[arin-ppml] Why should we do Proposal 121
owen at delong.com
Wed Dec 8 20:04:06 EST 2010
I've been asked by a fellow AC member to spend some more effort documenting
reasons we should enact proposal 121.
1. Current IPv6 policy is being interpreted to the detriment of ISPs that
have subordinate ISPs. Subordinate ISPs should be able to get PA
space from their upstreams equivalent to what they would be able
to get directly from ARIN. Currently, ARIN is not allowing for the
possibility that an ISP would reallocate /32s (or larger) to their
2. HD Ratio is confusing to people and of dubious value vs. basing
utilization on simple ratios.
3. A large number of historic outages have been the direct result
of people being generally bad at bitmath. By moving allocations
to nibble boundaries, we can reduce the likelihood of these
errors by removing complex bitmath from most network deployments.
4. The current complexity of getting allocations larger than /32 from
ARIN is resulting in many providers choosing, instead, to shrink
the size of assignments they give to end sites to less than /48.
If this practice becomes wide spread, it will likely reduce possible
innovations in the SOHO realm because vendors will generally
implement to the lowest common denominator. We have already
seen this with various consumer electronics that assume that
there is no way to reach the device from the global internet and
either depend on rendezvous/relay solutions or simply don't
provide features that take advantage of global reachability.
IPv6 allows us to restore the end-to-end model to the internet
so that these and other capabilities become realistic. It would
be unfortunate to artificially limit the potential for innovation
by reducing the number of bits available to do so unnecessarily.
Further, a concern was raised that we would potentially blow through
address space too rapidly. The particular AC member in question
works for one of the largest (potentially the largest) eyeball ISPs
in existence. He stated that under the proposed policy, his network
might qualify for as much as a /12. There are probably no more
than 16 organizations of comparable size (I estimate 10 or
fewer) worldwide. Let's assume we burned through twice that
number of /12s (>80% of IPv4 allocations in ARIN have been
to the 24 X-Large organizations).
Even if we assume the absolute maximum possible...
5 RIRs x 24 X-Large organizations = 120 /12 allocations.
So, 120 /12s would be 80% of the total IPv6 we would
need to issue, 120/0.8 = 150 total /12s consumed
world wide, if every RIR were to implement this proposal.
That's roughly 23.5% of the first 1/8th of IPv6 address space.
That means we still leave 97% of all IPv6 space on the
shelf for the foreseeable future.
The potential consumption from this policy simply isn't an issue.
I expect that our total consumption will be far less than the
worst case I show above. In fact, if we get through 20 /12s
in the next 5 years, I'd support reviewing the policy and
consideration of whether we should reduce allocation sizes.
More information about the ARIN-PPML