[arin-ppml] Change of Use and ARIN (was: Re: AFRINIC And The Stability Of The Internet Number Registry System)

Owen DeLong owen at delong.com
Sat Sep 4 19:46:41 EDT 2021


>> " Just curious Mike... Does this opinion on your part extend also to IPv6?"
>> 
>> Of course not. IPv6 is not a scarce resource and there is no market for it
>> to compete with the free pool.
> 
> IPv6 isn't scarce -today-.  If we look out 25 years it may become so.
> So then what?

One of two things needs to occur for this statement to work:

	1.	IPv6 allocation policies need to radically change
or	2.	You have to be really bad at math.

So far, I believe 2 RIRs have received their second /12s from the first /3,
so the total IPv6 distribution so far stands at 7 /12s + some earlier more
specific RIR allocations from IANA which, IIRC, taken together add up to less
than a /12, so let’s call it 8 /12s.

THat’s lasted us 20+ years of IPv6 already. RIRs seem to be burning through
roughly 1 /12 (total) every 5 years or so at current allocation rates. Let’s
take that and assume that the IPv6 allocation rate will double every 5 years
for the next 25 (very unlikely, but it’s OK.).

So we’re at 8 /12s now, and by 2026, we’d be at 10 /12s issued. To compute 2031,
we double to 4 /12s issued bringing the total to 14. By 2036, add 8 more for
a total of 22 and 2041 adds 16 bringing the total to 38. Finally at the 25
year mark, we’ll add another 32 bringing the total to 70 /12s issued.

The initial /3 IETF has put into play as IPv6 GUA contains a total of 512 /12s,
so in 25 years, even at the outlandish growth rate I posit above, we’ll still
be just slightly past 10% of the first 1/8th of the total address space allocated.

In short, absent some radical change in how addresses are used and allocated,
it is unlikely that:
	1.	Anyone alive will survive to see IPv6 scarcity
	2.	Address scarcity will be the first fatal scaling limit seen
		in IPv6.

> This seems to be what Lu Heng and others want, and strange as it may sound
> coming from me, nobody has ever persuaded me that this would actually be a
> Bad Idea... and I gather that you also would tend to view this idea favorably.

I don’t think that’s a fair characterization at all.

Owen




More information about the ARIN-PPML mailing list