[arin-ppml] Initial ISP Allocation Policy

[Owen DeLong]

On Jul 17, 2013, at 3:51 PM, "Alexander, Daniel" <Daniel_Alexander at cable.comcast.com> wrote:

> Owen,
> 
> Your reply sounds circular to me. Is the process of explaining resources
> complicated because they are end users or ISP's, or are they complicated
> because the policy requirements have made them so by creating all the
> distinctions? All the issues you mention are because the policies are
> different. I'm really not trying to give a cheap answer. It is a question
> to the larger point of the email.

I don't believe that the current process is complicated. I believe that if we were to try and shoehorn everything into one process, it would be complicated. My intent was to express how the current end-user policy does not map well to an ISP/LIR and how the current ISP/LIR policy would not map well to an end-user.

If you have language that you think could be used to meet both sets of users' needs, then I would be very interested in seeing that. I have tried to think of language that could do that and it seems an insurmountable task to me, but I am perfectly willing to accept that you may be more clever or better able to tackle that problem than I am.

> I was more curious about everyones thoughts on the high level concept.
> Should the distinctions in network definitions be applied to how resources
> are justified, tracked, or both?

I think they are at least necessary in terms of justification. I think that because of the fee implications, they should also apply to how they are tracked. Further, I think it would be very punitive to end-users to make them start paying the same fees as LIRs/ISPs.

> To make a change like this would require multiple proposals. David's
> initial ISP allocation question is a good place to start.

I do agree that the initial ISP allocation policy for IPv4 probably may be worth tweaking, not because I think that longer prefixes should be issued to ISPs as David proposes, but because I believe that we should remove some of the deadly-embrace conditions that exist in that policy and perhaps make it somewhat easier to qualify for a base /22 regardless of an ISPs immediate ability to utilize all of the space within 3 months.

However, for the free pool, I don't see much point in changing IPv4 policy at this point. However, since this would also affect a new entities ability to obtain resources through the transfer process, I would suggest that effort in this area may still be worth while.

Owen

> 
> Dan Alexander
> Speaking only for myself
> 
> 
> 
> On 7/17/13 1:13 PM, "Owen DeLong" <owen at delong.com> wrote:
> 
>>> The point I'm trying to make is that the distinction between the network
>>> definitions may apply to how the resource assignments should be tracked,
>>> but they should be less of a factor in how the resources are obtained.
>>> This becomes even more relevant in todays world where transfers will
>>> soon
>>> outnumber assignments/allocations, and we see organizations redefining
>>> what they are in order to save on member fees, or meet different
>>> qualifying requirements. I think there should be a single set of
>>> parameters regarding minimum allocation size, timeframes, utilization
>>> requirements, and qualifying requirements.
>> 
>> Respectfully, I disagree.
>> 
>> The process of explaining how one will use resources within a network
>> over which one retains exclusive control can be and often is quite
>> different from the process needed in order to explain how one plans to
>> delegate resources to other organizations and networks outside of one's
>> exclusive control.
>> 
>> Having made applications under both policy frameworks and having been
>> active in authoring policies on both sides of the spectrum, I think that
>> the needs of these two different communities in terms of how they justify
>> resources are, in fact, quite distinct.  I suggest this exercise for
>> anyone who doubts this is the case...
>> 
>> Imagine you are an IPv6 end-user wanting to apply under NRPM 6 for
>> resources. You have a single site and are hoping to obtain a /48. Now,
>> read through the LIR/ISP policy for IPv6 and imagine trying to write your
>> justification under that policy. It makes no sense whatsoever.
>> 
>> A little (very little) less nonsensical... Imagine you are an LIR/ISP
>> with 3,100 serving sites located throughout the ARIN service region. Your
>> largest serving site serves 40,000 customer end-sites. Now review the
>> end-user policies in section 6 and imagine applying under those.
>> 
>> Yes, there are many organizations which are hybrids of these two
>> environments these days. In most cases, those organizations are better
>> served by the LIR/ISP policy and should apply under those policies. That
>> is one of the reasons I suggested that we mostly let organizations
>> self-categorize and give staff guidance and support stating that if an
>> organization does not clearly fit within the end-user definition, they
>> should be treated as an LIR/ISP.
>> 
>>> Somewhere we blurred the lines of how resources are allocated with how
>>> they should be tracked and the result is the dichotomy you mentioned. I
>>> also feel that this creation of classes is contrary to the stewardship
>>> our
>>> RIR policies should provide. As this discussion continues to develop I
>>> would like to see the distinction between PI and PA allocations and
>>> assignment requirements be removed. I would suggest they all be resource
>>> allocations that are given to a network operator, with possibly
>>> different
>>> requirements as to how they should be tracked.
>> 
>> I simply don't see how that could be workable. Could you propose policy
>> language you feel would adequately implement such a structure?
>> 
>> Thanks,
>> 
>> Owen
>>