[arin-ppml] Policy Proposal 122: Reserved Pool for Future Policy Development
It seems we have a couple of issues/questions around NRPM 4.10. To
help me (and hopefully others) think through the issue, let me see if
I can identify several of them:
Q1: Should a reserved pool of addresses be reserved, and made
available after free pool exhaustion, to support IPv6 transition?
My answer would be yes, and I believe we have a strong consensus
(expressed through the original adoption of 4.10, and in subsequent
discussions) for this.
Q2: Does NRPM 4.10 need to be updated at some point?
Again, I think the answer is yes, and it seems we have quite a bit of
community support for some kind of revisions.
Q3: Can we agree on how to update 4.10?
Perhaps. We seemed to have some points of general agreement, but the
last proposal attempting a comprehensive rewrite of 4.10 failed.
Perhaps smaller tweaks would do better, or new policy proposals (like
Q4: What should happen if we fail to reach consensus on any
modifications to 4.10? Should it expire at some point?
This is where I think proposal 122 comes in:
IMO the solution proposed by 122, which is to revoke 4.10, reserve an
equivalently sized block for one more policy cycle, and then return it
to the free pool if nothing can be agreed upon, is a less desirable
solution than the status quo.
I seem to recall some people doing the math on demand for space under
4.10, but I don't remember the results, so I'll give it another try.
If we assume that every organization requesting space today will find
a way to request space under 4.10, and somehow manages to qualify for
the maximum number of space allowed under the policy, we can figure
out how quickly the 4.10 reserved pool could be used up. According to
https://www.arin.net/knowledge/statistics/index.html, ARIN is
currently processing <200 IPv4 address space requests per month. So
let's assume 2400 requests/year. Since ISPs can get 1 year of space
at a time today, let's assume that's ~2500 orgs actively requesting
space, and let's assume that number doubles after exhaustion, to say
5000 orgs. If each such org qualifies for a /24 every 6 months under
4.10, then it would take approximately 3 years to assign all of the
space from the /10 reserved under 4.10.
So IMO our best course of action is to get together a number of minor
tweaks that we think should be made to 4.10, and get them into the
policy process ASAP. If any of them get consensus, we should be able
to adopt them before more than a small fraction of the 4.10 pool is
I don't think it would be a good idea to put a near-term expiration
date on the /10 reservation currently defined under 4.10. I would be
less opposed to simply suspending 4.10 for a few months while we
discuss alternatives, but IMO that isn't really necessary given 4.10's
/24 maximum allocation and one-allocation-every-6-months restriction.
IMO it would actually be better to start using 4.10 upon exhaustion
and use our implementation experience to inform future tweaks to the
rules for using its reserved pool.
On Mon, Nov 22, 2010 at 4:17 AM, Hannigan, Martin <marty at akamai.com> wrote:
> On 11/20/10 1:27 AM, "William Herrin" <bill at herrin.us> wrote:
>> On Fri, Nov 19, 2010 at 7:07 PM, Hannigan, Martin <marty at akamai.com> wrote:
>>> I agree. My date was deliberate though. I don't want to drag this out to
>>> Philly as we risk being exhausted by then.
>> Then don't. Close the window in August, well before Philly. Pick which
>> meeting you want to be the last chance for community consensus and
>> close the window just enough time later for a draft policy to have
>> gone through last call and ratification. Don't drag the deadline out
>> to the close of the following meeting.
> That makes sense. It coincides with AC and BoT elections too which I think
> is a good thing with respect to this.
> On a higher level, I've read two points that have little substance with
> regards to why this proposal shouldn't move forward.
> 1. 122 backdoors this to the freepool! Dangerous! Harmful! [handwave]
> I had hoped that it would be clear that something has to be done to preserve
> this in a manner that works for most if not all of us. The implication that
> some would seem intent to actually battle this out to the point where we
> would lose transition addresses is egregious. I think timing this to
> coincide with AC elections is probably better. See below.
> 2. What about the rationale in 2008-5?
> What about it? That was over a year ago and several AC members have noted
> that the policy is insufficient. Are we really going to let something this
> important go forward in a half-baked manner?
> Three people seem to be fixated on the expiration. I removed the CI argument
> and I'm willing to remove that argument as well.
> I could rewrite this to suspend 4.10 and upon expiry, re-implement. That
> would seem to address all objections and allow for this to be fully
> addressed at election time if need be.
> You are receiving this message because you are subscribed to
> the ARIN Public Policy Mailing List (ARIN-PPML at arin.net).
> Unsubscribe or manage your mailing list subscription at:
> Please contact info at arin.net if you experience any issues.