Takashi Arano wrote:
My question is HOW. I personally can't imagine which kind of policy will achieve this goal.Even if there are any, who in the world can agree with a new policy which makes him/her giving up requesting addresses or just getting significatly fewer addresses?I guess It would be more tough discussion than the proposed policy will cause.
Arano-san, Fair question.Firstly I was not suggesting any reduction in allocations, clearly to avoid a chaos in the short term applicants need to continue to receive allocations of IPv4 space appropriate to their needs. It was also my suggestion that there would be no artificial retention of space, we'd see allocations down to the last block to allow as long as possible for migration.
The suggestion is that additional criteria for allocation be progressively brought in to make applicants implement IPv6 technologies rather than ignoring it for the most part. I was not envisaging any change to the current criteria.
There has been substantial discussion of IPv6 on the NZNOG list and one of the conclusions I have drawn is that, as with global warming, the threat of exhaustion is inadequate to drive significant action. Commercial imperatives see only the costs and the long term benefits of IPv6 implementation are too remote to allow money to be spent.
Indeed in this part of the world IPv6 transit is simply not available except by tunnelling. My suggestion is to artificially generate demand for action, rather like the Carbon Credits scheme is supposed to motivate people to take useful actions over global warming. It may not work but that is not an adequate reason for taking no action.
For the purposes of illustration only here is a small list of hurdles that could be imposed. I'm a little reluctant to do this as they all could all be objected to on various grounds but perhaps we need to be specific here. The dates are just for illustration!
1. Applicants could be asked to demonstrate that they have implemented a dual stack system somewhere on their network by end 2007
2. By end 2008 a requirement could be that an infrastructural item on their network is dual stack (eg a mail or web server)
3. All visible infrastructural items on the network to be dual stack by end 2009 (also an easy test to automate)
4. A commitment to only deploy dual stack CPE by end 2009 (just a promise but that's the case about a lot of the current requirements)
5. Demonstrate that dual stack equipment is being deployed and perhaps a percentage of current allocation is IPv6 capable by 2010
6. By 2011 a free IPv4 market is established and address space can be traded at market prices. RIR policies are modified to effectively sell new IPv4 space at the market price to prevent arbitrage opportunities.
Sure there are policing issues and people will perhaps not be as truthful as they might. However something needs to start driving IPv6 deployment or I believe we will see some serious problems as IPv4 runs out.
Thank you for your consideration. -- Robert Gray bob at brockhurst dot co dot nz