Re: [v6ops] draft-moreiras-v6ops-rfc3849bis-00

["Antonio M. Moreiras" <[e-mail]>]

On 16/08/13 18:12, Fred Baker (fred) wrote:
> Dumb question.
> 
> I wonder if there is a less expensive way to go about this. By expensive, I mean "choke. you want a /20?". It has been argued that we need something that is shorter than a /32, and that we need something for ULAs. Whatever we do, it needs to be consistent with class examples that need to get typed into operational equipment. There's a lot more that has been said, but that's what I draw out of it.
> 
> What if we shortened 2001:db8::/32 to 2001:db8::/29? I note that the prefix doesn't show up in ftp://ftp.ripe.net/ripe/stats/delegated-ripencc-latest, and the IANA counterpart mentions it only in a footnote.
> 
> We could also delegate fc00:db8::/29, or something longer (/44 perhaps, allowing for the description of several ULA prefixes in documentation but not chewing up as much address space), by the same logic.
> 
> I see the argument, but not for the size requested.

It's a very good question. For our BGP trainings at NIC.br, as they are
today, we need at least a /27 (we have 23 simulated ASs in our lab, each
one using a /32).

Since APNIC could shorten the 2001:db8:: only to a /29, which is not
enough for our classes, we saw as a better alternative asking for a new
prefix. We thought that maybe we shouldn't be so conservative, asking
exactly for what we need at the moment. The /20 is the shortest prefix
allocated to an ISP in our region, and it seemed to be a good choice.

Moreiras.