Global IPv6 policy question

[Iljitsch van Beijnum]

On 24-jul-2006, at 16:13, Tim Chown wrote:

> I guess one other question is whether there is any value in reserving
> larger chunks of sequential space for each RIR, whatever is actually
> allocated now, should any regional aggregation have any value in the
> future?   i.e. whether to think about where the /16's come from.

Currently IPv6 space is given out at these rates (in /32s):

mysql> select left(day, 4) as year, rir, sum(pow(2, 32-num)) as  
thirtytwos from addrspace where type='ipv6' and num <= 32 and rir =  
'ripencc' group by year, rir;
+------+---------+------------+
| year | rir     | thirtytwos |
+------+---------+------------+
| 2002 | ripencc |         73 |
| 2003 | ripencc |        173 |
| 2004 | ripencc |       7088 |
| 2005 | ripencc |      17504 |
| 2006 | ripencc |       6709 |
+------+---------+------------+

+------+-------+------------+
| year | rir   | thirtytwos |
+------+-------+------------+
| 2002 | apnic |         31 |
| 2003 | apnic |         39 |
| 2004 | apnic |       6211 |
| 2005 | apnic |       9378 |
| 2006 | apnic |       2175 |
+------+-------+------------+

+------+------+------------+
| year | rir  | thirtytwos |
+------+------+------------+
| 1999 | arin |          2 |
| 2000 | arin |          8 |
| 2001 | arin |         11 |
| 2002 | arin |         15 |
| 2003 | arin |         46 |
| 2004 | arin |         42 |
| 2005 | arin |         58 |
| 2006 | arin |         29 |
+------+------+------------+

This means it will take several years to use up a /16, but the RIPE  
NCC reserves a /29 when it gives out a /32, but still, a /12 will  
take a good numer of years, so I don't think being able to predict  
which RIR a given /12 is going to end up with that far ahead is of  
noticeable value.