[arin-ppml] Policy Proposal 124: Clarification of Section 4.2.4.4

[Scott Leibrand]

On Thu, Dec 9, 2010 at 8:02 PM, Hannigan, Martin <marty at akamai.com> wrote:
>
> On 12/9/10 10:04 PM, "Scott Leibrand" <scottleibrand at gmail.com> wrote:
>
>> On Thu, Dec 9, 2010 at 11:28 AM, Hannigan, Martin <marty at akamai.com> wrote:
>>> To be perfectly clear, I have proposed as a part of PP 124
>>> to remove the following from the policy statement in 4.2.4.4:
>>>
>>> "This  reduction does not apply to resources received via section 8.3. An
>>> organization receiving a transfer under section 8.3 may continue to  request
>>> up to a 12-month supply of IP addresses."
>>>
>>>
>>> The effect would be that transfers would also be subject to a three-month
>>> window of need.
>>
>> IMO this would have a substantial negative effect on the number of
>> routes in the IPv4 routing table.  The reason for keeping the 8.3
>
> Along with all of the other disaggregation?

I suspect most of the ARIN free pool will be given out as one block
per org per time period, similarly to how it's been given out
historically.  Draft policy 2010-1: Waiting List for Unmet IPv4
Requests, which was approved by the Board and is awaiting
implementation, directs ARIN to "fulfill the request with the largest
single block available that fulfills the request", and stipulates that
"an organization may only receive one allocation, assignment, or
transfer every 3 months".

Given that the free pool will be exhausted over a relatively short
timeframe (at least until we get down to legacy class C swamp space),
I don't think there'll be much impact on the rate of growth in the
routing table from the 3-month window of need for getting space from
ARIN.  However, I expect 8.3 transfers to be the only practical way to
get space for a lot more than 3 months, so I think it's reasonable to
leave that window at something longer.

>> supply at 12 months was because we expect that a large fraction of 8.3
>> transfers will result in deaggregation of a larger block into multiple
>> smaller blocks.
>
> I'm not sure why anyone is going to want to spend that far into the future
> with respect to the cost and lack of visibility into the state of
> transition. This is a great way to undermine the STLS.

As detailed above, any organization can get up to 12 months' worth of
space, and can return to the transfer market for more space after 3
months if they need more.  I believe that restricting organizations'
flexibility by limiting them to 3 months' of space (particularly while
preventing them from coming back for more space for 3 months after
that), as this proposal does, would actually undermine the transfer
market.

>>If transfer recipients are required to get a new
>> transfer every 3 months instead of every 12, we can expect 4 times as
>> much demand for deaggregation in such situations.
>
> That would be a guess. You don't know what the cost of addresses will be.
> That will directly impact demand. Be careful about being too excited about
> markets. They will be necessary, but it won't be something to be happy
> about. The cost of an address in a market using STLS will dwarf the costs we
> have now.

Yeah, there will be some organizations that prefer to get the bare
minimum (3 months' worth) of space.  If that's the predominant
behavior, then removing the 12-month option would be a no-op for those
organizations.  However, I know there are organizations that will want
to try to secure enough space that they're not constantly having to go
back to the market and deal with the volatility there.  Removing the
option to get 12 months worth of space would eliminate those
organizations' flexibility to do so.  I don't see a benefit to the
community of doing that.

-Scott