"Recommendations on IPv6 Route Aggregation" (was: Re: IPv6 multihoming)

[Pekka Savola]

On Tue, 8 Feb 2011, Rob Evans wrote:
>> Rob, please correct me if I have missed the more recent version of this
>> document.
>
> You haven't.  Equally there isn't that much to document, although
> curiously this is now the fourth time this has come up in the last
> couple of weeks.
>
> In my experience the number of people filtering on /32s appears to be
> diminishing, /48s are widely accepted (for unverified values of
> "widely"), and so "prudent subdivision" of a /32 is likely to become
> the norm where needed, especially as the proposal to allocate
> additional /32s for routing reasons did not reach completion in the
> RIPE region.

I would take the following into account in revision:

Whether you accept /48 is not a one-bit decision.  Specifically:

- quite a few still filter more specifics from PA ranges
- quite a few still reject /48's set aside for IX fabrics
- almost everybody accepts /48 (or sometimes shorter) in designated PI 
ranges

So, if you do break a PA /32 block down to smaller pieces, you should 
expect some to filter more specifics.

BCP in this field should include a statement that (also per address 
allocation policy), the whole aggregate MUST still be advertised as a 
single prefix even if you advertise more specifics. The onus is on the 
advertiser to make sure connectivity works to to the subblocks through 
the aggregate.

-- 
Pekka Savola                 "You each name yourselves king, yet the
Netcore Oy                    kingdom bleeds."
Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings