One Year On: IPv4 Exhaust

Owen DeLong owen at
Sun Sep 25 22:28:08 UTC 2016

> On Sep 25, 2016, at 10:19 AM, Paul Thornton <paul at> wrote:
> On 25/09/2016 01:54, Jay R. Ashworth wrote:
>> One year ago today, at 12:36pm EDT, Facebook On This Day reminds me, John
>> Curran announced that the last IPv4 address block in ARIN's Free Pool had
>> been assigned.
>> How's that been workin' out for everyone?
> If you'll all indulge a bit of a RIPE-centric reply on this; I've was allocated a /22 from around half-way through last week (185 being RIPE's final /8).
> Assuming that RIPE are allocating sequentially - and I believe they are - This means that they have consumed around 66.5% of their final /8. They started allocating from this in September 2012, which suggests a reasonably low consumption rate but the RIPE final /8 will be exhausted in around two years time.
> I can't find an equivalent ARIN page of "how much we've allocated from our last /8" - the statistics show that just over 2x /16s worth have been assigned/allocated between January 2016 and July 2016, so a lower rate by some margin than RIPE - but there are of course policy differences at play there.

The reason you can’t find such a thing is because ARIN doesn’t have a last /8 policy, per se, like RIPE and APNIC. Instead, ARIN set aside blocks well before the last /8 for critical infrastructure (Key high-level name servers, IXPs, etc.) and IPv6 transition. The IPv6 transition space has a pretty limited set of valid use cases as does the critical infrastructure block, so ARIN is probably allocating those relatively slowly, but they aren’t coming from the “last /8”, to the best of my knowledge. The last /8 was allocated business as usual from the free pool and may well have provided the last allocation from the “virgin free pool” (as opposed to reclaimed blocks).

> Now the operational question of "How has this affected us" is probably best answered with "We've had to pay real money for IPv4 addresses since then."  What may be much more interesting is what happens when the fairly ready supply of IPv4 addresses in the secondary transfer market starts to dry up.  Just throwing additional money at the problem will probably not be an effective or viable solution then.

IMHO, sane organizations see this writing on the walls and are deploying IPv6 at an increasing rate. If people act at a responsible pace, they should be able to get IPv6 deployed before we run out of readily available secondary market supply. If not, then, well,
it’s not like they didn’t have 20+ years warning so I don’t exactly feel a great deal of sympathy for their self-inflicted wound(s).

> I'm sure that Geoff Huston has a much more accurate and colourful set of predictions than my back-of-envelope calculations for those interested!


IPv6 is the present. IPv4 is the past. The sooner we get more networks to regard the world in this way, the quicker life gets better for everyone.


More information about the NANOG mailing list