[arin-ppml] Set aside round deux

Benson Schliesser bensons at queuefull.net
Fri Jul 30 11:48:06 EDT 2010

On 30 Jul 10, at 8:54 AM, Joe Maimon wrote:

> Policy that dictates that requests for global resources must demonstrate a global need for those resources is not out of order.

Agreed in principle, but "global need" doesn't equal "Internet routed".  E.g. A provider (of information or communication services) might run a global network that interconnects with many other networks (some of which may run RFC1918 internally) but isn't present in the global routing table, as seen by Tier 1 ISPs.  That network has the need for globally unique address space, just like any other global inter-network.

> Demonstrating justified global need should not nearly be as easy for resources intended to not being addressable on the global network as otherwise.

How would you propose that "demonstrating justified global need" is different between big-"I" and little-"i" internetworks?  Just because a prefix exists in the DFZ routing table doesn't mean that anyone can reach those addresses; firewalls and other methods of policy enforcement are ubiquitous.  Would a network like the one I described above hold more esteem simply by adding their prefixes to the global routing table, regardless of connectivity?

In short, I think it's a messy situation if any RIR starts discriminating based on the way addresses are used.  Clearly, it's a good idea to verify that "globally unique" is a requirement before providing an allocation of global resources.  But beyond that, I think we have to accept that not every organization has the same perspective as ourselves.  The RIR exists to foster number resources, not define the Internet.


More information about the ARIN-PPML mailing list