[ppml] Policy Proposal 2007-16: Ipv4 Soft Landing - a simulation analysis

[Geoff Huston]

I've performed some analysis on this policy proposal, attempting to 
understand to what extent adoption of this proposal would materially 
alter the consumption of the remaining IPv4 unallocated address pool.

The original model I've used here is the model documented at 
http://ipv4.potaroo.net. This model is based on an order 2 polynomial 
extrapolation of the advertised address count, As noted on that page the 
current projections using that model are the exhaustion of the IANA Ipv4 
address pool as of 1 June 2011 and the exhaustion of the first RIR's 
unallocated address pool on the 8th August 2012.

There are a lot of assumptions in this model, again as noted on the web 
page, but question of interest here is what would be these two dates if 
ALL the RIRs were to adopt this policy proposal.

I've adopted a conservative approach to the simulation of this policy 
proposal, simplifying the proposal to be a utilization level of 80% for 
allocations until IANA reaches 25 /8s, when the utilization level is set 
of 85%, and resetting this to 90% when the IANA pool reaches 10 /8s.

The first step has been to  generate the following table, which says 
"what would be the net reduction in total allocated addresses if the 0.8 
address utilisation factor was changed to a different factor value"

Factor   afrinic    apnic     arin   lacnic  ripencc
0.80       1.000    1.000    1.000    1.000    1.000
0.81       0.994    0.994    0.993    0.994    0.993
0.82       0.987    0.988    0.988    0.989    0.988
0.83       0.981    0.982    0.981    0.982    0.981
0.84       0.973    0.975    0.975    0.976    0.974
0.85       0.970    0.969    0.969    0.968    0.969
0.86       0.961    0.963    0.963    0.961    0.963
0.87       0.956    0.956    0.956    0.957    0.957
0.88       0.950    0.950    0.951    0.949    0.950
0.89       0.943    0.944    0.943    0.945    0.943
0.90       0.935    0.938    0.937    0.936    0.937
0.91       0.931    0.932    0.931    0.934    0.932
0.92       0.922    0.923    0.924    0.927    0.925
0.93       0.921    0.920    0.920    0.918    0.920
0.94       0.913    0.911    0.912    0.914    0.913
0.95       0.903    0.906    0.907    0.904    0.906
0.96       0.905    0.900    0.900    0.901    0.901
0.97       0.897    0.894    0.892    0.895    0.894
0.98       0.890    0.885    0.886    0.889    0.888
0.99       0.882    0.883    0.882    0.878    0.881
1.00       0.877    0.875    0.875    0.876    0.875

Taking the values for 0.85 and 0.90 and plugging them in to the 
projection model as per the policy proposal results in the following:

IANA exhaustion date: 8 July 2011 (was 1 June 2011)

First RIR to exhaust its unallocated address pool: 17 October 2012 (was 
8 August 2012)

Like all simulations there are a lot of assumptions at play here, and 
I've had to make a number of simplifications in modelling the policy as 
proposed, but I trust that this result gives the ARIN PPML folk some 
quantification of the impact of this proposal on the projected 
consumption rate of the unallocated IPv4 address pools.

Thanks,

    Geoff Husotn
    APNIC