I believe that we'll see ISP NAT for customers before we see widespread adoption of IPv6 in the USA. The dynamics are all backwards for encouraging people to switch. Web site operators (and other content providers) will, in general, pay extra to have an externally reachable IPv4 address, in order to stay accessible to the ~3b nodes already on IPv4. Most home users don't know or care about IP addresses, so they're not likely to go through the trouble to fix their systems to talk to the set of systems that are reachable via IPv6, especially so when considering that the most popular sites are available on IPv4 anyway.
For these common users, being able to send unsolicited traffic to their nodes is actually a misfeature (SQL slammer from a few years ago comes to mind). I've always thought that an ISP could sell a NATted connection at a premium as a security feature, and the cost to deploy a solution like that would be a small fraction of the cost of engineering IPv6 into a network infrastructure if it wasn't designed with it in mind initially.
Perhaps my opinion is dissenting, but I'm not sure we'll ever see IPv6 (except in that alternate universe of Japan, where everything is way more awesome anyway). I have a feeling we'll keep piling on the hacks until a paradigm shift of some type.
> I've always thought that an ISP could sell a NATted connection at a premium as a security feature
The problem is all the new aged people who use P2P file sharing, multiplayer console games, or any other internet application that relies on at least one or two inbound ports being open to listen on. An ISP couldn't possibly NAT-forward ports to individual users, but an end-user can easily decide to NAT-forward ports to their game console / desktop / laptop...
The good thing is that that population generally coincides with the people who use the most recent fancy gadgets with working IPv6 support.
The danger here is the legacy software in the hands of grandparents who can't meaningfully upgrade without huge personal costs. Those are the folks well-served by an ISP-managed proxy solution.
Along with tens of millions of lines of C code with 4-bytes-per-IP hardcoded into them, many on devices that can't be upgraded without walking up to them in person.
Uh oh. It looks like sometime in the next 3 years, I'll have to change my kernel configuration to include IPv6, and add ipv6 to my USE flags. The horror!
I know I'm trivializing, because I'm sure there are a lot of computers out there that don't support it, including some important routing devices. I just couldn't help pointing out an instance where Linux users definitely have it easier.
This is scary... but only because just yesterday the security logs on one of our boxes indicated an attempt to brute-force SSH and the IP resolved to...
For these common users, being able to send unsolicited traffic to their nodes is actually a misfeature (SQL slammer from a few years ago comes to mind). I've always thought that an ISP could sell a NATted connection at a premium as a security feature, and the cost to deploy a solution like that would be a small fraction of the cost of engineering IPv6 into a network infrastructure if it wasn't designed with it in mind initially.
Perhaps my opinion is dissenting, but I'm not sure we'll ever see IPv6 (except in that alternate universe of Japan, where everything is way more awesome anyway). I have a feeling we'll keep piling on the hacks until a paradigm shift of some type.