[nycbug-talk] IPv6 and CIDR confusion

[Jonathan Stewart]

alex at pilosoft.com wrote:
> On Fri, 23 Mar 2007, Isaac Levy wrote:
> 
>> I see, but to reconfigure my question, is this a feature, a bug, or a
>> defined part of the routing specification?
> I'm leaning towards 'bug' or a 'practical constraint which should be taken 
> into account while determining local addressing scheme'.
> 
> I'm not positive if it is documented in any rfc, or not. I think not.

See section 3.1 and 3.6
http://tools.ietf.org/html/rfc2374

The last 64 bits are defined as the "interface identifier" as part of 
the specification.  I know part of the reason for leaving it so large 
was to have globally unique interface specifiers in case that ability 
ever actually becomes useful for something.  I guess they also figured 
64 is "neat" because it splits the address evenly rather than 80/48 or 
something like  that.

Jonathan