[arin-ppml] v4 to v6 obstacles

[Matthew Kaufman]

Owen DeLong wrote:
> If the top 50 web sites make it to IPv6 dual-stack, then, that's 
> probably enough
> for the rest of the web to decide they'll start playing catch-up, 
> frankly.  I'm not
> sure what the critical mass of web sites and other services on 
> dual-stack is for
> eye-ball ISPs to start issuing IPv6-only customers once IPv4 becomes 
> difficult,
> but, I'm betting that as IPv6 deployment grows and IPv4 depletion passes,
> the difficulty of issuing IPv4 addresses will increase and the 
> threshold of
> acceptability of IPv6-only will continue to drop until some point 
> where those
> values meet.
If the top 50 web sites are dual-stack, then dual-stack customers whose 
IPv4 comes via NAT will have happy ISPs, as the NAT doesn't need to work 
nearly as hard if the top 50 destinations don't go through it. But the 
ISPs will still need those NATs to reach the IPv4 Internet (assuming, of 
course, that we really ever run out of purchasable address space)

But my prediction is that "IPv6-only" will *never* be acceptable to 
customers except in specialized circumstances (customers who are 
attaching a specific device that is IPv6-capable that they only need to 
reach from the IPv6 Internet) or if IPv6-to-IPv4 translation is 
perfected to the point where an IPv6-only user can still reach the 
entire IPv4 Internet that still exists. (But even the latter is painful, 
as to talk to your local printer you bought last year you need IPv4 
enabled on your hosts anyway and so dual-stack from your ISP is really 
much more likely to work properly for reaching the IPv4 Internet)

Matthew Kaufman