On Dec 2, 2009, at 4:03 AM, <michael.dillon@bt.com> <michael.dillon@bt.com> wrote:
We can solve this problem the same way that we solved IPv4 runout.
Heh. We haven't solved IPv4 runout. We've largely investigated interesting geometries we can rearrange the deck chairs into on the Titanic. As far as I can tell, the jury is still out as to whether IPv6 will win over IPv4 + multilayer NAT.
As long as we have IPv9 ready to deploy by 2050, we can avoid any risks of IPv6 addresses running out.
One of the reasons IPv6 has had some difficulty getting off the ground is due to the large installed base of IPv4 that sees no particular benefit in moving to IPv6. IPv4 is now a core component of critical infrastructure all over the planet. Switching out critical global infrastructure is not something that is trivially done; you need deployable transition plans, business cases, backwards compatibility, etc., all stuff that was not really addressed with IPv6 because it is really, really hard. And (assuming IPv6 does win) you're suggesting that after we've spent half a century of deploying critical infrastructure globally on IPv6 that we'll just go and redeploy IPv9? Are you assuming the Singularity will have been reached and a god-like AI will solve the problems for us? Regards, -drc