[arin-ppml] 2013-5 LIR/ISP and End-user Definitions

[David Huberman]

Hi Bill!

Thank you for the reply.  You wrote:

> I agree that we should work toward eliminating the distinction.
> It seems like this would declare a default so that instead of trying 
> to find a "best fit" ARIN would say that "everyone not clearly an 
> end user is the default." Over time, end-user can be tightened up 
> or eliminated leaving a single functional user type. Is that not a
> reasonable step in the right direction?"

I'm sorry to disagree my friend, but I think it is bad policy text. Here's my thinking:

1) I think the proposal is totally no-op.  I was an ARIN hostmaster from 1999-2000 and again from 2003-2013.  I'm thinking back to the thousands of customers I worked with, and I can only think of one type of network operator who fit this scenario: the university.  Universities are the clearest example I know of for the question, "what am I?", and the staff elegantly solved that problem a long time ago by allowing universities to self-select if they want to be an ISP or an EU.

I can't think of a single customer who had such a big problem with classification that staff would have to invoke this new policy section.    My experience has been that when there's a disagreement, the customer and ARIN talk about it.  They go back and forth, giving each other their perspective, and make a decision accordingly.  It's a conversation, and I can't think of a single example where ARIN staff waffled AND the customer was unhappy with the result.  [I completely concede that I may be forgetting an instance or three, but that's kind of my point: it's a total one-off scenario.]

2) I think the proposal's solution ("if the customer and ARIN can't decide whether the customer is an ISP or an EU, then ARIN shall choose ISP") is actually harmful.  For a policy construct like 2013-5's text, I think the correct prescription would be selecting EU.  Current ISP policy (for both initial and additional requests) is much more punitive than the EU policy.  Off the top of my head, the major differences are:

- ISPs have to be using upstream space already before qualifying for an initial allocation.  EUs do not.
- ISPs only get a three month supply of addresses.  EUs get twelve months.
- ISPs have to SWIP.  EUs cannot SWIP.

So, I re-state my original response. The real problem here is that the terms "ISP" and "end-user" don't work in 2013.  ARIN has setup internal procedures for how to classify requestors, and a lot of requestors either don't understand these rules intuitively, or disagree with them.  The problem isn't with ARIN's procedures on classification; the problem is with the two-class delineation.  

With regards,
David