Re: [Softwires] MAP-E question -- first prefix

[Tom Taylor <[e-mail]>]

On 09/04/2013 3:14 PM, Ole Troan wrote:
> Tom,
>
...>
>> I think that captures the intent. However, I see a possible difficulty: differing lengths of the End-user IPv6 prefix for different CEs. We have to add two conditions:
>>
>> -- the configured subnetwork identifier is at least
>>             64 - min (n + o)
>> bits long, where the minimization is taken over all FMRs in the domain
>>
>> -- the subnetwork identifier used with a given calculated End-user IPv6 prefix uses the right-most (64 - (n + o)) bits of the configured subnetwork identifier, where (n + o) is taken from the FMR used for the calculation.
>>
>> Seems simpler just to restrict to the all-zeros value, but please comment.
>
> I can't see how different End-user IPv6 prefix lengths within a domain can work.
>

The End-user IPv6 prefix length obviously has to be the same for all 
nodes using the same BMR, but couldn't, for example, the PSID width 
(i.e., the sharing ratio) vary from one BMR to another?