Recently, it was reported that the last unallocated top-level blocks of Internet addresses have been allocated. The top-level blocks, which contain about 17 million addresses each, are distributed on a continent-by-continent basis, and there is still some time time before they are broken into smaller blocks and allocated for use. But sooner or later the pool will be exhausted.

The solution is to replace the current IPv4 protocol (which allows for only 4 billion Internet addresses) with IPv6 (which allows for 3.4 undecillion, aka a whole lot). The problem is, software that works for IPv4 does not automatically work for IPv6 as well: it has to be upgraded to support IPv6. Much of the software already has been upgraded. All major operating systems support it. Your web browser supports it (unless you’re using Lynx or NCSA Mosaic). But a lot remains to be converted. This will be the greatest economic cost of switching to IPv6.

Here’s how I expect the transition to go:

1. Pre-exhaustion stage. Unallocated IPv4 addresses for end-users remain. A few consumers, companies, and governments start to adopt IPv6, but most behavior remains unchanged. Toward the end of this time, markets for IPv4 addresses will appear. Some people will grab unallocated blocks to sell at a marked-up price later.
2. IPv4 market stage. The last IPv4 addresses have been allocated. People who control sizeable blocks will sell them for a quick buck. This will extend IPv4 for longer than people expect. (Most articles I’ve read predict it will extend IPv4 by a few months; I say it’ll be a year at least. Never underestimate raw capitalism.)
3. IPv6 hype stage. As IPv4 market prices continue to rise, companies will start to hype up the advantages of IPv6 [1] in an attempt to get consumers to change, and to make a buck in the process. Comcast will create ads that say, “Get our new-fangled IPv6-enabled modem!”. Linksys will come out with routers, “Now with IPv6!”. They will both quietly offer free flash upgrades to reward their smarter, and thus better, customers. But the end result, shady though the process be, is consumers will start to use IPv6 en masse.
4. IPv6 critical mass stage. When a critical mass of consumers, infrastructure, and content is converted to IPv6, some providers will decide not to bother with IPv4, and some knowledgeable consumers will disable IPv4. New software will start to be developed without IPv4 support. Old software will start to drop IPv4. I’m guessing this won’t happen till 2014.
5. Legacy IPv4 stage. IPv4 prices peak. Large corporations and goverments that have money and a need for maximal exposure will start buying out the ipv4 addresses so that legacy users can still connect to them. Almost no new software will support IPv4. Legacy software that only supports IPv4 will fall into disuse.
6. IPv4 wind-down phase. IPv4 blocks fall to near zero value. IANA and friends will start asking companies to return IPv4 blocks for permanent retirement.
7. Microsoft will drop support for IPv4, finally making it obsolete.
8. Top-level domain servers stop serving IPv4 addresses. IPv4 will survive only in off-line legacy networks, and Gophernet. Guessing this will be about 2020.

We’ll check back here in 2020 to see how my predictions went.

Currently, I haven’t even begun to adopt IPv6. This is mainly because my ISP doesn’t support it yet, so what’s the point? I actually have my ipv6 module blacklisted so it doesn’t load. (I run Linux so I can do stuff like that.) As soon as my ISP does that (and they should do it pretty soon), I’ll start my own conversion.

Footnotes:

[1] IPv6 has other advantages besides the larger address space.