[AfriNIC-rpd] Value of N for Global Policy for the Allocation of the Remaining IPv4 Address Space

[Andrew Alston]

Hi All,

I've been having a brief look at the data on potaroo.net trying to figure
out in my mind what a sane value for N would be when considering the data
actually available with regards to current allocation trends.

I must admit, having looked at that data I'm probably even more uncertain
now than I was before, I do however believe that setting N to either 1 or 2
could cause problems for certain regions.

If I'm reading the data correctly, and if I'm not I'm open to correction,
the current allocation rates show that RIPE is allocating at a rate of
between 4 and 5 /8s per year, APNIC is sitting at between 4.5 and 5.5, ARIN
is sitting at between 1 and 2 per year, LACNIC seems to be sitting at
between 1 and 1.5 and AFRINIC is sitting between 0.3 and 0.35 per year.

Now, if N = 1 AFRINIC and LACNIC are relatively ok, we have some time to
play with to give people SOME time to adjust to the fact that there is now
probably a soft landing policy in place etc etc.  APNIC and RIPE could have
as little as 2 months, maybe less (in June, RIPE NCC allocated 0.75 /8s), I
see this as being relatively problematic for them!

Its kind of hard to model, because of how variable the allocation is, and
how much it fluctuates, but I'm leaning towards believing that if this
policy is to be sensible, it needs to allocate the remaining space such that
the biggest RIR has at least 6 months at time of implementation of policy.

Now, considering the growth in allocation rates of late, and due to the fact
that it is entirely possible that there is an explosion in demand once the
space depletes to these levels, I'd have to say that N should be at least 3,
if not 4.

I believe very strongly that when AfriNIC debates the value of N in this, we
need to be very very careful to take the global view into account on this,
we cannot debate this from the perspective of purely African merits, as at
the end of the day, what hurts one part of the Internet will end up in the
long run hurting us as well, hence my belief that you have to model N around
the largest of the RIR's and around growing demand.

Anyway, those are my thoughts.

Regards,

Andrew Alston
TENET - Chief Technology Officer