Re: FW: I-D ACTION:draft-ietf-v6ops-ent-scenarios-03.txt

[Tim Chown <tjc@ecs.soton.ac.uk>]

On Fri, Jul 02, 2004 at 10:14:50AM -0400, Margaret Wasserman wrote:
> 
> Hi Tim,
> 
> My preferred "solution" to this problem is to avoid the situation.  Nodes
> that  need to speak to IPv4-only nodes must implement IPv4, and nodes
> that need to speak to IPv6-only nodes must implement IPv6.  Nodes
> that need to talk to both IPv4-only and IPv6-only nodes must be dual
> stack.  An IPv4-only  node would not be able to communicate directly
> with an IPv6-only node.

This is the IETF mantra, and I agree with it.  The question is how reasonable
will it be?   In some places, in order to run dual-stack it'll have to be
IPv4 private addressing and IPv6 public addressing together.   Perhaps some
(mainly greenfield) networks will just decide IPv4 isn't worth deploying
and will offer translation (probably ALGs) for legacy applications like mail, 
web and FTP and rely on IPv6 for all new or intranet applications?   Do
you think this would be unreasonable, especially in less well IPv4-endowed
areas?
 
There are actually many academic sites that rely on ALGs, e.g. student
dormitory networks, which might only have web ports open outbound, and
use private IPv4 addressing.  These could be candidates for IPv6 deployment
experiments, although in many cases the restriction is of course not because
of IPv4 address shortage, but policy decisions.

Tim