[arin-ppml] 2600::/12 LOA

[Matthew Kaufman]

On 3/29/2014 10:29 AM, Joe St Sauver wrote:
> ...
> Moreover, given BGP route selection rules, I'm not particularly disturbed
> by the presence of that covering announcement: any more specific route should
> immediately be preferred to a broad covering route of the sort employed by
> the IPv6 darknet research effort.

If the BGP listener is able to hear the more-specific route. If the 
more-specific is temporarily missing, the traffic goes to the wrong 
place. If the more-specific is not heard due to length filters at the 
listener, the traffic also goes to the wrong place. In both cases, that 
is production network traffic being diverted from where it should be going.

>
> I believe that ARIN acted properly in supporting this network research, and
> I'd be quite disappointed if ARIN (and other RIRs) discontinued support for
> research of this sort, particularly when carefully done by leading academic
> networking research organizations.

I don't see how ARIN had any business at all supporting research other 
than to issue *unique* address space to researchers. In this case, it 
wasn't unique, and so there was no way ARIN could supply that space 
under the existing numbering policy, and so ARIN should have said "I'm 
sorry, we've issued addresses in that space, so we can't issue you 
2600::/12 as a unique assignment under section 11 of the NRPM... if 
you'd like to announce a covering route in that space, we encourage you 
to talk to the organizations that are already using parts of that space, 
and the organizations to which you intend to announce that route to"

Matthew Kaufman