[arin-ppml] Against 2013-4

Owen DeLong owen at delong.com
Mon Jun 3 23:35:33 EDT 2013


On Jun 3, 2013, at 17:20 , William Herrin <bill at herrin.us> wrote:

> On Mon, Jun 3, 2013 at 7:11 PM, David Farmer <farmer at umn.edu> wrote:
>> Eventually, yes that was the case, and was definitely the case by the time
>> RFC 1366 was published, However it was still technically possible to get a
>> class A even then, look at Section 4.1.
> 
> Hi David,
> 
> Regardless of what might have been possible under conditions now past,
> what actually did happen is that the first IPv4 /8's were preemptively
> assigned without any kind of needs analysis at all. For smaller
> requests (for shrinking versions of smaller) the practice of assigning
> addresses without substantive consideration of "need" continued until
> 1997. That's what -actually- happened.

THis is not true. What is true is that the needs analysis was rather limited and
consisted, essentially, of determining that anyone running an NCP network
at the time would need to run an IP network to move forward.

This seems like a pretty reasonable determination of need at the time.

> At a guess, I'd say Jon's response was never "come back later." I'd
> guess something more along the lines of, "Are you sure you what you
> want to do really takes that many addresses?"  I never met the man, so
> I can't say that with any confidence.

Jon was a very reasonable and personable man. I have no trouble believing
David's characterization that he would likely have told someone whose need
up to 5 years out was 0 hosts that they should come back when that number
was larger than 0.

I can also accept that he would likely have asked the other question in a case
where someone offered 32 hosts in a request for a class B.

Nonetheless, the point is that your documents do show an implementation of
needs-basis as a principle that predates ARIN and those that knew Jon have
consistently stated that he always applied an expectation that addresses were
issued to a specific organization for a specific purpose and should be returned
when that purpose was no longer valid.

While I realize this latter part has not been all that well followed and that there
is a fair amount of money and therefore a fair number of parties pushing for
that part to just sort of fade into obscure history, it does not change the fact that
he stated as much to numerous people on  numerous occasions.

>> Therefore, I believe operational need is a principle that MUST be included.
> 
> I respectfully disagree. Needs assessment is a tool for conservation.
> It's a coarse tool at that... consistently either too permissive or
> too restrictive. And we use it badly, demanding predictions that
> depend on unknowable variables, shortening prediction periods to the
> detriment routing scalability and too frequently failing to check on
> how the predictions panned out.

Needs assessment serves purposes other than conservation. It also works
to ensure that resources are handed out in a fair and consistent manner. It
provides for a certain level of transparency in the operation of the registry
while still protecting some privacy of resource holders.

> "Operational need" is a just a tool that supports conservation for the
> sake of sustainability. If a better tool proves itself, we shouldn't
> hesitate to jettison operational need.

Which can be done through the PDP when and if such a tool presents itself.

If adopted, 2013-4 is not an immutable document. It will simply be an additional
part of the NRPM still subject to the same PDP as the rest of ARIN policy.

Owen




More information about the ARIN-PPML mailing list