[AfriNIC-rpd] Last Call: IPv4 Soft Landing Proposal

[McTim]

All,

I'm going on safari to USA soon, and will be unable to participate in
policy discussions until August, so here is my preemptive reaction to
the upcoming last call:

I must object quite strenuously to this policy and to it's being put
to Last Call on both policy and process grounds.


Policy objections:

1. It aims to promote adoption of v6 by trying to lengthen the
lifetime of IPv4.  This seems contradictory to me.  By making v4 last
longer, we are only retarding the growth of IPv6
usage.

2. The stated objectives of all RIRs are Conservation, Aggregation and
Registration.

See slide 33 of http://www.afrinic.net/training/materials/Member-Training-v2.pdf

This policy negates 2 of those objectives.   It actively promotes
deaggregation by setting an artificially low maximum allocation size.
Currently our maximum allocation size is a /10.  This policy limits
LIRs to a /23 in the exhaustion phase, thus mandating the announcement
of multiple /23's.  This does not limit routing table growth, instead
it encourages it.

In terms of conservation, the intended consequence of this policy is
to reserve a /12 for future use, which is an idea I support.  However,
the UNINTENDED consequence of this policy is to reserve the majority
of the last /8, by "locking up" most of the 16 million IPs of the last
/8.

quoting GB:

"You limit each LIR to a /21 of addressing resources. That provides
for 7680 (excluding the reserved /12) LIRs to obtain maximum
allocations. AfriNIC currently has 1009 members and a growth rate of
less than 100 members per year[1]. This means that at the end of 5
years 80% of our final /8 will be locked up and un-allocatable under
this policy."

Conservation is one thing, however making a large portion of the /8
"un-allocatable" as an unintended consequence is not only overly
conservative, it is absurd and bad policy making IMHO.  In other
words, while trying to extend the lifetime of v4 in this region, we
are making it much shorter artificially!!

The response I got to this objection at he meeting is that "none of us
know what the environment will be in the future".  While this is
correct, it does not seem unreasonable to conclude that the current
growth rate of LIRs will not change substantially due to consolidation
in the ISP market.  Even if we quadrupled the number of LIRs in 2
years, we would still be "locking" a substantial portion of the last
/8.

While my crystal ball may be just as foggy as others, in the
theoretical situation where each of the existing LIRs (~1000 now and
approaching 1500 in 3 years time) has gotten all of their possible
allocations AND demand by endusers for IPv4 is still high, then there
is only one possible outcome (unless we change this policy at some
point).  That outcome is a surge in demand for PI (EndUser) blocks.

Both Aggregation and Conservation depend upon LIRs requesting
resources they need.  This is our mantra (some of our critics call it
our "religion" see Milton Muelller).  We are changing that due to the
perception (not the reality) of a shortage of IPv4 blocks in this
region.  While it still holds true for v6, we are shooting ourselves
in the v4 foot.

On process, I just didn't see the consensus that the PDP-MG did in the
room or on the list on this one.  Points out that we do indeed need a
new PDP.  But that one is for another day ;-)


-- 
Cheers,

McTim
"A name indicates what we seek. An address indicates where it is. A
route indicates how we get there."  Jon Postel