[arin-ppml] Fee Philosophy (was: Re: Draft Policy ARIN-2013-3: Tiny IPv6 Allocations for ISPs)

[Owen DeLong]

>> 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.
> 
> No NRPM change is needed because of the Revised Fee schedule; fees under 
> the new schedule will be lower for smallest ISPs in any case.
> 
> The question is whether the community also provide support for a xx-small 
> category which is even lower ($500/year) but distinguished by only a /40 
> IPv6 allocation.  This is being discussed in Draft Policy ARIN-2013-3, and 
> while it is enabled by the Revised Fee schedule, it is an independent item
> for the community to consider and can be adopted or not based on its merits.
> 

No, that is not the only question. Because of the situation created by the
restructuring of the fee schedule, we have several negative incentives
newly created. It is true that we can choose one negative incentive[1]  by
not modifying the NRPM. Policy 2013-3 attempts to replace that negative
incentive with a different negative incentive[2].

No matter what, the fee schedule as it currently stands creates negative
incentives and I believe that is what Michael is calling into question here.

[1] The fee structure and policy as currently adopted will create the
negative incentive for organizations in the xx-small IPv4 category to
not deploy IPv6 in order to save $500/year.

[2] The combination of the adopted fee structure (assuming it is modified
to /40 instead of /48) and proposal 2013-3, if adopted would replace [1]
with the negative incentive for those providers in the XX-Small category
to obtain massively undersized allocations in order to save the same
amount of money, allowing them to deploy IPv6 without additional
address space cost, but very likely inflicting undue limitations on the
space issued to their customers. This negative incentive already
exists in both policy and the fee schedule for the X-Small category
in the form of support for /36s.

>> 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.
> 
> Correct, and this has been covered on this mailing list before
> <http://lists.arin.net/pipermail/arin-ppml/2013-March/026396.html> -
> 
>  "that is precisely the benefit of the revised fee schedule;
>   every size ISP category now includes both IPv4 and IPv6, so every
>   ISP can add an IPv6 allocation and see _no_ change in fees at all.
>   (This does mean that we can get ISPs for whom there is a "mismatch"
>   between their IPv4 and IPv6 allocations, and they end up in a higher
>   category, but to be truly fair we'd need to have distinct proportional
>   fee for each of IPv4 and IPv6, and that's exactly what you don't want:
>   any addition of IPv6 means an additional fee.)"

Another alternative would be to base all revenues on IPv4 until a flag
day to be determined by the board at a later date with at least 12 months
notice to the community after which flag day, all revenue calculations
would shift to IPv6. Organizations which did not hold IPv6 after the flag
day would have the option of maintaining their IPv4 records either by
obtaining an IPv6 allocation or by continuing based on their IPv4 fee
category.

I'm sure there are also other possible ways to avoid some or all of the
issues of mismatch. In fact, have we considered the possibility that
instead of MAX(IPv4,IPv6) we charge based on MIN(IPv4,IPv6)?
I'm not sure that would be a good idea, either, but I think it may not
have ever been considered and I would be interested to see what
the implications of such a structure might be.

Owen