Re: [v6ops] [Fwd: I-D Action: draft-ietf-v6ops-icp-guidance-03.txt]

[Brian E Carpenter <[e-mail]>]

Nick,

On 31/08/2012 13:54, Nick Hilliard wrote:
> On 31/08/2012 12:12, Tore Anderson wrote:
>> The document continues in the same vein: «only large content providers
>> can justify the bother and expense of obtaining a PI prefix». Can't say
>> I agree.
> 
> +1.
> 
> PI is easy to get and there are plenty of smaller organisations who depend
> on it for their business requirements (at least for ipv4).  

It's a solution, not a requirement. It may well be the *only* solution in
IPv4, but this is about IPv6 which needs to be much more scaleable than
IPv4 to make any sense.

> There are no
> substantial issues with routing PIv6 address space, because the minimum
> assignment size is /48, which corresponds with the RIRs' recommendations on
> minimum prefix filter length.

That is, however, just a pragmatic choice and with the *current* size of the
IPv6 RIB, simply not an issue. The point is that we (the IETF) need to set
things rolling in a way that will scale to an IPv6 RIB many times bigger
than the current IPv4 RIB. Recommending PI unconditionally is, in my
opinion, highly irresponsible in that context.

As document editors we will of course do what the WG chairs determine
is the WG consensus. As an individual contributor, I will continue to press
for recognition of the PA approach.

    Brian

> 
> Nick
> 
>