[arin-ppml] Using fees to encourage route aggregation

Milton L Mueller mueller at syr.edu
Thu Oct 15 17:36:55 EDT 2009



> -----Original Message-----
> 
> The RIRs role is NOT to afflict routing policy with their judgment of
> how it should be done.
> The RIRs should focus on allocation policies that meet the needs of
> the community and leave routing issues to those that run routers and the

I pretty much agree with that on principle (see proviso below), but it's interesting that on Tuesdays I get excoriated for not understanding how essential RIR central planning of the address space in order to enforce route aggregation and on Thursdays we hear how RIRs have nothing to do with routing. Let's get the story straight. 

To be consistent, the disclaimer above is pretty clearly false. It's clear that many, many things RIRs already do are based on routing considerations. Think of your approach to PI allocations, for example. The whole thrust of post-CIDR policy is to require ip address users to aggregate under providers, which is all about routing. 

Bearing in mind that I am not at this point arguing for an aggregation fee, it is nevertheless important to understand that such a fee doesn't tell people how to route (any more than policies that promote provider-based aggregation do), but sets parameters or incentives around which people base routing decisions. 


> Which view of the routing table would be
> considered the authoritative reference for 
> disaggregation? Would momentary leaks of
> more specifics count against someone if 
> they just happened to coincide with the
> billing sweep through the table? Over what 
> period of time should a prefix have to be
> disaggregated in order to increase the 
> billing for the prefix? Should the bill increase with
> the amount of time the route remains in the table? 

These are the _interesting_ and _useful_ questions that my floating of the aggregation fee idea was meant to elicit. Thanks! 

But your ability to raise such questions, which are clearly unanswered at this point, does not mean that they cannot in principle be answered. Regulatory systems are full of such metrical issues 





More information about the ARIN-PPML mailing list