Re: [74attendees] The great emphasis on IPv6 - a positive look

[Brian E Carpenter <brian.e.carpenter@gmail.com>]

On 2009-03-29 05:18, Jeff McAdams wrote:
> Philip Homburg wrote:
>>> Fortunately 2002::/16 orlonger is pretty well filtered.
> 
>> You are saying that if an ISP gets a /32 from a RIR it is no problem,
>> but a
>> 2002::xxyy/32 suddenly is a problem?
> 
> Yes, because 2002::xxyy/32 or whatever, as a direct import of the IPv4
> routes results in much more routing table explosion than an ISP getting
> a /32.
> 
> The ISP that I used to work for is relatively small and has to inject 7
> IPv4 routes into the DFZ.  When they go with IPv6 (alas, they haven't
> started that work yet, to my knowledge...I keep poking, though), they'll
> have 1 that will provide them with IPv6 space for pretty much any
> anticipated needs long long into the future.

Of course. But as Jeroen said, please read RFC3056. The only more-specifics
under 2002::/8 that make any sense whatsoever are /48, and there would
*potentially* be about 300,000 of them today, one per announced IPv4 network.
More specifics under 2002::/8 are very much not allowed. Mr R. Bush gave
us a very hard time about that before RFC3056 was approved.

My point was that the same danger applies to a NAT64 well-known prefix.
unless we carefully avoid it.

    Brian

     Brian