[ppml] Policy Proposal: Resource Reclamation Incentives

[Owen DeLong]

On Jul 5, 2007, at 10:51 AM, William Herrin wrote:

> On 6/28/07, Owen DeLong <owen at delong.com> wrote:
>> Here's an attempt to partially drain the swamp and create some
>> incentives
>> for legacy holders to both return available IPv4 space and start  
>> using
>> IPv6.
>
> Owen,
>
> I don't object to your proposal but I question its value.
>
> I'm a legacy registrant with a /23 down in the swamp and until a
> couple years ago I was the lead engineer for a registrant holding two
> legacy /18's. The policy offered by your proposal does not appeal to
> me in either role.
>
OK... As the /23 holder, you're really not the target.

> As a /23 holder, why should I return the /23 or part of it (a /24)? I
> pay ARIN nothing now so a fee waiver is meaningless. A full fee waiver
> for the initial assignment of an IPv6 block might be nice but I'd want
> a more definite statement than "if you qualify" before both making the
> effort to apply and turning in IPv4 addresses.
>
ARIN has no way to determine if you qualify for an IPv6 assignment
before you apply for one.  However, when you applied, the only
thing you would be obliged to is filling out the template and emailing
it.  If you didn't qualify, it would end there unless you choose to
pursue it further.  If you do qualify (either initially or after further
effort), then, you would have the OPTION of taking an election under
my policy to receive the IPv6 space for free (for 5 years) by bringing
your IPv4 space (and ASN(s)) under RSA.
> As the engineer for the /18s registrant, I recognize that renumbering
> folks out of part of that space is a major undertaking. It will cost
> me many thousands of dollars of manpower and will impose additional
> and unexpected engineering costs on my customers leading some to
> reconsider their service contracts.
>
Depends.  If you're sparse-allocated all over the /18s, then, you're  
right.
If, OTOH, like many of the legacy /8s out there, you have vacant /20s
all over the place, then, there's virtually no cost to returning them.

> Even if I approach it opportunistically and just don't reallocate
> space in a particular part of the block when old customers depart, I
> fail to see how giving up precious IPv4 space (and doubly-precious
> fee-free legacy space) could possibly be compensated by saving a
> pittance on my new IPv6 block.
>
And lots of people will probably feel that way.  Obviously, there will
be no value to this proposal in those cases.  However, some people
will actually choose to do what is best for the community if they can
do it without taking on significant additional risk or cost to  
themselves
in the process.
> It just doesn't make good business sense.
>
Community minded action rarely makes good business sense from
the perspective you are approaching this with.
> I don't want to be the kind of guy who just says, "No!" so what would
> it take to get me to sign an RSA, turn in part of my space or both?
>
> Before I'd step forward with my legacy registration and either sign an
> RSA or give back part of it, at least one of two things would have to
> be true:
>
> 1. I'd have to realize some appreciable gain for my activity, to
> offset the loss.
>
> What if, for example, I could trade up to a /48 of IPv6 addresses with
> no initial assignment fee and no justification for each /24 of IPv4
> addresses I turn in with the requirements that I also place any
> retained IPv4 addresses under the RSA and that do so no later than
> 12/31/2008? Now you have a real enticement. I can get something
> cheaply now that may not be available later at any price but I have to
> behave in a way that meaningfully benefits the community to get it.
>
I'm already offering you pretty close to that.  The barrier to  
qualifying
for IPv6 space has been reduced to the same requirements as IPv4
space has today.  So, if you are multihomed and have 500 or more
hosts, you can get portable IPv6 space.  It does not make sense to
me to hand out a /48 for every /24 returned.  Each /48 is 65,536
subnets.  There's no way to carve a /24 up into more than 256 unique
subnets, and, to do that, you have to make it into /32s.  Even if you
go with point-to-point link numbering, you're maximum number of
subnets from a /24 is 64 /30s.  Why should we trade you 1024:1?

If you have a need for more than a /48, you can easily justify it.
If you are an ISP and would be issuing /48s to other organizations,
you can easily get a /32.

Other than that and the "lack of justification", you haven't presented
anything my proposal doesn't already offer.  Best of all, you can
actually "justify" and see if ARIN accepts your justification _BEFORE_
making any commitment.

>
> 2. My action would have to REDUCE future uncertainty about the status
> of my registration.
>
> At present, ARIN guarantees that legacy assignments will be managed
> under the policies then active while assignments under the RSA are
> subject to whatever policies we folks here on ppml can convince the
> board to implement. Thus signing an RSA and undertaking related
> activities would serve to INCREASE my uncertainty around continued
> holding of the address space.
>
I'm not sure about that original guarantee.  While I think that ARIN is
obliged to exactly that, there are a number of people with different
opinions, and, the angry mob mentality will only get worse as IPv4
free space exhaustion gets closer.

> On the other hand, I presently have no contractual rights associated
> with my legacy registration. Any rights I might have had expired with
> the implied contract with Network Solutions when they quit the IP
> registry business. ARIN has chosen to obligate itself to maintain that
> registration, but if they reinterpret that obligation to my
> disadvantage I might not even have standing to sue.
>
IANAL, but, the way I interpret the situation, you would not be able to
sue ARIN for dropping your WHOIS record(s) or your IN-ADDR
delegations, but, you'd probably have a pretty good case if they
issued any of your resources to a third party or encouraged the
use of any of your resources by third parties.

> If I could sign a modified RSA that contractually obligated both ARIN
> and I to follow the policies in place today but exempted me from any
> future policy until such a time as I found it advantageous to accept
> the regular RSA, that might well REDUCE the uncertainty associated
> with my registration. As a legacy registrant, I would at least find it
> worth considering.
>
Understood.  There is effort being expended on exploring ideas like  
this,
but, that effort is not yet ready to be brought out into public light.


Thanks for your input.  I'd really like to continue the discussion on
section one of your response above as I think we are much closer
together and that perhaps some tweaking on both sides would
facilitate a more attractive proposal and acquire your support.

Owen