BCP for multisite multihoming

[David Conrad]

Thomas,

On May 22, 2007, at 8:15 AM, Thomas Narten wrote:
>> Irrelevant for the question being asked, however I believe the IETF
>> and RIRs have spoken that address use efficiency isn't relevant with
>> IPv6.
> This is going to far in the other extreme, from my perspective.
> Address use efficiency is (of course) still important, but the overall
> balance between "efficiency" and "aggregation" has shifted with
> IPv6.

The question at hand is how to implement multi-site multihoming.   
According to existing policies, the highest likelihood of obtaining  
the desired end is to allocate to each site a /32.  In such a case,  
neither aggregation nor efficiency is gained.

In general, given the liberalization of PI allocation policies either  
implemented or being discussed (and likely to be implemented) in all  
RIRs, it isn't clear to me that aggregation remains a goal.

The question is, is this a problem?  I have a gut feeling it is (both  
in terms of aggregation and efficiency), but folks at router vendors  
are telling us not to worry about the routers scaling and folks at  
the IETF are saying not to worry about address assignment efficiency...

Rgds,
-drc
(speaking personally, of course)