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

[Michael Sinatra]

On 04/06/13 19:31, John Curran wrote:
> On Apr 6, 2013, at 12:00 PM, Michael Sinatra <michael+ppml at burnttofu.net> wrote:
> 
>> The second issue is that it's not even clear what the goals of the fee
>> schedule are.  
> 
> <https://www.arin.net/fees/faq_fee_schedule.html>
> "   
>    ARIN adopted the new Fee Schedule in order to:
>        • Ensure members receiving comparable services are paying comparable fees where feasible
>        • Meet the community's expectations for new and future services such as RPKI
>        • Maintain and reduce, where possible, cost for smaller ISPs
>        • Provide a revenue model based on long-term expenses
> "
> (This is also very similar to the information in community 
> consultation held in October 2012 on the Revised Fee schedule.)
> 
> Note that the present fee schedule does not provide an xx-small 
> size category at all, and the fees associated with the current 
> x-small category are also lowered by the Revised Fee schedule.

That doesn't really answer my question.  My question revolves around the
following *types* of questions:

o Is the fee schedule intended to be based on fairness?
o Is the goal to reduce complexity at a (reasonable?) cost of fairness?
 Or increase fairness at a reasonable cost of complexity?  If I suggest
that we tweak the fee schedule, am I going against one of the
fundamental goals of not making it "too complex" by some definition thereof?
o Is fairness determined in terms of revenue, capitalization, or just
address space?
o How do we define terms like "comparable service."  Given that ARIN
provides registration services, not addresses, how *should* "comparable
service" be defined.

>> Are we trying to provide incentives for IPv6 uptake?  For IPv4 return?
> 
> We're trying to provide more more uniform fee categories and
> introduce lower fees for both x-small and xx-small ISPs as 
> noted in the Revised Fee schedule faq.  
>
>
>> Are we trying to charge ISPs and end sites based on their size?  
>> On their profits?
>>  (Address space allocations measure neither.)
> 
> ARIN has always had fees based on relative size of organizations,
> generally based on the total address space held.  This is the case
> with both the present fee schedule and the Revised Fee Schedule. 

That may be true, but where is that codified and/or hashed out?  What is
that basic philosophy discussed and reviewed?  I am not saying it needs
to be right now, but I have a hard time understanding why we need to
contort the NRPM to patch over bad incentives in the fee schedule.

Moreover, that standard is called into question by the fact that ARIN
charges based on the larger of the two address family allocations, with
no regard to the situation where there are radical differences between
IPv4 size and IPv6 size.

What I am trying to say here is that there is huge amount of slack
between the actual implementation of the fee schedule and what it
appears to be trying to do.  So let's stop pretending that policy ought
to be beholden to the vagaries of the fee schedule and let's be willing
to tweak the fee schedule where necessary to provide the right
incentives, since we acknowledge that the fee schedule is imperfect.

That said, I disagree with the comments section of the proposal that
there is no obvious solution to the problem that still provides for
sustainability of ARIN.  One obvious possibility is to treat a /32 IPv6
allocation as a special case for ISPs: If you have both IPv4 and IPv6
allocations, and your IPv6 allocation is a /32, then your fees are based
on your *IPv4* allocation, even if it drops you into a lower fee
bracket.  If we're reserving the /32 indefinitely anyway, then I don't
think that creates a sustainability issue for ARIN.  It's still a single
allocation, so it's not costing ARIN anything more to provide a /32 but
simply charge for only the smaller IPv4 allocation.

Obviously, this would have to be revisited at the time that ARIN ceases
to provide IPv4 registration services.  But I suspect that at that
point, it would be time to revise the fee schedule anyway.

michael