[arin-ppml] Draft Policy ARIN-2014-21: Modification to CI Pool Size per Section 4.4

[Martin Hannigan]

On Tue, Dec 23, 2014 at 5:57 PM, David Farmer <farmer at umn.edu> wrote:

> On 12/23/14, 14:50 , Martin Hannigan wrote:
>
>>
>>
>> On Mon, Dec 22, 2014 at 5:48 PM, David Farmer <farmer at umn.edu
>> <mailto:farmer at umn.edu>> wrote:
>>
>>     So far there has been very little discussion on this policy.
>>
>>
>> Not really. I recall a huge thread previously that demonstrated much
>> support. What's the hold up?
>>
>
> No hold up, the proposal was submitted a little more than week before ARIN
> 34. The AC Chair assigned shepherd, but there wasn't sufficient time to get
> it ready for the October AC meeting held at ARIN 34.  At the November
> conference call the AC accepted the proposal on to its docket, promoting it
> to Draft Policy, and it was posted to the list for discussion.
>
> I went back and reviewed the thread, there was some support for the ideas
> that spawned the proposal, but no specific text was posted at that time.
> Since the Draft Policy and the specific text was posted there has now been
> two statements of support and no opposition.
>
>
Its fair to expect that the OIX commentary, documented in full public view,
should count as support. That would be more than two.



>      Therefore, as one of the AC shepherds for this policy I would like
>>     to initiate some discussion of this policy.  Here are a few
>>     questions for the ARIN community to think about and provide feedback
>> on;
>>
>>     - The current CI reservation is for all CI not just IXPs, the
>>     problem statement discusses growth primarily in the IXPs as
>>     justification to expand the reservation.  Should we split off a
>>     separate reservation pool for IXPs?  Or, keep the current common CI
>>     pool?
>>
>> If you want to complicate this further, yes, let's do that. If not, are
>> you suggesting that there still isn't a large enough reservation?
>>
>> I see some merit in your suggestion, but dragging this out beyond
>> exhaustion doesn't sound like a wise idea. I'd go with a larger
>> reservation if you are concerned and have data points.
>>
>
> I think /15 is probably big enough.  But, there is no guarantee that the
> /16 that will be added will be used for IXPs.  As shepherd, I'm just trying
> to make sure the community considers the issues.
>
>
I agree, I think the /15 is big enough. With the change to three as the
minimum requirement and the addition, that at least solidifies some CI
addressing for IXPs. If we have to adjust again, it'll be doable with the
reservation in place. I think its big enough. For now.




>      - ARIN-2011-4 the policy that made the original CI reservation had a
>>     Policy Term of 36 Months following implementation, but this was not
>>     in the policy text itself and therefore did not get included in the
>>     NRPM.
>>
>>     https://www.arin.net/policy/__proposals/2011_4.html
>>     <https://www.arin.net/policy/proposals/2011_4.html>
>>
>>     If applicable, this would have expired July 2014.  So, should there
>>     be expiration date included in this policy text?  If there should be
>>     no expiration date, should we explicitly note the removal of any
>>     expiration date in the discussion of this policy?
>>
>> I was the author of that revision and the author or this one. That was
>> based on a belief in 2011 that ARIN would have been long out of v4
>> addresses and v6 adoption would be well underway. Expiry in retrospect
>> seems unnecessary since dual stacking is likely to prevail in CI for
>> quite some time to come. As it should.
>>
>
> Ok, then I think we should make that intent explicit, "regardless of the
> status of the ARIN-2011-4 Policy Term, this policy intentionally does not
> include an expiration date for the CI reservation."
>
>
I believe I specified "non" when I submitted the template so I'd argue that
its part of the modification.

     - There was discussion of smaller and larger than /24 IXP
>>     allocations, like /26 on the smaller side and that some very large
>>     IXPs are starting to need as large as a /22.  Also discussed was,
>>     sparse allocation for IXPs to allow expansion without renumbering.
>>     Should this policy includes any changes along these lines?  Why or
>>     why not?
>>
>> There's nothing that codifies that an CI prefix can not be routed so
>> linkage to the minimum allocation makes sense.
>>
>
> So keep a /24 minimum?
>

If later "we" managed to codify a variable MAU and hten have policy
reference that, it would be a reasonable approach. Until then, /24 is "ok".



>      - Should we try to get this to the PPC at NANOG 63 in San Antonio as
>>     a Recommended Draft Policy?  Or should it wait go to the PPM at ARIN
>>     35 in San Francisco as a Recommended Draft Policy?  What about ARIN
>>     free pool run-out timing?
>>
>> Speaking as an Open-IX community (board) member http://www.open-ix.org/
>> and referencing previous discussions here that pointed there as well,
>> there is a substantial and demonstrated amount of support for this.
>>
>> San Antonio.
>>
>
> I'd be happy to work toward getting this to San Antonio as a Recommended
> Draft Policy.  However, the normal path would be Draft in San Antonio and
> then Recommended Draft in San Francisco.  So, to justify accelerating this
> policy we need clear and strong support from the community for this
> policy.  Lukewarm support with no opposition is not sufficient to take this
> to San Antonio as a Recommended Draft Policy in my opinion as the shepherd.
>

I'm happy to cross post and invade ppml rms style with OIX. That doesn't
seem necessary based on the transparency of our discussions.



>
>      Do you support the policy as written, if not are there any changes
>>     that could be made that would allow you to support the policy?
>>
>> As written.
>>
>> Best,
>>
>> -M<
>>
>
> Thanks.
>
>

:-) You're welcome.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.arin.net/pipermail/arin-ppml/attachments/20141223/9ef84f66/attachment.htm>