[arin-ppml] Draft Policy ARIN-2019-17: Returned Addresses to the 4.10 Reserved Pool

Martin Hannigan hannigan at gmail.com
Fri Sep 13 15:58:20 EDT 2019

On Tue, Aug 27, 2019 at 5:10 AM Owen DeLong <owen at delong.com> wrote:

> On Aug 26, 2019, at 18:09 , Martin Hannigan <hannigan at gmail.com> wrote:
> On Thu, Aug 15, 2019 at 5:10 PM Owen DeLong <owen at delong.com> wrote:
>> Really, it seems to me that this proposal is another attempt at
>> eliminating the waiting list for unmet requests.
>> The first attempt (ARIN auctions the space) met with resistance from
>> ARIN’s legal team (for good reason), so now this attempts to sequester the
>> space where it will be hard to distribute rather than allowing the waiting
>> list to have any potential to compete with the transfer market.
>> The proposed targets (4.4 and 4.10 pools) are well stocked and unlikely
>> to run out in any useful IPv4 lifetime.
> Data: see "reserved"
> https://www.arin.net/vault/knowledge/statistics/2018.html
> If you look at the numbers you can see a mild acceleration and then what
> looks like it could be a deceleration of assignments from this pool (and
> many others ironically). Note, the time period is only a year. My
> conclusions aren't rock solid as a result. I'm really just eyeballing.
>> As such, restocking them from returned space strikes me as just a way to
>> sequester this space where it cannot be used.
>> IMHO, this is counter to ARIN’s mission and should not be allowed.
> Do you think having pools of the magnitude and current use profile of 4.10
> and perhaps even 4.10 are in need of being re-evaluated?
> Admittedly, its a little re-arranging of the deck chairs. However, if the
> waiting list is truly stocked with people who have a need and aren't just
> using it to hedge against a transfer they already completed it may make
> sense to make adjustments. And to potentially apply it there. I won't speak
> to the authors intention on 4.10, but I had a hand in 4.4. I determined and
> argued for the pool size based on conditions on the ground at the time.
> Today I would argue IXP growth in NA is somewhat flat, that most are
> commercial in nature and won't have trouble using the transfer market. Yes,
> there are some that aren't for profit, but even those usually have an easy
> time fundraising.
> I think that IXPs are a sufficiently important benefit to peering density
> and interconnection that even if they are commercial, I don’t mind holding
> a carve-out for them. Note also that 4.4. covers some TLD and Root server
> addressing needs as well.

Noted. And agreed on IXPs. The pools primary intention, at least as I
recall the expansion some years back, was to protect against the unknown of
the market. I don't think anyone had a clue about what the cost of
addresses were going to be and when. Today, from my seat on the Internet, I
see stability.

> I think that the current sizes are working well for not. If we get a
> couple of years down the road and see that they really aren’t being used
> (and IPv4 is still perceived as a vital growth commodity), I’d be
> potentially willing to re-evaluate at that time, but in that case, if we
> shrank the pools, I’d probably want to use whatever we released to drain
> the waiting list queue (which I know will make the author’s skin crawl).
Whether it is the waiting list or something else, I'm not sure it matters.
And we said we'd re-evaluate the pools a few years down the road then IIRC.

If anything should have a pool, a small pool for the Caribbean makes sense.
> The contrast in economic conditions and market norms justify it easily in
> my opinion. Such as the US large volume transit price per mb/s between
> $0.05 and $0.15 nd for most there we're substantially north of a $1.00.
> I’m not sure what the relationship is between transit pricing and the need
> for IPv4 addresses. If anything, most of the Caribbean economies have a
> slight advantage in the IPv4 market place as few of them need short
> prefixes.

Sorry if I wasn't clear. Transit pricing is orthogonal to IXP value. The
more costly transit is, the more valuable in many ways the IXP becomes.
Thinking about ARIN's service region mainland bits are cheap. Island bits
are expensive due to much history, incumbency and granted monopolization.

> I also think that Caribbean transit prices will continue to fall and while
> they might not reach parity with the US in the next few years, the gap will
> close.
Its sort of interesting in that a successful IXP benefits transit,
especially on an island. Curacao is the case study. The more caching that
took place and saved costs for local networks, the more transit was needed.
I'm near certain that it caused growth in transit use albeit the ratio is
where the savings and offsets made that occur.

> However, if you do feel that such a carveout would be desirable, please
> submit  a policy proposal. Make sure to cover where the addresses to stock
> the pool are supposed to come from and what, exactly, are the criteria for
> consuming from the pool. Lots of details to mask many devils in there, IMHO.
I know a lot about the Caribbean as much as the folks who live there do and
they are here now. I'll defer to them. I would only argue if such a thing
is needed we would go back to these pools and instead of re-arranging deck
chairs we burn a few to create smaller ones.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.arin.net/pipermail/arin-ppml/attachments/20190913/067fe749/attachment.htm>

More information about the ARIN-PPML mailing list