives you absolutely no assistance.

kre