[Community-Discuss] Exhaustion of IPv4 addresses, hypothesis or reality

[Owen DeLong]

> On Aug 11, 2017, at 05:11 , Kangamutima zabika Christophe <funga.roho at yandex.com> wrote:
> 
> While Afrinic experts across Africa to predict IPV6's good word and good practices to break down public IPV4 addresses, most of the West African embassies operating in African countries use The principle of 1PC for a public IP. This means that each workstation operating on the local network is directly connected to the internet through its public address (not even dynamic NAT with a public IP pool).
> This is to ask questions if the reasons for migration to Ipv6 apply or will only be for a specific category of Internet users. In my opinion, at IANA level, it should be advisable to make an inventory of used IPv4 addresses and actually used. Since there are apparently many entities that have sold a war treasure in terms of public IVP4 revolving or transferable from one area to another but not yet used.

This is a red herring. As the IANA free pool neared its end, APNIC alone was going through approximately 1.5 /8s per month, in addition to the approximately 0.75 /8s per month in the RIPE region and 0.5 /8s per month being consumed in the ARIN region at the time plus miscellaneous consumption in the LACNIC and AfriNIC regions.

The most optimistic estimate I’ve ever heard for reclaimable (unused/fallow) IPv4 addresses was 24 /8s worth. Let’s round that up to 32 /8s to make the math easy and make sure we are overly conservative. Now, let’s compare the amount of usable lifetime IPv4 gets from reclaiming all those addresses to the time and manpower it would take to do so…

First, the most optimistic estimate for reclamation I ever saw was ~80% over the course of about 5 years, assuming complete cooperation, no litigation, and rapid, painless, unambiguous identification (each and every one of which is utterly impossible in the real world). 25.6 /8s. With a burn rate north of 2.75 /8s per month, that’s a life extension of approximately 9 months.

Under the above overly optimistic assumptions, the minimum (again most optimistic) estimate was that each RIR would need a minimum of 10 FTE working the process to achieve those numbers (as if they were even possible). Let’s assume we only really need these efforts in the larger 3 RIRs and leave LACNIC and AfriNIC alone for the moment. That’s 30FTE * 5 years = 150 FTE Years at 2000 hours per year = 300,000 Man Hours.

Each and every person who has looked at this in any depth has come to the conclusion that there’s a lot more to be gained from those 300,000 man hours going into deploying IPv6 than could possibly be achieved by sinking them into reclamation efforts.

Owen