[ppml] Efficient allocation? What's that?

[Michael.Dillon at radianz.com]

In some internal discussions I've been asked to demonstrate what is the 
official method for calculating address utilization rates. The discussion 
arose because different people have different ideas of what is a "used" IP 
address. I though it would be simple enough to find a public document that 
would backup the method that I'm using to calculate this, but I was wrong.

Apparently, there is nothing on the ARIN website nor in any of the RFCs 
that defines when an IP address is "used" for the purposes of calculating 
the usage rate. 

If there really is a need to continue minimizing waste of IPv4 addresses, 
then I would think that now would be a good time to tighten up 
definitions. There is less than 40% of the IPv4 address space left. If 
IPv4 addresses really are valuable then it won't be long before we start 
to see hoarding behavior which will lead to premature exhaustion of the 
space. Having clear and unambiguous definitions of things like "used" and 
"justified" will help prevent hoarding.

On the other hand, some people, including me, don't care whether or not 
the IPv4 space runs out in the next 3 to 5 years. It wouldn't be a 
catastrophe because we do have tools like IPv6 and NAT and MPLS to keep on 
running the net. I think we should now be loosening the rules of IPv4 
allocation in various ways because we no longer need to fear the day when 
they are all gone. However, there is still a need to provide clear 
definitions for things like "used" and "justified". Rather than leaving 
the vagueness in, define them in a way that is simple to administer.

In fact, there are two kinds of usage rate to calculate. One is for an 
assignment where you probably want to count the number of device 
interfaces configured as "used". But at the allocation and sub-allocation 
level, you probably want to count an entire justified assignment as "used" 
because that's easy to administer.

Am I making sense with this?

--Michael Dillon