Why there will never be a IPv6 transition

There are many nervous announcements that we are running out of IP-addresses and that we must change this year, 2011 to the new IPv6 standard. It is correct that we are running out of IP-addresses, but we will not change to IPv6 this year - on the contrary this transition will never happen.

All current users could keep their numbers and new users just got numbers with a different area prefix. (Of course there have been small changes in which number-changes were forced, but that was the exception, usually the current users did not lose their numbers.) The result was that there was no change for the current users: The own number stayed the same and also all numbers in the phonebook stayed the same - usually the users did not even notice that new numbers were created for new users and that is exactly how it should be.

One might believe that because of such a long experience in telecommunications one might find a similar solution for IP-addresses, but that is not the case.

IPv4 and IPv6 are not compatible to each other - to take the example of telephone numbers it would be like changing from digits to letters: Nobody could keep his number.

There are about 4 billion IPv4 addresses. There are about 200 to 300 Million used domain names, which are all connected to a IPv4 address. (Added to that are subdomains, for example not just domain.com but also www.domain.com, mail.domain.com, etc.) The full count of domains including subdomains is hard to estimate, but I think 1 billion is realistic.

To introduce IPv6 successfully, all owners of websites would have to:

• Make sure, that all their servers, routers, switches, etc. support IPv6.
• Apply for an IPv6 address for every IPv4 address they currently use.
• Add all their IPv6 addresses to their DNS-entries.
• In the "transitional period" (which is now already lasting for over 10 years) every configuration change has to be done twice, because of course you cannot turn off IPv4 as long as practically everyone is using it.

Even when you are incredibly naive and optimistic and estimate that only 5 minutes are used for each domain, the total requirement is 40000 man-years.

So far the theory, in the real world, things are more complicated. For example Google writes that they cannot activate IPv6 for all their offerings, because this would lead to networking problems. When even Google fails at introducing IPv6 for all their servers, this shows that it's not just about configurating addresses, but that there are a lot of much more complicated problems.

But even all that would be somehow manageable, there are legions of network administrators in the world. The technical problems are solvable in principle, but the real problem is not technical in nature but the belief, which is bordering on megalomania, that an international, global network can be reformed, by letting all users find new addresses.

What dooms each attempt to introduce IPv6 is the chicken-egg problem:

Because almost no webservers are reachable with IPv6, there is understandably no interest by users for it. Without IPv4-address the Internet is practically useless, because apart from some exceptions no websites are available. And because there are almost no IPv6-users, there also understandably no reason for webmasters to apply for IPv6-addresses.

When you are an early adopter in the IPv6-transition, you have no advantages, on the contrary: The early adopters have to worry about buggy software, unwilling business partners, expensive hardware and untested systems. Why be an early adopter then? The rational approach is to wait that IPv6 "becomes mainstream" and only then join in. But because everybody does that, that will never happen.

The mistake in the design of IPv6 is not really in the technology. The mistake does not lie in the design of the protocol itself, but in the incompatibility of the addresses: If the addresses would have been compatible, or to put in other words that every IPv4 automatically becomes a valid IPv6 address (just like hundred years ago in the telephone networks), then IPv6 would have had a chance, but that is not the case. And because it is not a technological but a administrative problem, there will hardly be a technical solution for it.

Conclusion

In the subconcious this seems to be even understood by technophiles, because support for IPv6 has become worse rather than better in the last years. For example Cisco/Linksys still sells routers that don't understand IPv6 and open-source projects like DD-WRT have disabled IPv6 from their default install to make room for features that may actually be used. When even projects of people who like to tinker with new technology turn their back on IPv6, who can seriously believe that average companies and users will be interested in it? According to Google, IPv6-usage remains relatively flat at 0.30%.

The future is hard to predict and it will be hard to predict how long IPv4 will live, but one thing is sure: What ever will come after it, it won't be IPv6.