I disagree with you. RIR YES would need a policy for doing what you are suggesting. In fact, the policy proposed is a particular case of what you are suggestign.

When there are only 5 /8s IANA ask all the RIRs what would be your future (short term) needs. And obviously the answer would be "At least one /8".

If I were IANA then that would not be an acceptable answer. At some point, IANA will be allocating large percentages of the global free supply of IPv4 addresses to RIRs and they will have to satisfy themselves that they are not treating the other 5 RIRs unfairly. In the scenario that you describe, you expect IANA to give out 20% of the total IPv4 address space just because you claim that you need it. This is not sufficient. At a minimum, you should justify your request by showing the full status of all your previous IPv4 allocations and this information should be made public. After that, all other RIRs and the public should have an opportunity to file comments with IANA. Finally, after considering all the comments, IANA should decide how many IP addresses to allocate. IANA would be within its rights to begin doing this tomorrow, not waiting for the last 5 /8s. My main concern with all these types of proposals is that they should not cause an IPv4 address shortage to happen sooner than it would happen without intervention. Some proposals, like the one which locks up the last 5 /8 blocks would cause shortage effects sooner, and are therefore bad. However, if IANA/ICANN starts taking its responsibility seriously and applies more scrutiny, in public, of all RIR allocations, then we would be unlikely to cause shortage effects earlier. But IANA and ICANN need to take the initiative on this, not wait for the RIR/NRO two to three year policy cycle. The criteria for RIR allocations can and should vary on a case by case basis. For one thing, I'm aware that the RIRs have a lot of addresses tied up in "reserved" status which should be taken into account when analyzing the overall picture. I'm also aware that by taking up the slack, we may extend IPv4's lifetime by another year or two but when we do run out it will be like hitting a brick wall. Nevertheless, companies which cannot implement IPv6, within the next two to three years, are failing in their duty to shareholders and customers. I don't think we should be trying to protect these companies in any way, just extend things as far as we can into the future to give time for the network operator and vendor project cycles to deliver results. In the current world of business, project cycles are longer than they used to be back in the early 1990s. Many people don't like this but we have to live with it, since we all depend on other companies to play their role in IPv6 deployment. We need better support on the hardware, we need upgraded OSS/NMS software, firewalls, load balancers, and consumer broadband gateways. In addition we need scalable v6/v4 translation clusters. Any policy changes, and behavior changes, need to support all these companies in providing the necessary upgrades within their plodding project cycles. I don't think that the proposals currently before any of the RIRs do this, and I don't believe that the RIR/NRO policy processes work quickly enough to deal with this issue. Only ICANN and IANA can take the needed action in the required amount of time. Fortunately ICANN is also an open and responsive organization so we have a chance to fix this in their forums. -- Michael Dillon