[arin-ppml] What do you think of 2011-1 (now in Last Call)?

[Owen DeLong]

> -- I *also* recognize that there are great disparities in the allocation
>   of IPv4 address space relative to regional populations.
> 
>   When I last did the math, I saw (sorry if you're using non-monospace
>   fonts):
> 
>   Region      Population         IPv4 /8's        Ratio of %'s
>   Asia        4,121,097 60.3%    44.24 31.6%      0.54
>   Africa      1,009,893 14.7%     2.29  1.6%      0.112
>   Europe        732,206 10.7%    31.42 22.5%      2.103
>   L America     528,418  8.5%     6.28  4.5%      0.529
>   N America     348,360  5.1%    27.65 19.8%      3.882
>   Oceania        35,387  0.5%    (part of APNIC)  
>   Total:      6,829,360         139.85
> 
>   [Population in thousands, mid year 2009 estimates from the UN; address
>   usage from www.nro.net/statistics (Sept 2011).]
> 
>   If everything was proportionate, you'd see a 1.0 ratio for each
>   region. Obviously, the actual ratios aren't very close to those values.

>   In considering that table, note that while APNIC has one of the lower
>   ratios (0.54), it actually already has the largest number of reported 
>   IPv4 /8's
> 

When you consider that APNIC has more than 4 times the population of
the next most populous region, the fact that they have less than twice
the address space held by north america seems pretty irrelevant to the
argument. Especially when you consider that they have more than 10x
the population of North America. In fact, they have 60.8% of the population
(almost 2/3rds of the world population) and less than 1/3rd of the address
space, while North America is approximately 1/20th of the world population
and holds almost 1/5th of the address space.

The numbers above make a very good case for 2011-1 IMHO.

>   But also note that Africa, with the second largest tranche of the world's
>   population, has only a little over *two* /8's, or roughly 11% of the
>   number of /8's one might expect.
> 
> -- Thus, my overarching concern would be that if additional IPv4 resources
>   are transfered out of the ARIN region *now*, in an attempt to meet the 
>   short term demand for additional resources in the APNIC region (which,
>   recall, already has over 44 /8's), there will quickly be no IPv4 
>   resources available for transfer to the AFRINIC region in the 
>   *intermediate* term (when that region finally is able to participate 
>   more fully in the Internet community as a result of 
>   just-recently-improving transoceanic fiber availability).
> 

I think this is a red herring. If there is a continuing transfer market by
the time AfriNIC runs out, there will be addresses available to them in
that market. If there is not, then, keeping the addresses away from users
in the APNIC region will not in any way help AfriNIC at that time.

> -- I also understand and sympathize with the desire to "end the suffering"
>   and just "finish using up" what IPv4 address space remains, thereby
>   forcing the reality of IPv6 upon a reluctant audience, however I'm
>   not willing to force a precipitous exhaustion scheme just to "get it
>   finally over-and-done-with." Suicide pacts don't make for good public
>   policy.
> 

This isn't about a suicide pact. This is, however, about the damaging
consequences of asynchronous runout. Geoff Huston expresses these
much better than I can, so, I will refer you to his presentations from Busan
and Philadelphia RIR meetings.

Owen