[arin-ppml] ICP2 and the role of ARIN in an ipv4 address market

[Milton L Mueller]

Interesting suggestions, Matthew. Worth considering. What matters now is not whether those are exactly the right solutions or whether someone on the list can pick at one or two details. What matters is whether this group understands the problem and takes the problem seriously enough to start working on it. 

In that regard, I view John's invocation of ICP2 as off-target. 

ICP2 is about how to establish a new RIR. What is on the table now is not a new RIR, but a different regulatory model for the existing RIR in the North American region. ICP2 is completely irrelevant to the problem we face once legacy holders start to sell address blocks to others who may or may not choose to enter into an RSA with ARIN.

The RIR model assumes a territorially exclusive address allocation and post-registration services entity, one that is exactly like the ones we have now. Benson S.'s proposal was not for a new RIR, nor are the Denuo/Addrex people proposing a new RIR. They are proposing to alter the relationship between the address registry/allocation function and the Whois and other post-allocation services functions.

The (now undeniably real) possibility of v4 transactions that occur without ARIN's approval or participation means that we need to deal with that problem. John's desperate attempt to get a contract - any kind of a contract - from the buyer of Nortel's addresses proved how existing policies are founded on the false assumption that everyone has to go through ARIN. It also revealed how community policies had to be bent almost beyond recognition as a result. To me, that's a sign of stress on the system and is ignored at our peril. 

--MM

> -----Original Message-----
> What about adding a mandatory field to the Whois database output:
> "Agreement:" which would list one of three possible values:
> LRSA, RSA, or None.
> 
> Networks could then make decisions about whether or not they wished
> to accept routes from that organization based on the nature of the
> relationship
> between that organization and the RIR.  A paranoid or fascist network might
> opt to not accept routes from entities whose Agreement field was set to
> anything other than "RSA".  A slightly less paranoid network might allow
> RSA or LRSA, but drop prefixes from entities with "None".  Liberal network
> policies would accept prefixes from anyone, regardless of their relationship
> to the RIR.
> 
> Would that provide a means for the community to add some teeth into the
> mix, to potentially ostracize networks that do not opt to adopt some level
> of formal relationship with the RIR?
> 
> Matt