[ppml] [sig-policy] Policy Proposal: IPv4TransferPolicy Proposal

Stephen Sprunk stephen at sprunk.org
Tue Feb 12 11:59:56 EST 2008


Thus spake <michael.dillon at bt.com>
> In fact, is there currently a barrier to doing an ERX style
> transfer?

ERX was only for legacy registrations, which were all on convenient classful 
boundaries, limited in number, and in specific /8s that were already a mess 
from a DNS perspective; supporting an unknown number of transfers on 
arbitrary boundaries in any /8 is a rather more complex and painful 
undertaking.

> For instance, imagine that a European ISP built a network in the USA,
> using RIPE addresses,

They're not supposed to do that; while it's not written into the policies 
anywhere I can find, it seems to be accepted that addresses are only to be 
used in the region of the RIR that issued them, minus incidental use like a 
large network on one continent dropping a handful of routers at IXes on 
another continent.  If you've got actual customers in a given region, you 
should be getting addresses for them from the respective RIR, not the one 
where your headquarters is arbitrarily located this year.

How well the above is enforced in practice, I can't tell; my guess is that 
we don't see much "registry shopping" today because most members of the 
community voluntarily do the right thing, the RIRs' policies are largely 
equivalent, and the RIRs' staffs are doing a bit of enforcement via internal 
policies _not_ voted on by their members.  This probably needs fixing, but 
it doesn't seem urgent at this point.

> Assume that the ownership transfer of IP addresses within RIPE has
> already been done. Can this ISP transfer these addresses to ARIN
> under current rules?
>
> If not, why not? What are the specific barriers? If this was
> a US incorporated owner would it make a difference?

AFAIK, there is no process in place to transfer non-legacy blocks from one 
RIR to another.  If you have addresses in a /8 delegated by IANA to a given 
RIR, status quo is that they'll be registered with that RIR forever. 
Perhaps that needs to be changed, but we'd need a global policy to do it and 
AFAIK there is no such proposal on the table at present -- and we're past 
the deadline for any new proposals in the current ARIN cycle.

S

Stephen Sprunk         "God does not play dice."  --Albert Einstein
CCIE #3723         "God is an inveterate gambler, and He throws the
K5SSS        dice at every possible opportunity." --Stephen Hawking 




More information about the ARIN-PPML mailing list