[arin-ppml] simple question about money
michael.dillon at bt.com
michael.dillon at bt.com
Fri Jun 13 11:05:51 EDT 2008
> But as you yourself implicitly recognize, once the addresses
> have been assigned, they don't come back.
Don't put words in my mouth. It is rude and you can't get
away with it on this list.
The fact is that addresses do come back. ISPs regularly recover
customer assigned addresses. ARIN also gets back address blocks
for various reasons. Even IANA has received addresses back such
as the 14/8 block.
> And as you also
> recognized in your reference to the telecom collapse, in many
> cases it is perfectly rational for organizations that have
> been assigned addresses to hold on to them even if they have
> more than they presently need, because they may grow in the
> future, and the process of "demonstrating need" to a central
> planning authority is costly, time consuming and unpleasant.
Part of the process of "technical justification" does allow for
intent to use the addresses. You don't need to buy equipment,
configure it, and give ARIN a tour of the boxes.
> So your comparison of these organizations to car thieves is
> just inflammatory rhetoric that doesn't solve any problem.
Car thieves do not have a justification for owning the car that
they are selling. In the same way, an ISP who has an ARIN allocation
which is more than they need, and which they DO NOT INTENT TO USE,
is breaking ARIN laws/policies. The car thief analogy is perfectly
valid and I expect that most people realize that I am not suggesting
that ARIN policy has the same weight as US law. I am simply pointing
out that an ISP with a surplus of addresses that they do not intend
to use, has the same relation to ARIN policy as someone who finds
a wallet on the sidewalk. If you return the wallet, then you are
good upright citizen. If you keep the wallet, then you are just like
a car thief selling auto parts.
> Good policy pays attention to how people actually behave and
> what their rational incentives are. If your policy is founded
> on the assumption that commercial ISPs and corporations will
> voluntarily give up resources that don't cost them anything
> to hold in order to help other ISPs or organizations, it
> probably won't work, because that assumption doesn't square
> with how people actually behave.
Let's see. I have surplus IPv4 addresses which can only mean that
I am either on the verge of bankruptcy or I have a healthy IPv6
deployment with good customer takeup. Do I give my surplus back to
ARIN in order to help my poor IPv4 brethren or do I hoard it. If I'm
on the verge of bankruptcy, hoarding won't make a difference because
the IPv4 market is shrinking and those who most need IPv4 addresses
are living off their internal supply from customer churn. If I'm
IPv6 then I don't give a damn what the IPv4 ISPs do because they can't
me. I may as well make a big public fuss about handing back my unneeded
IPv4 adresses to ARIN, after having hired some extra sales people to
the rush of new business.
> So, what do you think is a better way to get them back? Get
> people on the ARIN PPML to call them bad names? Or provide
> them tangible economic benefits to transfer them to someone
> who does need them? If ISP A needs addresses and ISP B is
> willing to give up addresses that it doesn't need as long as
> it receives some compensation, why should anyone object?
Because it is selling. It either happens in private or it happens
in an open public market. To create an open and public market is
not easy, and is even risky because of the maze of laws and regulations.
The ARIN BoT could end up in jail if they don't do it right, and
doing it right costs a lot of money. It is easier to not do it
at all and continue with the current policy which discourages
selling entirely. A few wierdos hand over money anyway, but since
there are no guarantees that BOUGHT addresses will work, this has
never picked up any momentum. It has been happening since the mid
1990's if not earlier, but it is a drop in the bucket, not a market.
> I think you misunderstand the issue here. The fundamental
> insight behind the address transfer policy is that there is a
> huge difference between "exhaustion" of _assignments_, which
> is purely nominal, and the exhaustion of the _actually used_
> address space, which we may not have reached.
There is a hierarchy of assignments and they all run out in
the same way. First IANA runs out but the RIRs have some. Then,
one by one, the RIRs run out but in an unpredictable order. Then,
the large ISPs begin to run out PoP by PoP, also in an unpredictable
order. Nobody can know how many addresses are "free" in the ISPs since
"free" is hard to pin down. If I shut down a dialup pool and put those
addresses in limbo for 4 months while I try to cleanse the DUL lists,
are they really "free"? If I notify all small business customers that
when contract renewal time comes up, I will not renew contracts unless
they agree to go to NAT, hand back their address allocations, and live
behind a single statically assigned address per circuit, then are those
handed-back addresses "free".
> There may be
> large amounts of address space that are assigned, but unused.
> The assignees nevertheless hold on to that space because they
> get no benefits from giving it up, but incur real costs and
> risks if they do.
Yes. One benefit is that they encourage more IPv6 deployment which can
lead to more business for forward thinking IPv6 ISPs. The risk from
giving it up is that less people buy IPv6 services.
> Nevertheless, broadcaster industry organizations (NAB,
> etc.) have invested IMMENSE political and financial resources
> into hanging on to that spectrum and preventing a
> reallocation to mobile telecom.
Yes, but in this industry the opposite had happened. We have created
a new kind of spectrum which is so vast that it will be hard for
anyone (except the RIRs) to hang onto enough spectrum that it will
block other users. That new spectrum is IPv6.
And that is the reason why we don't need FCC or other regulatory
oversight, or any kind of IP address markets.
More information about the ARIN-PPML