[arin-ppml] Draft Policy ARIN-2013-3: Tiny IPv6 Allocations for ISPs

[Michael Sinatra]

On 03/27/13 17:45, David Farmer wrote:
> On 3/27/13 18:00 , Michael Sinatra wrote:
> 
>> Or, to put more bluntly, if ARIN's fee structure is itself creating
>> disincentives for proper IPv6 adoption, then let's go back and (re-)fix
>> that problem.
>>
>> Oppose 2013-3.
> 
> Michael and others opposed,
> 
> What about modifying the proposal to /40, require a minimum reservation
> of /32 (or maybe /28) be held for ISPs that elect for /40 or /36
> allocations, allow subsequent allocations to expansion from /40 to /36
> and then to /32 without evaluating there current IPv6 usage.  Thereby
> ensuring they can grow their allocation in place and allowing policy
> flexibility that enables the fee structure equity that the new xx-small
> category seems to provided.

Sorry to be responding to an earlier part of the thread, but I was on
vacation and lost track of this thread, and you did ask me a direct
question.  I owe you the courtesy of an answer.

The answer to your question is no.  If I start out with a /40 or /36 and
then rapidly grow into a /32 (and can justify the fees), then I am going
to end up with a largely organic addressing plan.  We're giving
incentives for people to cram all of their addressing into a corner of
the total space that they should be using and it will create a really
messy IPv6 deployment.

Again, when did we decide that ARIN should encourage parsimonious *IPv6*
adoption?

One thing that I understand from 19 years working at a large state
university is a phenomenon called "ignoring the obvious solution."  It's
like the doctor who keeps prescribing more drugs to a patient to deal
with the side effects of the drugs the patient is already on, rather
than finding the one drug that will solve the patient's problem with
minimal side-effects.  When we face a big problem that should be tackled
head on, we instead try to work around it.

The big problem is that the fee schedule doesn't make sense.  A
university with 5 /16s (IPv4) and 1 /32 (IPv6) is an "X-Large" in IPv4
and "Small" in IPv6.  (Note that such a fee schedule will definitely not
encourage (L)RSA uptake, BTW.)  It's increasingly hard to cut this new
fee schedule in a way that treats ISPs or end users consistently across
the two address families.  This policy doesn't solve that problem, since
a big part of the problem is on the IPv4 side.

The second issue is that it's not even clear what the goals of the fee
schedule are.  Are we trying to provide incentives for IPv6 uptake?  For
IPv4 return?  Are we trying to charge ISPs and end sites based on their
size?  On their profits?  (Address space allocations measure neither.)

[snip]

> Finally, even if you continue to not support the proposal would you
> support making the changes to the text about for the text to discuss at
> the PPM?

I couldn't quite parse this statement, but I think my answer is "yes."
I really appreciate the original author's work on this and your work as
shepherd, and I think the policy needs to be discussed.  Still, I can't
support a policy that simply codifies the bad incentives that inhere in
the fee schedule.

michael