Re: Re: IPv6 only devices in home network

Keith Moore <[email protected]> Fri, 11 Oct 2002 12:09:16 -0400
Newsgroups gmane.ietf.ngtrans
Message-ID <[email protected]>
> > No. NAT (including NAT-PT) breaks applications. a big
> > reason to use IPv6 is to avoid having your applications
> > broken by NAT.
> 
> I suspect this is not a real problem. By the time ISPs
> are assigning their clients IPv6 addresses rather than
> public IPv4 addresses virtually all public servers will
> be ready to accept IPv6 connections.

I don't buy it for a millisecond.  Why in the world do people think
that all ISPs are going to migrate to IPv6 in lock step when the
evidence overwhelmingly contradicts that?  For that matter, why do 
people assume that all users use the same applications and have the
same needs, or that all applications have 'clients' and 'servers'?

> To whatever extent the ISPs lead the servers, however,
> you are ultimately going to have to rely on NAT. 

NATs will have to be used in corner cases.  They are not, and cannot
be, general solutions.

> While neither is desirable, it is far more realistic to
> have application developers work in NAT compatible ways
> than to get all NAT boxes removed.

that's a ridiculous statement.  what is realistic for some applications
is completely infeasible for others.

> Do you really think there is a large body of code that
> is IPv6 ready that is *not* NAT tolerant?

yes, there are many such applications.  and IPv6 will allow more 
of those applications to exist.

Keith