[arin-ppml] Policy discussion - Method of calculating utilization

[Jimmy Hess]

On Fri, May 2, 2014 at 8:04 PM, Jeffrey Lyon
<jeffrey.lyon at blacklotus.net> wrote:
> Jimmy,
> I would not support scaling this beyond 80% except at the larger
> allocation levels (eg. perhaps /17 and shorter, aggregate).

The essence of it is, that the 80%  utilization criterion is ancient,
and before resource scarcity, before technical improvements such as
unnumbered interfaces, before host-based virtual hosting,  and  /31s
as point to point links.  And  a greater utilization requirement may
slow exhaustion.

The number allows plenty of extra wiggle-room luxury, in addition to
the occasional oddball patch which can't be allocated even in an
efficient planned address allocation strategy,  which the free
resources don't exist for, any longer.

80% in the aggregate criterion is unfairly strict against resource
holders  that have  a small total allocation size.  And unfairly
lenient against resource holders that have a large total allocation
size.

And removing the per-allocation utilization requirement would serve to
exacerbate this problem.

For example, under the current rules a holder of a /10 equivalent, can
 call their existing allocations "efficiently utilized",   even if
there are most recent allocation, an entire  contiguous /20s  has been
completely untouched and unused.

Whereas the resource holder that has a /20,   cannot have a single
/23's worth  untouched.

--
-JH