[arin-ppml] Utilization policy is not aggregate

[Brett Frankenberger]

On Fri, Nov 16, 2012 at 06:22:53PM -0600, Jimmy Hess wrote:
> 
> It's essentially like saying  "You used your previous allocation _SO_
> efficiently, that we will give you a bonus,  and  let you not use the
> next allocation so efficiently, and still obtain more resources."

Or it's like saying "because we value form over function, you can't
have any more addresses now, unless you waste time reassigning some of
your customers from your previous allocation to your next allocation".

> Instead it should just be  "lesson learned"  for the applicant;  if
> you ever actually exceed 80% utilization, stop allocating from that
> block, start allocating from the new one, and there is no need to
> change policy.

So you believe that we should encode in policy a preference for making
assignments from newer allocations over older allocations, and reward
providers who do that?

An applicant who has used 95% of his first /16 and 75% of his second
/16 has more efficiently utilized his space than an applicate who has
used 81% of his first /16 and 81% of his second /16.  Yet we are
willing to give more addresses to the second, less efficient,
application, but we will not give more addresses to the first
applicant.

Moreover, the first applicant *can* get more space without any
increase in utilization, simply by renumbering customers equivalent to
5% of a /16 from the his first /16 to his second /16.  Should policy
really be motivating/rewarding this sort of behavior?

A counterargument is the repetitive assignment possibility. Someone who
is at 95% of a /16 justifies and receives a /20.  Without any
utilization of it whatsoever, they are at 89% and can essentially
submit the same paperwork and get another /20 the next day.  However:
(a) even under current policy, they can do that, as long as they move
80% of a /20 of customers from their original /16 to their new
/20, and (b) in the situation I describe, the real cause of the
potential for repetitive assignment is the applicant's voluntary delay of
their request: they could have requested when they got to 80% of their
first /16, and then their /20 would have placed them below 80% in
aggregate.

The goal is efficient utilization of a limited resource. But current
policy treats entites who are equally situated in terms of existing
space, utilization of existing space, and projected need going forward,
differently, based on *which* of their existing addresses they have
used.

If the government started taxing purchases paid for with a $20 bill
differently from purchases paid for with two $10 bills, we'd all think
that's silly.  But we are treating providers who made allocations from
their older assignment differently from providers who make those same
allocations from their newer assignments.

     -- Brett