Raul Echeberria wrote:
Geoff. One observation.Accepting transfers only within the region is a way to keep the IP addresses within the regions in which they were originally allocated.I don't know if it is good or not, but a fact.If the transfer of legacy space is also admited, it has a great impact in other regions, since most of the unused space belonging to the legacy blocks, will feed regional markets in the developed countries. This is the promotion of regional markets instead of global markets.My opinion is that the regional approach to a global poblem that is the IPv4 deployment is not the right approach.
There are two issues that I can readily see here - the first is the issue of defining the mechanics of a cross-rir transfer. and the second is the assessment of the value of the intended outcomes and the risk of other unintended outcomes.
Concerning mechanics, in looking at the APNIC and ARIN proposals they both take the approach of "qualification" of the two parties to a transfer. One approach would be to 'recognise' the qualification from another region - i.e. taking an ARIN perspective a "transferor" (is that really an english word? ;-)) meets the criteria listed in section 8.4.1. To extend this to allow cross-RIR transfers it would be a case of adding "or meets the criteria as listed (insert reference to the transfer policy of another RIR) for members of (RIR). Similarly the conditions of the transferee could be augmented by reference to the relevant qualifications in the policies of other RIRs. So in terms of extending the mechanics of the policy proposals to encompass cross-RIR transfers then I'd suggest that there are ways to achieve this though the use of mutual recognition of each RIR's qualification processes.
(I should note a slight inaccuracy here from my previous posting - I thought that the presentation at the October 2007 RIPE meeting :
http://www.ripe.net/ripe/meetings/ripe-55/presentations/vanmook-v4policy-change.pdfhad been submitted into the RIPE policy process as a proposal, but it looks like this has not happened as yet as there is no proposal listed at http://www.ripe.net/ripe/policies/proposals/index.html)
The second issue is perhaps the one that deserves further consideration. There is a strong argument in favour of looking at this issues of transfers from a global rather than regional perspective. The regional distribution of IPv4 addresses today, the regional levels of demand for IPv4 addresses, and the projections of demand within each region do not appear to be well-aligned, and an imposition of a regional 'containment' of transfers may well lead to less than desireable outcomes.
One question I have is whether a global transfer model would run the risk of unintended outcomes. Would a global transfer domain create inequities and imbalances in the residual IPv4 internet that may require some other form of intervention or mediation to redress? What are the risks of such outcomes, and it is possible to propose some policy mechanisms that may mitigate such risks?
regards, Geoff