[arin-ppml] Q1 - ARIN address transfer policy

[Owen DeLong]

On Jun 22, 2008, at 4:04 PM, Scott Leibrand wrote:

> Owen DeLong wrote:
>
>>> Do you have reason to believe that deaggregation pressures will be  
>>> larger under 2008-2 than under a no-transfers exhaustion scenario,  
>>> where incumbents with space lease/SWIP it to downstreams needing  
>>> the space?
>>>
>> Yes... I think that the market will lead to a larger number of  
>> organizations
>> multihoming in order to qualify for the ability to transfer space  
>> in because
>> their ISP is out and they can't get space any other way.
>
> What about the case where ISPs are the ones that acquire the address  
> space via transfer?  Do you believe that something about a transfer  
> market will change current practice?
>
I do not believe a transfer market will generate large enough blocks to
satisfy the demands of the large(r) ISPs.  Therefore, those ISPs will  
begin
to instruct their customers on how to obtain their own space by  
"multihoming"
and then purchasing from the market.

I don't think ISPs will be the majority of recipients in the market  
soon after
the market is created (if we do this).

> Today, large holders (ISPs) get most of the address space from ARIN,  
> and reassign it to their downstreams.  This has several benefits,  
> most of them related to economies of scale (fewer interactions with  
> ARIN, staff who get more knowledgeable in IP issues since they deal  
> with it all the time, etc.).  In a post-exhaustion transfer market,  
> it would seem that all those advantages would remain, and in  
> addition large holders would be in a better position to acquire  
> space on the transfer market, since they would be more knowledgeable  
> about price trends, etc.  So I would think that the only real change  
> to your average small organization would be that after exhaustion  
> some ISPs will begin charging slightly more for Internet service to  
> folks who need a large number of public IPs.
>
Except that in a post-exhaustion transfer market wit ha 2 year  
timeframe between
successive acquisitions, ISPs won't be able to get enough from the  
market to
satisfy the demands of their customers.

As such, I believe that ISPs will pressure their customers who want IPv4
address space to bring their own from the market in smaller chunks.

>>> As Randy has pointed out, most of the deaggregates in the current  
>>> table are more-specifics of RIR-allocated routes.  Under a no- 
>>> transfer or 2008-2-type policy, I would expect that to remain the  
>>> case, at least unless/until routing table growth starts to  
>>> outstrip router capacity growth.
>>>
>> Yes.  I don't see that as a bad thing. That form of growth is  
>> somewhat
>> manageable and when those more specifics get filtered remotely,
>> little or no damage occurs.  This is not so of fractions of address
>> space that no longer have any affiliation with a larger aggregate.
>> Especially in scenarios where someone is still announcing said
>> larger aggregate and may not be carrying the more specific.
>
> Agreed.
>
> -Scott

Preserving the above scenario in a market seems impossible to me.
How do you see this happening in light of my comments above?

Owen