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

[Owen DeLong]

> a. The end result is not an increase in the number of aggregatable blocks held by the organization.
> 

This is problematic. This would allow an increase in the number of non-aggregatable blocks. While I'm not 100% sure that it is possible to create such a situation, I would rather see us express the direct intent. How about:

a. The resulting number of aggregate retained blocks must not increase.

> b. Whole blocks are returned to the extent practicable.
> 
> c. Partial block(s) retained must conform to current applicable allocation or assignment policies, as to size, alignment, etc…
> 
> d. Block(s) retained are within a single reserved space or aggregate set aside for the organization in the ARIN database to the extent practicable.
> 
> e. All returned block(s) must not be in use by the organization or its customers.
> 
> Comments:
> 
> The author acknowledges the shortcomings of providing an ISP with an allocation of a size that is more traditionally associated with end sites. In order to avoid possible bad effects on the routing table, the author encourages ARIN staff to adopt the same sparse allocation practice as currently exists for larger allocations, ideally even reserving a block as large as the /28 that is reserved for /32s currently.  Note the policy intent of part 1 requires a minimum of a /32 be reserved.
> 
> Part 1 brings ARIN's allocation policies in line with the upcoming fee schedule, with the addition of an expected correction of the XX-Small fee category to "/40 or smaller".  This makes it possible to qualify for each ISP fee category while holding IPv6 number resources and allows expansion up to /32 without renumbering or additional justification as a subsequent allocation.  The selection of a /32, /36 or /40 allocation is only driven by an ISP's own internal business justifications.
> 
> Part 2 codifies and expands upon current practice for selective return in the manner described by John Curran on the arin-discuss mailing list (7-Mar-2013 in 8DA1853CE466B041B104C1CAEE00B3748F9239EA at CHAXCH01.corp.arin.net ). It specifies the generic requirements that should be met for such returns.
> 
> A more practical approach might to figure out a way to apply graduated fees to ISPs at the very small end of the scale using some metric other than prefix size. Fee schedules are outside of the purview of the Policy Development Process; such responsibility lies with the Board should they choose to take it up.
> 
> Summary of community discussion:
> 
> The fundamental argument against this draft policy is that the primary problem being solved is a billing or fee structure issue and not a number resource policy issue in itself.  A significant minority of the community would prefer /32 be the sole minimum allocation size for ISPs and other LIRs, and they feel there is no need for smaller /36 or /40 allocations.  They would prefer to solve the problem with changes in the fee structure rather than contorting number resource policy to solve the problem.  However, there are to many ISPs that fit into the /32 allocation category for the fee level associated with the XX-Small category to be fiscally responsible and sustainable for ARIN. Furthermore, there are no obvious solutions to this problem within the fee structure domain that are fiscally responsible and sustainable for ARIN, especially in the long-term.

I would not call this "A significant minority". Rather, I would say that a significant minority are adamant to the extent they oppose this policy. I believe the vast majority of the community would prefer to see /32 be the sole minimum, but that most are willing to accept the tradeoffs incorporated into this policy. I believe that if we had an effective way to solve the problem through fee structure changes that was acceptable to the vast majority, it would be the preferred solution. I encourage anyone who has ideas of such a solution to please make them known to the community either here or on arin-discuss.  While fees are outside of the policy process, at the point we start contorting policy to address problems with the fee structure, I certainly think it is appropriate to discuss the problems with the fee structure directly in that context.

> Everyone agrees making /36 or /40 allocations to ISPs seems less than ideal from a number resource policy perspective.  However, this is mitigated by ensuring that all ISPs have a /32 available to them without renumbering or additional justification and from a number resource policy perspective the selection of /36 or /40 allocations is completely voluntary.  This allows each ISPs to make the decision to select from a /32, /36 or /40 initial allocation based solely on their own internal business justifications, and eliminating structural disincentives in the fee schedule for IPv6 adoption.  This seems like the best balance available at this time of number resource policy, fiscal responsibility and sustainability for both ARIN and the ISPs that it servers.

I would rather see the reservation set to a minimum of /28, personally, but /32 is better than smaller.

Is there anyone in the community that objects to a minimum /28 reservation (which is the current situation for recipients of /32s)?

Owen