This is my first post to sig-policy, so be gentle :-)
We're an ISP in Australia who are in the process of productionising our dual stack trial. We've run into this kind of issue where a /32 does run into issues for us allocating /48s to customers and have looked at allocating less to them. (We obtained the /32 back before most of these policies existed and when we were much smaller).
We have a /32 and approx 250k customers which clearly isn't enough (65k /48s). But /56s as per the current policy with HD ratios (http://www.apnic.net/policy/ipv6-address-policy#5.3) in 5.3.1 means we do have enough but are in a bit of a gray space (16M /56s) as we do have a reasonable number of customers who match the /48 allocation policy.
We have 11 serving sites for the bulk of our broadband customers. The largest terminates approximately 50k customers.
As I understand it this means:
Y=20 (50k * 4/3 is slightly more than 65k (2^16) so becomes 20)
N=48-(4+20) = 24
Which gives us 16M /48s assuming my ability to use basic maths hasn't degraded that much :-)
Effectively this means we can allocate /48s to all customers.
This would mean a much easier allocation strategy for us and allow us to allocate to customers space as they expect. I can't see we'd ever come back for another allocation (more efficient in the global table).
On 12/01/2011, at 9:29 PM, Gaurab Raj Upadhaya wrote:
Peering Manager and Team Lead - Commercial and DSLAMs