Are we really ready for IPv6?

I would love to believe the Number Resource Organization (NRO), the group that oversees the allocation of all Internet numeric resources, claim that 84% of ISPs and other major Internet groups are now using, or are about to use IPv6. But, I don't.

Mind you, 84% of serious corporate Internet users should be at that point. We're quickly running out of IPv4 addresses. Let me check right now. Yep, according to the IPv4 Address Report, we're still on schedule to run out of all IPv4 Internet addresses on January 14, 2012.

I actually expect us to run out faster than that though. I strongly suspect speculators are already snatching IP addresses up the same way they did potentially popular domain names and for the same reason: to make money from selling them off in a seller's market. In fact, I just noticed that in the week since I last checked the Internet IPv4's drop-dead date, it's already moved up from January 25th 2012. The buy-outs are already happening.

I don't buy the most optimistic view of the IPv6 Deployment Survey (PDF Link) for several reasons. First, it's a global, self-selected survey of the people and groups who work closely with regional Internet registries (RIRs) the group that issues Internet addresses to ISPs. In short, they're the people who are closest to the coming Internet address famine. If they don't get it, no one does.

My other problem is that I've looked closely at the survey's results. They're not as rosy as they might first appear. Almost half of those surveyed reported that their biggest problem with IPv6 deployment is a lack of user demand while over 40% reported that they actually have any experience with putting IPv6 to work.

This sounds like a lot of groups and people have good intentions about using IPv6, but have actually done very little with it. This is not good. We all know that good intentions are what the road to hell is made from right?

I also couldn't help noticing that the vast majority of those who said that they're using IPv6 are doing so with dual IPv4/IPv6 stacks. That's fair enough. It's how I expect almost everyone will deploy IPv6 until we're in the 2020s. But, at the same time, these same users report that they're using address translation technologies like NAT (Network Address Translation) for their IPv6. Ah... I don't think so.

This suggests to me that even this 'expert' group is a little unclear of what they're doing. You can use NAT for IPv6, but I don't know anyone who knows it well who recommends it. Its one advantage is that it's simple to deploy. This technology, NAT-PT (Network Address Translator-Protocol Translator), simply doesn't work that well in practice.

I like to think of NAT-PT as trying to use a band-aid for a severed leg. Sure, it's quick to put on, but there's a host of problems that comes with it. These include trouble with timeouts, keep-alives, and DNS Security Extensions (DNSSEC) incompatibilities. Oh, and did I mention headaches with authentication and encryption? NAT-PT, and other attempts to paste IPv4's NAT approach on top of IPv6, are destined for failure.

Finally, the survey reported, not to any surprise of mine, that end-users and customers still aren't using IPv6. I know businesses. Without a pressing reason staring them in the face, company executives aren't going to invest in IPv6. The day is coming when they are going to have to see that they'll need to address this Internet issue, but what I picked up from this survey is that even those who are living next door to the problem are still giving it more lip-service than real study and work.

Darn it.