[ppml] "Who's afraid of IPv4 address depletion? Apparently no one."

On Sat, Feb 09, 2008 at 02:36:31PM -0500, John Curran wrote:
> At 7:11 PM +0000 2/9/08, bmanning at vacation.karoshi.com wrote:
> >
> >	right model, but (imho) 180 degrees out of phase.
> >	think IPv6 only islands w/ a thin veneer of IPv4
> >	(say a /28) on the outside.  the ALG's to support
> >	this are not well defined, with the IETF reconsidering
> >	work in this area.  some call this NAT-PT, my favorite
> >	is the CERN-IVI box... which might see the light of
> >	day 3q08...
> 
> Bill -
> 
> Is it safe to presume that these are architectures proposed for
> new customer connections?  It is hard to imagine an existing
> Internet connected (via IPv4) site having any reason to evolve
> its internal network into an IPv6 island intentionally in the near
> future...
> 
> /John

	i think a safe presumption is that this may be a 
	predominant structure as long as there are "arrogant
	twits" who maintain the fiction that only IPv4 transport
	is needed to get to their content/eyeballs.  e.g. if 
	facebook never supports IPv6 transport, this will be common.
	Facebook will never see IPv6 demand and claim "all is well"
	with IPv4 and the IPv6 hype is just that.

	I expect it will be common for new builds - so they can
	participate with that "other" address family.  Less common
	for existing builds - in part because the clients/content
	are already there.  UNTIL ... some compelling new content
	or feature is brought up ONLY on IPv6... then the installed
	base has to decide how to move.

	the middle ground, that you have espoused, where some content/
	applications are available over -both- address families - will
	be the predominant mode - but such will be an operational 
	suite from outerdarkness...  because the transports are non-overlapping.


--bill