[ppml] ARIN IP conservation and FREE IP Addresses

Jeremy H. Griffith jhg at omsys.com
Sat Oct 6 20:23:06 EDT 2007

On Sat, 6 Oct 2007 18:58:30 -0400, "Howard, W. Lee" 
<Lee.Howard at stanleyassociates.com> wrote:

>> >The "Local Internet Registries" do. If you're a SOHO or hobbyist 
>> >customer its not unusual to pay anywhere from $10 to $60 per 
>> IP address 
>> >per year for as much as 32+1 static IP addresses.
>> >
>> >Why should a Regional Internet Registry, one step up the food chain, 
>> >not charge the LIR's $1 per IP address per year? Fair's fair, right?
>> >They charge the end users per-address so why shouldn't ARIN 
>> > charge them the same way?
>Because two wrongs don't make a right.

Why is charging according to usage a "wrong"?  It's the
model used for virtually every transaction in the economy.

>> You're onto something.  All the blather about deadbeat legacy 
>> holders is pure misdirection.  The real issue is with the 
>> large holders, under RSA or not, who have *no* reason to use 
>> their assignments efficiently as long as the incremental cost 
>> of new IPs is zero.  In fact, with v4 exhaustion on the 
>> horizon, it's better for them to remain inefficient to the 
>> end of the free pool, since that will give them breathing 
>> room at that point.
>The large holders who are active have to show efficient 
>utilization, which means assignment to customers, in order to get
>more space.  If they don't (and I speak from experience), they
>don't get more address space, which means they stop turning up
>new customers.

I'm not clear on your point here.  There are many ways of
looking at "efficient", and all involve a cost/benefit
anaylsis.  Efficient when marginal cost is zero is different
from efficient when marginal cost is *something*.

>> So, to change this, we have to work the system.  And that 
>> means, as this thread has made blindingly clear, making every 
>> IP cost money, per year.  
>> the fees have to be enough, at the top end, 
>> to inspire more efficient use.
>Is it just me, or does it seems like everybody wants to raise
>fees on somebody else?

It's just you.  ;-)  I think everybody is trying to find
some way of using fees, among other strategies, to deal
with a real problem.  I've seen as many posts advocating
*lower* fees (for IPv6) as increases.  That too is an
appropriate approach.

>I'm not convinced that this particular stick would have
>the intended effect.

Why is it a stick?  It's simply a policy which has worked
rather well where it has been used, specifically in
California for energy and water, both in short supply.
So from experience, not from political belief, it seems
worth trying for the shortage ARIN is facing.

>> The fees have nothing to do with ARIN's costs, and everything 
>> to do with the public policy needed to keep the Net 
>> functional. 
>The fees were set by figuring out roughly what ARIN's budget
>needed to be, and setting up some simple tiers to meet those

Exactly.  And that's the problem.  The "simple tiers"
have a discount structure meant to maximize use, which
may have been fine in early days but isn't any more.

And exhaustion is certainly going to have an impact on
ARIN's budget.  Could be a major decrease (no applications
to process any more, so sorry), or an increase (paying
for demo projects to get IPv6 out there more, or for 
investigators and attorneys to claw unused space back).
The one thing it *won't* do is stay the same.

>> Applying this idea to ARIN, what if we had a basic fee for 
>> any IP addresses of $25/year, which would include the first 
>> 256 addresses (such as a legacy Class C, /24).
>> Then beyond that:
>>   /22     $ 0.20
>>   /20     $ 0.35
>>   /18     $ 0.50
>>   /16     $ 0.75
>>   /14     $ 1.00
>>   /12     $ 1.50
>>   /10     $ 2.00
>>   /8      $ 3.00
>>   more    $ 5.00
>> which is *still* at the very top less than *half* what the 
>> LIRs charge at their *lowest* rate for a single static IP.  A 
>> 100% markup would seem quite sufficient.
>Interesting idea.  
>From what I've seen, assignments to organizations (with 
>dedicated service) are included with the service.  Does that
>make any difference?

Not really.  I'm sure the cost of service far, far exceeds
the max in the scale above.  I pay about $200/month; it's
hard to see where $5/year is going to impact that.

>> This would be fairer *to the community* than the current 
>> scale, and would certainly promote fast action on IPv6...
>> especially with the fee waivers already in effect for v6.
>> Anybody want to draft a formal proposal?  ;-)
>This would go through the Suggestion Process, not the Public
>Policy Process.  https://app.arin.net/suggestion/

Why is that?  I've seen other proposals that affect the fee
structure here, and this is specifically about public policy.
Just wondering...


More information about the ARIN-PPML mailing list