When this policy was presented at APNIC, Doug Barton of IANA was not convinced about establishing the /12 for additional allocations. There exists the risk of allocating too much space. It is necessary to establish a threshold for how much additional space may be requested. In other words, that the initial allocation be a fixed allocation but that the second allocation be left open.
APNIC's proposal was also discussed at ARIN last week. Among the aspects of this discussion that I would like to highlight is that it was stated that publishing or announcement should not be part of this policy. There is no consensus about the size of the minimum allocation, although it is agreed that it must be somewhat larger than the current /23. There was a tendency to accept the period of 18 months instead of the 36 months proposed in APNIC's policy, in addition to considering a measure similar to the IPv4 allocation policy by IANA to RIRs for calculating additional allocations instead of 50% of the usage established in APNIC's policy.
Minimum allocation must be designated through global discussion and a bottom-up procedure, and we must not provide the opportunity for IANA to seek comments from third parties that will delay the adoption of this policy. Seeking third party opinions is not compatible with the global Policy Development Procedure. The policy that will be developed must be precise and specific.
The recommendation to establish the size of minimum allocations as a /12 and a time frame of 36 months is based on the investigations carried out by Geoff Huston, which are available at APNIC's website (http://www.apnic.net/mailing-lists/sig-policy/archive/2004/08/msg00066.html) Similar discussions have taken place at APNIC. This is recorded at APNIC's mailing list, where you can find more information on this matter.
The IPv4 policy refers to allocation units and, although the unit is established as an /8, this does not mean that more space cannot be received. I think it is very appropriate to mention the /12 as the allocation unit. I think that the criterion used for additional IPv4 allocations is appropriate for this IPv6 policy. I would not be concerned about whether there will be excessive allocations. The original policy included IPv6 space reserves by region which may contribute to some governments' concerns as to whether the resources will be available when they are needed.
About the concern of making excessive allocations to "small" registries, this was not a personal opinion but the opinion of Doug Barton of IANA. A good idea would be to assess the needs of each region. One option would be to measure a /48 per inhabitant of the region.
Regarding the matter of creating regional IPv6 reserves, this was also discussed at the ARIN meeting. It was suggested that the routing table be observed, being careful with the aggregation. Currently IANA is allocating /23s to the RIRs which are not contiguous to the region, which places aggregation at risk. Therefore, it would be prudent to manage space reserves for the regions, not only from the political point of view but also from the technical point of view. During the ARIN meeting the size of these reserves was mentioned, approximately an /8. IPv6 aggregation is important.
It is important to bear in mind that the current discussion about IPv6 concerns only one eighth of the total universe. The other seven eighths remain untouched, and this must also be considered. The creation of IPv6 architecture meant to include the distribution of this resource not only to individuals but also to different devices that require public and unique addressing. There is the concern that the measurements on which the distribution will be based will consider the population of the regions.
In order to provide the audience with some context, the intention of the discussion that is taking place is to have a global policy, one which is the same for all the regions. This is the reason why reaching consensus is such a difficult challenge.
I am not saying that a /48 should be assigned to each individual, but that this is used as a measure at high or global level.
In reference to Germán's comment, as it is a global policy it must be as simple as possible. We should not try to include every possible case or consideration. This will make its adoption in each region easier.