[arin-ppml] Policy Proposal 2008-6: Emergency TransferPolicyfor IPv4 Addresses - Last Call

[michael.dillon at bt.com]

> OTOH, the market will be unable to supply the mega-ISPs with 
> space for long, if at all, simply because of their voracious 
> appetites.  They will be weighing the cost (and odds) of 
> acquiring space on the open market with the cost of moving to 
> IPv6, and IMHO they will, in short order, decide that the 
> latter option is cheaper.

I believe that most of the larger ISPs have already made
this assessment in 2008 and decided, as you say, that IPv6
is cheaper. In addition, spending money on IPv6 is viewed
as related to technology refresh, while buying IPv4 addresses
is viewed as yet another patch to an aging vessel. Most large
companies realize the necessity of periodic technology refresh
in order to keep pace with changing markets and their strategic
vendors.

The fact is that almost all of the vendors which don't have
IPv6 support today, sell software-based platforms which could
have IPv6 capabilities in a flash, if the vendor could see a 
good reason to support that capability. The canonical example
is the consumer internet gateway products which are almost all
implemented using embedded Linux or BSD.

> My view is that they will go to IPv6 as soon as it's cheaper 
> than staying on IPv4.  An address market which increases the 
> cost of the latter will, hopefully, pull in that date rather 
> than push it out.

Another good reason to oppose transfer policies is to delay the
date at which we are forced to go to IPv6. Most people now know
that this is inevitable, but there is still a lot of work to be
done, and paid for. Operational Support Systems all have to be
upgraded and tested. IP addresses are in all kinds of databases.
This is one area where smaller companies still have an agility
advantage over larger ones, and new entrants will soon be able
to completely ignore legacy IPv4 services altogether.

--Michael Dillon