The IPv4 Exhaustion Wormhole

I’m starting to wonder if IPv4 exhaustion has opened a wormhole plunging us all back to 2008 again.

Australian researchers have begun collating information from networks across the world in an attempt to figure out how many usable blocks of Internet Protocol version 4 (IPv4) addresses remain unused.

This would have been a good wake-up call project in 2008.

 The study had already verified a couple of hundred million IP addresses that were currently not used – but could be reallocated…

They appear not to have considered burn rate. Around April 2010, APNIC had 13 /8 IPv4 blocks left – about 218 million addresses – and one year later they were gone. That’s just in the Asia-Pacific region. Worldwide, a ‘couple of hundred million IP addresses’ would have negligible impact on IPv4 depletion. (If they were even recoverable, given corporate self-interest.)

…incumbent ISPs with plenty of IPv4 addresses had market power because of the overall scarcity of addresses and therefore little incentive to lead the migration to IPv6.

Any incumbent trying to blackmail their customers into paying more because IPv4 is so ‘scarce’ will probably see those customers exit, stage right, for real ISPs who know what they’re doing, like the market-dominant gorillas AT&T, Verizon Wireless and Comcast.

Very little content is available on IPv6 and very few users [can access it] so if you just gave content providers an IPv6 address without having IPv4, they’re not going to have much custom…

Yes, that was the situation in, oh, 2008. Today Google alone takes the IPv6 content available on the Internet to around 50%. Facebook 46%, YouTube 33%: even with overlap that’s probably 80% of the most sought-after content on the planet, and it’s there on IPv6. (And red-herring alert: no-one currently advises giving out IPv6 addresses only. Dual-stacking is the new black.)

My apologies to the researchers if they’ve been horribly misquoted. But if they haven’t, I’m reminded of prospectors panning old mine tailings, desperately seeking grains of gold. And right beside them, gleaming benignly in the sunlight, is an Everest of precious ore just there for the taking.


The IPv4 Exhaustion Wormhole — 4 Comments

  1. Thanks, Kate, for taking an interest in our research.

    This is just one part of our research into the transition to IPv6, which is being carried out in conjunction with APNIC.

    Just to clarify, we are not advocating “panning for addresses”. However we acknowledge that some companies are likely to try it, and want to understand the process. For example:
    – how will the fragmentation of prefixes affect global routing table sizes?
    – Should ICANN take action to prevent a repeat of the Nortel address sale, or will such sales be inevitably small enough to be harmless?

    For more details, see our web site:

  2. Hi Lachlan, your research site suggests a lot more behind the project than was reported through the journalist’s filter, so thank you for the link. Certainly, address fragmentation is always an issue, the NAT estimates will be interesting, and estimating energy use is excellent stuff. I guess I was amazed at the reported statements on the numbers of recoverable addresses (with the implication they were significant in global terms), and there being little IPv6 content on the Internet, which today is simply not correct.

    And I’m not convinced the time frame for IPv6 deployment has much to do with the amount of IPv4 remaining. If the relationship was coupled then we’d have seen people taking up v6 as v4 stocks fell. The fact we (sadly) didn’t suggests to me they have different incentives.

    Still it will be fascinating to see what happens over the next few years, and I’ll look forward to seeing results from this project.

    Regards, Kate

  3. Hi Kate, I’m one of the people involved in the STING project. I very much agree with you on this point (assuming you meant remaining unallocated):

    “And I’m not convinced the time frame for IPv6 deployment has much to do with the amount of IPv4 remaining. If the relationship was coupled then we’d have seen people taking up v6 as v4 stocks fell.”

    Clearly, if the existing players still have larger numbers of unused IPv4 addresses there is no immediate pressure for them to move to IPv6. The runout of allocated addresses only puts pressure on new players, but I think this effect is limited. However, I think once existing players will run out of unused IPv4 addresses, they will come under pressure.

    I’m not convinced at all that the high “burn rate” of allocated space (APNIC) translates into the same high burn rate of unused space. I think we may have seen at least somewhat of a “buffet effect” or “forward demand effect” or whatever you like to call it.

    So we’re not only looking into used IPv4 addresses, to understand the possibility for “panning”, but to get a better understanding of the whole transition process! As you said, the number of unallocated IPv4s is probably not a good indicator. I think the number of used IPv4s is a better indicator, but unfortunately it’s also harder to estimate.

  4. Hi Sebastian, thanks for explaining more about the STING project, and yes, understanding this kind of transition is important, especially for dealing with future events. I guess after experiencing years of IPv6 denialism I assumed this was part of the same thinking: just find enough of those nuggets of v4 and the need to face v6 reality can be postponed even longer.

    Perhaps as part of the project you should involve colleagues in the Psychology field as well, because logic is clearly not driving the transition to v6 – something a lot more irrational seems to be in play. I wrote that as a joke, but now I’m not sure it is!

    Anyway, good luck with the project and keep us informed about its progress.
    Regards, Kate

Leave a Reply

Your email address will not be published. Required fields are marked *