[arin-ppml] Policy Proposal 93: Predicable IPv4 Run Out by Prefix Size - Revised

[David Farmer]

On 18 Jun 2009 Martin Hannigan wrote:

> On Thu, Jun 18, 2009 at 7:14 PM, David Farmer<farmer at umn.edu> wrote: >
> On 18 Jun 2009 Martin Hannigan wrote:
>
> [ snip ] 
>
> > I don't want NYBISP (Name Your Big ISP) taking the last /10 ARIN has
> > and cleaning out the free pool.  This will create a political @#$$%
> > storm, if that happens.
> 
> What will create a larger political storm is if we dramatically change
> our policy to be lopsided towards one party or another.
> 
> If "big ISP" has the need for a /10 why shouldn't they be able to
> acquire it after justification?

Tell me how big ISP2 gets a /10 when ARIN has nothing left because big ISP 
1 cleaned out ARIN getting a /10, 5 minutes before?

With this policy big ISP1 gets a /12, big ISP2 gets a /12, big ISP3 gets a /12, 
big ISP4 gets a /13, big ISP5 gets a /13, big ISP6 gets a /13, big ISP7 gets a 
/14 ... big ISP30 get a /20

So 29 other big ISPs get partially filled along with big ISP1.

Under this policy maybe big ISP1 sues for not getting the whole /10.  But that 
is one case for ARIN to defend, and one judge to convince that ARIN was a 
good steward of resources. 

Under the current policy why wouldn't big ISP2-30 sue ARIN for getting 
noting.  But, now you have 29 case to defend and 29 judges to convince that 
ARIN was a good steward of resources.

This isn't about the big guys or the little guys it about how we approch 0 
resources available.

> >If it happens, it will not be NYBISP's fault it will be ours because
> > they will only be doing what we are telling them to do.  Big ISP are
> > not evil, they are Big and use lots of addresses, if you tell them
> > they should clean out ARIN they can, will, and should, because it is
> > what we are telling them they should do.
>
> We have policy. That policy is followed by our members. If our members
> have a need, they make a request. They request what they need. We
> review their requests for allocations against our policy. Analysts
> then determine that there is indeed a need. We've implemented checks
> and balance and even officer certifications.

The current policies are based on an assumption you just go get more when 
ARIN runs low.  When ARIN can't more the assumption the policy is based 
on breaks.
 
> How do these ISP's or others expect to clean us out again?
> 
> >
> > In the example above instead of NYBISP get the last /10, they would
> > get a /12 and could come back in 3 month if there is anything left,
> > this maximum goes down and down until you get to a /20 and them
> > someone big or small get the last /20 then ARIN runs out.
> 
> Right, but again we're faced with one requester[1] fulfilling 100% of
> their need and another fulfilling 1%. This is a situation very much
> like the one with the recent global policy trying to return space to
> the IANA from within the region[2].

So your suggestion will fill 100% of everyones until we can fill 0% of 
everyone elses


> > I was trying to find a way make the end-game work without rationing,
> > that is what people seemed to want comming out of San Antonio.
> >
> > If you don't like this way what do you suggest?
> >
> 
> I suggest that we instead change the minimum allocation unit as we
> always have on a meeting to meeting basis and avoid creating a
> specific policy with automatic triggers.

Explain how changing the minimum allocations unit does anything?  Are you 
suggesting we raise the minimum allocation unit to /10 so only the big guys 
can get addresses?

Specificall what are you suggesting?

> So, in order to [attempt to] avoid another five hundred post thread,
> let me summarize:
> 
> 1. Not in favor

OK fine

> 2. Unfair to change mid-game

What happened to the /13 maximum allocation that ARIN originally had 
then?  Why wasn't that an unfair change mid-game?  Change is not unfair it 
is what policy does.

> 3. Creates an unfair condition where some get satisfied and others
> don't. 

This policy doesn't do that, running out of IPv4 addresses does, this policy is 
just trying to find a way to deal with the facts-of-life.

> 4. Could create an unfair advantage for competitors[3]

The current policy will (no could) create a unfair advantage for competitors.  
This policy is just trying make sure there are no big winers out of the run-out, 
it is trying to spread the pain around a bit.

> 5. Invites BoT emergency power action since it has automatic triggers
> that could fail on [unknown] existing conditions at any time.
> 
> I'd rather see the BoT invoke their emergency powers for something
> that we "missed", not created.


No your suggesting the we do nothing even though we see a big problem 
coming at us.
 
> Best,
> 
> Martin
> 
> 1. Not all requests are from service providers. There are big
> enterprises and infrastructure players who have large amounts of space
> as well and are likely to have continued needs.

Yes, they are included in this too, the policy covers allocations and 
assignements

> 2. I'm still baffled as to why the lawyers agreed with me when I
> stated the same thing re unfair and lopsided fulfillment with this
> policy, but disagreed with me on another, similarly functioning
> regional policy
> 
> 3. IANAL, but seems to at least offer a whiff of anti-trust i.e.
> restraint of free trade.



===============================================
David Farmer                                      Email:farmer at umn.edu
Office of Information Technology
Networking & Telecomunication Services
University of Minnesota		       Phone: 612-626-0815
2218 University Ave SE		       Cell: 612-812-9952
Minneapolis, MN 55414-3029	       FAX: 612-626-1818
===============================================