[arin-ppml] Q1 - ARIN address transfer policy

[John Curran]

At 9:33 AM +0900 6/22/08, Randy Bush wrote:
>John Curran wrote:
>> where's all of the modeling from the open market folks regarding
>> consequences to the routing table?
>
>i believe lucy's theory.  the vast majority of blocks will trade pretty
>much straight across in a year or so, flushing out the unused space; the
>goal of the exercise.

Agreed on the goal (flushing our the unused space), but see no reason
why it will not deaggregate in the process of being flushed out if that
has the highest return and is allowed by policy.

>follow the cash.  what this discussion is really about is who gets the
>money, the rir bureaucracies or the holders of the resources. 

I don't know about other regions, but it's certainly not about that in
the ARIN region.  The ISP community has historically been pretty vocal
about preventing completely uncontrolled block fragmentation (even
though the routing table has plenty of examples of ISP's carrying more
specifics for their own business reasons), and that's the real question
on the table. 

Personally, I think RIR's need to plan for a long-term future where
their mission with respect to registration services is rather staid, as
IPv6 availability and allocation sizes will tend to preclude follow-on
allocation requests, but that's a different topic for discussion...

>the impact on the table will be that of bringing the current unannounced
>space back into use, multiplied by the idiocy factor that defrags
>current blocks no matter how acquired.

Bringing currently unannounced space back into use definitely creates
more entries, but currently blocks don't fragment from speculators trying
to maximize value by breaking up blocks, nor to do they deaggregate due
to constraints from the current transfer policy.  The transfer policies in
consideration in the various RIR's introduce new factors for fragmentation
and as kc said, some realistic simulation of routing and market dynamics
here would be useful.

/John