[arin-ppml] Draft Policy ARIN-2020-3: IPv6 Nano-allocations

[Michael Sinatra]

I oppose the policy *as worded* for the reason(s) expressed below.

On 2020-03-24 10:20, ARIN wrote:
> On 19 March 2020, the ARIN Advisory Council (AC) accepted
> "ARIN-prop-285: IPv6 Nano-allocations" as a Draft Policy.

[snip]

> Draft Policy ARIN-2020-3: IPv6 Nano-allocations
> 
> Problem Statement:
> 
> ARIN's fee structure provides a graduated system wherein organizations
> pay based on the amount of number resources they consume.

The wording of this problem statement is, IMO, incorrect.  ARIN's RSA,
Section 7, explicitly states that number resources are not the property
of the holder.  As such, I don't see how a reasonable person can view
them as being "consumed" by the registrant.

This may seem like a semantic issue, but part of the problem is that the
statement is really correct inasmuch as the ARIN fee schedule does scale
based on the amount of IP addresses registered to an organization,
rather than based on the registration services actually provided by ARIN.

The elephant in the room is that ARIN's fee schedule treats IP addresses
as some sort of scarce property rather than properly recovering the
costs of the services ARIN provides.  And yes, I understand how hard
this is: I was on an ARIN fee schedule committee about 5-6 years ago.
That experience taught me that creating a fee schedule that properly
accounts for the costs that accrue to ARIN for the services it provides
to the community is really hard.  But it also gave me an inkling that
there might be a better way than treating IP addresses as consumable
property and charging for their scarcity.  The alignment between that
and ARIN's operating costs is just too tenuous.

The policy is represents good intent, and from a technical perspective,
I don't have a problem with it.  /40s-/48s appear in the IPv6 DFZ as a
result of direct assignments obtained by end sites via the "PIv6"
policies ratified by ARIN and other RIRs over a decade ago (which I
supported).  I agree that ship has sailed.  I also acknowledge, as I did
during the PIv6 debates and despite my support for PIv6, that the costs
of micro-allocations are actually borne by the operator community who
must provision additional FIB capacity to support bigger routing tables.
 I would state that routing capacity has arguably scaled better than we
thought back in the mid-00s, and we have dealt with that, as is our job
as operators.

But it underscores that ARIN's fee schedule doesn't account for such
externalities.  One could argue that it shouldn't, but if that's the
case, why does it try to account for the externality of scarcity of
number resources?  Why doesn't it just account for ARIN's costs?

And more broadly, how many more band-aids are we willing to put on the
current fee structure?  If the answer is "oh, we can handle at least a
few more band-aids," then fine.  Change the word "consume" to something
else, and prepend the problem statement with "For better or worse," and
we're done.  But I feel like we should at least think about the larger
question.

michael