[arin-ppml] IPv4 Transfer Policy Change to Keep Whois Accurate

[David Farmer]

On 5/11/11 22:09 CDT, Mike Burns wrote:
...
> I'm sure there are many more that I cannot think of. I agree with you
> that most buyers will have need, and I agree with you that most buyers
> will see the value of maintaining a valid ARIN whois record pointing to
> their authority.

Unfortunately, most rules are meant to enforce something on those who 
are not necessarily acting reasonably.  Above you save "most buyers will 
see value" what about those that don't.  Or, what about those, that see 
a value in ensuring that their competitors can't get addresses and that 
the whole system doesn't shift to IPv6.  I don't want to see something 
like that happen and I'm not entirely sure it can't.

> But the policy in APNIC was changed to remove needs requirements for
> transfers for the same reasons I am requesting its removal here.

There are other policy differences between APNIC and ARIN then just 
needs or non-needs based transfers that are relevant to the changes due 
to IPv4 run-out. There is also a fundamental difference in the way APNIC 
and ARIN plan to operate the pool of addresses they have left, APNIC is 
reserving their whole /8 and allowing each organization a single /22. 
The ARIN community has reserved /10 of IPv4 for IPv6 purposes, but is 
allowing need to driver the run-out of the rest of the pool, limiting to 
3 month need for fairness with new entrants and to reduce the 
competitive disparity created by IPv4 run-out.  Additionally ARIN added 
officer attestations, to ensure organizational accountability for 
requests, I believe this will strongly suppress the tendency for ARIN to 
have a run-out curve like APNIC had during the first 4 months of 2011. 


So there are a number of differences in how ARIN and APNIC are 
approaching the fundamental changes that are occurring, and it would be 
a mistake to evaluate only one of these issues out of context with the 
others.  I'm not trying to say APNIC did or is doing something wrong or 
ARIN did or is doing it right, but there is more to the picture than 
just needs or non-needs based transfers.  If ARIN is going to change to 
a non-needs based transfer regime lime APNIC we may also need to change 
many more parts of policy as a consequence.

The point here is that ARIN's policies are consistent, if you have need 
you can have addresses, be that from the free pool or from transfers. 
Much of the ARIN community didn't like to idea of addresses sitting at 
ARIN, if there was need for them.

APNIC is also consistent, in that once they get down to the last /8, it 
doesn't matter if you need addresses you only get a /22 once from the 
last /8 pool, and you can transfer without need.

I believe it would be unwise for ARIN community to change only one part 
of its policies to emulate APNIC.

> My policy proposal also has the benefit of incentivizing legacy
> resources to come under RSA, and it serves to even the playing field
> between the disparate rights of legacy versus non-legacy holders.
>
> And my underlying point is the obvious one, that the very act of paying
> for address space is a very good indication of need, or at least
> perceived need on the part of the buyer.

I think it is a good indication of perceived value in all cases, it is a 
indication of need in some cases, but not in all cases.  If you want to 
corner the market on IPv4 there maybe sufficient value to someone to do 
so, but that is not a valid need.  And, this is the case where your 
analysis breaks down.

That said I'm not certain that current criteria for evaluating need will 
even be valid once there is no longer a free pool.  The basic idea that 
you get a years worth of need based on your past years consumption of 
addresses assumes the existence of a free pool.  Once we get more than a 
year into a market-like transfer system, what you used last year has 
more to do with how much money you were willing to spend that what your 
actual need was.  And if the price was high the previous year and if you 
simply evaluate your need by how much addressing you used last year when 
the price was high that isn't necessarily a true representation of your 
actual need.

So I am equally skeptical of maintaining the current needs based system 
as is, as I am of completely abandoning the concept of a needs basis. 
Right now I'm thinking allowing transfers up to some size limit over 
some period of time without justifying need more than putting up the 
cash to do it, on the premise that within some safety limits it might be 
a good idea to let the market work things out.  However, if you want to 
transfer more than those limits you can, but you must justify your need 
beyond those limits.  Either based on your previous years usage or based 
on the fact that you have put previously transferred addresses into 
documented use.

Just throwing some numbers out, I'm thinking up to the equilivant of a 
/12 (or a little more than 1 million IPv4 addresses) of transfers 
received within 12 months per organization without documenting need. 
There needs to be a mechanism that prevents gaming by creating multiple 
organizations by the same real organization, maybe this is another place 
for officer attestations.

For most organizations a /12 is more or less unlimited, but that is 
probably not enough to corner the market of IPv4 addresses and anyone 
who really needs more than a /12 can probably justify it without much 
more hassle then now.  If they really need more than a /12 they are 
probably quite adept at working with ARIN already.  I think this would 
give the invisible hand enough room to do its magic, without allowing 
someone to corner the IPv4 market and destroy the Internet as we know 
it.  Depending how the actually market progresses, someone could maybe 
still corner the IPv4 address market, but if they can do that with a /12 
then it will really be the time for everyone to move to IPv6.

What I mean by "destroy the Internet as we know it", is anyone big 
enough, or that anyone thinks is big enough, to stall the transition of 
the Internet to IPv6, getting enough IPv4 addresses to attempt to do so. 
  I'm not sure anyone is big enough, but I'm just not sure no one is.

In my view this doesn't abandon needs basis, it significantly changes it 
based on new the realities we are facing.  We are using the market to 
determine need up to a cretin extent, and beyond that requiring 
documented need.

-- 
===============================================
David Farmer               Email:farmer at umn.edu
Networking & Telecommunication Services
Office of Information Technology
University of Minnesota	
2218 University Ave SE	    Phone: 612-626-0815
Minneapolis, MN 55414-3029   Cell: 612-812-9952
===============================================