[ppml] Proposed Policy: IPv4 Countdown

[Stephen Sprunk]

Thus spake "Ted Mittelstaedt" <tedm at ipinc.net>
>>IPv4 space means the collection of 32 bit integers (when used
>>as IP addresses).  However, the portion of the IPv4 space over
>>which ARIN provides stewardship is limited to that which has
>>been delegated to ARIN by IANA.
>
> Well, Owen, it looks like you and John have differing ideas about
> that. Check his response out to this question.

I see no disagreement.  You are quibbling, I think, over some 
assignments/allocations made directly by IANA or InterNIC, which have since 
been handed over to ARIN to maintain.  ARIN does provide stewardship for 
that space, though the terms it's done on are a bit different.

>>There is no direct simple answer to that question because of the
>>multiplecategories of addresses under current ARIN stewardship.
>>In the case of addresses issued under an ARIN Registration
>>Services Agreement, generally, yes, there is the authority to
>>revoke an assignment or allocation.  However, this ability has
>>never been put to a legal test and it is unclear whether ARIN
>> could prevail in such a suit.
>
> Once more you and John differ.  I would tend to side with John here
> anyway, because to get assigned addresses the organization must
> agree to a contract.  Thus it is a simple contract violation, courts
> have been dealing with these for years.

Nothing is simple when lawyers get involved, and this sort of contract is 
novel in a variety of ways, so until it's been tested in court, we really 
don't know how enforceable it is.  Courts throw out contracts all the time 
for various reasons; anti-trust issues in particular are of serious concern 
to RIRs.  Unfortunately, for now ARIN is in a much better position using the 
carrot than the stick.

>>In the case of legacy assignments and allocations, originally
>>issued prior to the existence of ARIN, it is even less clear that
>>ARIN has any ability to revoke them.
>
> If ARIN's charter is responsible stewardship of IPv4 space and IPv4
> space is defined to be all space even that not assigned to an RIR
> then it seems to me that there is an issue there.  But, this is the
> heart of the issue of IPv4 conservation - unless IPv4 space issued
> prior to the RIR's is brought under control of an RIR then there is
> no way to fairly allocate IPv4 space, once it becomes constricted.

All of the address space is either reserved or assigned to an RIR.  The 
issue is that legacy assignees/allocees(?) are not bound by any contract 
with their respective RIR that dictates the terms of that relationship.

In theory, since there's no contract ARIN has no legal obligation to 
maintain those registrations, but the community has, to date, felt that 
there is a moral obligation to maintain them at no cost.  Again, if you 
disagree, submit a policy proposal and we'll see if things have changed.  I 
doubt it, though.

>>Currently, there is no policy to support such action.
>
> Exactly, and this issue must be faced squarely or we are just wasting
> our time on the whole issue of extending the life of IPv4 on the
> Internet.

Then quit arguing about it and submit a policy proposal.

> This is also an issue that must be faced squarely.  Any plan to
> extend use of IPv4 past the actual end of allocatable blocks of
> numbers must deal with the mechanism of finding and obtaining
> previously-allocated blocks that are now unused.  Because there
> isn't going to be anywhere else to get the IP space from.

Do note that the projections for how long address-reclamation efforts will 
extend the exhaustion are on the order of six months.  That means it'll take 
us longer to reach consensus and implement the changes than the period of 
time we'll buy by doing so, meaning we're better off _not_ doing it and 
instead spending our time figuring out how to get people on IPv6.

> IP addresses are abstractions.  In my view, the concept of IP
> addressing was first defined in BSD UNIX many years ago. ...

Hardly.  You really need to read your history...

S

Stephen Sprunk      "Those people who think they know everything
CCIE #3723         are a great annoyance to those of us who do."
K5SSS                                             --Isaac Asimov