Hi Izumi. On 28/09/2007, at 12:30 AM, Izumi Okutani wrote:
David Conrad wrote:Suppose there are 6 /8s remaining in the free pool. An RIR comes to IANA and indicates they want another allocation. Current practice is to allocate 2 /8s (if justified). IANA allocates the 2 /8s, leaving 4 /8s. The obvious approach would be to allocate the remaining 4 /8s to the other 4 RIRs. Is that the intent?right, thanks for clarifying.... and a single /8 will be distributed to *5* RIRs if the RIR requestedfor 1 /8 instead of 2 /8s.(i.e.the requesting RIR gets 2 /8s in total, remaining four RIRs get 1)
If we are going to the trouble of reserving 5x /8s now for allocation in this manner, then why not allocate these as soon as an RIR has a policy in place for it's eventual use?
Given that the basic intent of this proposal is to provide RIRs certainty of address allocation at the point that the IANA pool dries up, this approach provides even more certainty.
(Thank you Izumi and Randy for making the point of this proposal clear at APNIC24).
4. RIRs should maintain the current address distribution criteria until the IANA Exhaustion Date.Perhaps not too surprisingly, I disagree with this particular clause. By analogy, we're driving down a road at 100 KPH and we see a brick wall ahead of us. This clause requires us to put the car on cruise control and close our eyes until we're about a meter from the wall. What is the rationale for this clause?The idea is to ensure LIRs can receive IPv4 address space they need(based on justifications) until the last minute with minimum confusion.Going by road analogy, you increase confusion for drivers if you add extra rules changing from time to time, which may lead to accidents/traffic jam. Our intention is to avoid confusion by maintaining a consistent rule.That could be one approach, and this is the part we intend to discuss asI would think a more rational approach would be for each RIR to encourage IPv4 conservation using whatever policies make sense in their region.regional policies after IED (presented as informational in APNIC24).- Is this proposal addressing a real need or problem?It isn't clear to me what problem this policy is attempting to address.When the remaining IANA pool is 5 /8s (or less), there are not enough blocks for all RIRs on consumption basis. You can't tell if your turn to request will come before the IANA pool runs out as it all depends on timing of your + other RIRs' request. This makes it more difficult for RIRs to plan distribution of the available pool in their regions.
At the point that an allocation request is made that IANA can't fulfil from the free pool, uncertainty is going to exist unless that RIR is requesting a single /8. If an RIR requests 2 /8s and only receives one, what will they do? Presumably they had planned their available pool distribution based on obtaining both of these /8s. So uncertainty still exists for the RIR who makes the last allocation request.
The only benefit that this proposal offers is the guarantee that all RIRs will have at least one /8 at the point the IANA pool runs dry. It does nothing to change allocation behaviour before this point. It doesn't make allocation any more equitable between RIRs. It does nothing to incent RIRs and their members to not get greedy as we get closer to exhaustion.
I think we are best served to allocate this last pool now, then move on to putting policy in place that makes the lead-up to address depletion as fair as possible to the entire internet community.
:-) I understand your point, but I imagine a single /8 won't attract too many investors. It probably won't last for more than few months to meet- What advantages are there to distributing the last remaining /8 blocks equally to the RIRs?Encouraging investment in developing countries by large ISPs in developed countries?their needs. I know quite a number of people are concerned about this point, so I'd be interested to hear more details on what people see as an issue.
Given that we are talking about a time when 'free' IPv4 addresses will be at their most scarce, and given the current lack of IPv6 deployment, we have to accept that this is *will* happen. It would be irresponsible to think otherwise.
Cheers, Jonny.