[arin-ppml] ARIN-prop-178 Regional Use of Resources

[Owen DeLong]

Sent from my iPad

On Jul 16, 2012, at 11:59 AM, "Tony Hain" <alh-ietf at tndh.net> wrote:

> Owen DeLong wrote:
>> On Jul 15, 2012, at 1:22 PM, Jimmy Hess wrote:
>> 
>>> On 7/15/12, David Farmer <farmer at umn.edu> wrote:
>>>> creates an unacceptable danger that people may implement IPv6 using
>>>> separate unaggregatable prefixes from multiple RIRs.  When they may
>>>> be
>>> 
>>> Prefixes from other RIRs are aggregable along the Geographical address
>>> assignment hierarchy lines.
>>> 
>>> Aggregation of all prefixes to  outside  RIR  region X    to one route
>>> for external Region X.
>>> 
>>> "One global prefix"  is in conflict with better efficiency  achieved
>>> through that method.
>>> 
>>> --
>>> -JH
>> 
>> That's not universally true. If you operate a highly efficient backbone
>> network, you actually get greatest efficiency from accepting your traffic
> on a
>> hot potato basis and controlling it across your own network as soon as
>> possible. Advertising a single global prefix facilitates that quite well.
> 
> This highlights why it is insane to restrict a resource to 'only for use
> within the allocating region'. 
> 

Which the proposed policy does not do.

>> 
>> The backbone operator should have the choice and this proposal
>> accomplishes exactly that... Either number your global network out of a
>> single RIR (mostly) or number your network out of each region (mostly)
> with
>> room for some flexibility to accommodate the differences between
>> theoretical optimizations and real world deployments.
> 
> And the 'flexibility' creates ambiguity, which in turn leads to conflict,
> which in turn makes the proposed language unenforceable. Simplify and make
> the requirement 'must justify existing global resource use against policy of
> the RIR being asked'. 
> 

I think it is time to agree to disagree on this point.

Owen

> Tony
> 
>> 
>> Owen