August 2009 Archives

Our local area code (814) is running out of phone numbers. When discussing IPv6 with non-technical folks, I frequently use the hypothetical scenario of running out of phone numbers as an analogy for IPv4 address depletion. The conversation usually goes like this:

Imagine if we were running out of phone numbers. One way of solving that problem would be to make them bigger. Instead of ten digits, what if we made them thirty digits? If we did that, how many other things would we have to change? Some mundane things like business cards, letterhead, and phone books. But also more substantial things, like form processing software, backend provisioning software, and legal intercept software. All of that would take years to design, test, and deploy.

(By the way, we're not running out of phone numbers. The NANP projects that the US is fine through at least 2039.)

All analogies break down at a certain point. Technically, the phone number analogy isn't accurate, but it's a reasonable way to explain to my parents what I do at work. Technical details aside, there's (at least) one significant difference between running out of phone numbers and running out of IPv4 addresses: People don't deny that we'll run out of phone numbers.. As I said above, we're in no danger of doing so right now, but the day will come when our population will grow to a point that it will exhaust the telephone numbering plan. That day is far away, but when it comes, we'll have to start adding digits.

On the other hand, some people do deny that we're running out of IPv4 addresses. I don't understand this. The data is unequivocal: We are running out. Fortunately, over the past few years, the data became more clear, and most of the deniers have changed their opinions.

Still, we've never had people advocate for a telephone equivalent of NAT. We've never heard people claim that it's a security threat to have their phone "exposed to the public telephone network." When ten-digit-dialing was introduced years ago, some people complained about the hassle, but speed dial and phone address books solved that problem. When it's noticed at all, ten-digit-dialing is seen as a technological impact of population growth, whereas IPv6 is often seen as some sort of personal assault on the sysadmin asked to deploy it.

I am very concerned that the Internet community has waited too long to begin serious efforts for IPv6 deployment. Only 10% of the IPv4 address pool is left. It will probably be gone 2.5 years. Yet only 5% of the Internet supports IPv6 (as measured by BGP announcements). I just don't see a way to get from 5% to 100% before we run out of IPv4. The next few years will be interesting, to say the least.

Two more IPv4 address blocks have been allocated. Both went to the Asia-Pacific region, as did the two blocks that were allocated in April.

Only 10.9% of the IPv4 address space remains unallocated:


If you exclude the blocks reserved by the "n=1" proposal, only 8.9% of the IPv4 pool is left.

Folks, it's way past time for IPv6 trials. It's time to get serious. How much more grim do the data have to get before we start taking action?