[ppml] FW: No transfer policies are needed

[Ted Mittelstaedt]

> -----Original Message-----
> From: ppml-bounces at arin.net [mailto:ppml-bounces at arin.net] On 
> Behalf Of Edward Lewis
> Sent: Tuesday, April 22, 2008 12:19 PM
> To: ppml at arin.net
> Cc: ed.lewis at neustar.biz
> Subject: Re: [ppml] FW: No transfer policies are needed
> 
> 
> I have not worked for an ISP/LIR and haven't managed address 
> assignments.  So I want to ask some questions related to the 
> "reality" of this scenario.
> 
> At 11:40 -0700 4/22/08, Ted Mittelstaedt wrote:
> 
> >In other words, suppose you have a small ISP "SmallCo Inc." 
> that owns 
> >an IPv4 block it doesen't need.
> 
> In that case, isn't ARIN justified in recalling the block? 

As long as SmallCo keeps paying the yearly fee and does not
request additional space that would require justification, how
is ARIN going to know there is a block to be recalled?

> Or is 
> this a block that is "legacy?"
> 
> The context of my question is that I have long assumed that the issue 
> with the markets is the large amount of legacy (=RIR unencumbered 
> space) and not the space covered by registration service agreements.
>

No.  The legacy space isn't significant even though people think it
is, this has come up multiple times on this list and several people
have bothered to do the research and run the math, and the amount of
IPv4 that could be freed from legacy holders isn't going to significantly
extend the IPv4 runout date.

  The issue is that post IPv4
runout your going to have an Internet with mixed IPv4/IPv6 on it.  Orgs
that are moving to IPv6 will eventually have proxies or translators or
gateways or whatnot so that IPv6 customers can talk to an IPv4 Internet,
and IPv4 customers can talk to an IPv6 Internet.

Ultimately your going to enter a period of time where some orgs will
have all IPv6 customers, and will be able to use proxies or gateways
or translators for those customers to access the remaining IPv4 on
the Internet.  At that time they will have unused IPv4.  What to do
with it?  If it's returned to the RIR then it will be available for
reassignment to orgs who have not yet converted to IPv6.  That's what
the community should want, of course.  However, if the org sells it,
the org may get a windfall if they are an early IPv6 adopter.  So you
can see that a time will come when the early adopters may want to
have ARIN bless such transactions.  And some people think that if ARIN
blesses these transactions it will encourage orgs to abandon IPv4 all
the earlier - which will ultimately add pressure to the rest of the
Internet to switch to IPv6.  The downside of course is by making IPv4
available to slow-starters, it can also act to delay IPv6 introduction.
The other downside is that as the RSA that the org signed disallows
consideration of IPv4 as "property" there is no legal mechanism for
the seller to "sell" the numbers - thus ARIN would have to be involved
in such a transfer, (perhaps allowing an org to sell "reservations")
but even setting up such a scheme greatly weakens that section of
the RSA.

I could see a court case where someone claims that ARIN setting up
an "IPv4 Market" invalidates section 9. of the RSA as ARIN themselves
are now treating IP addressing as property.  And if that section is
invalidated then the entire idea of an ARIN-controlled market falls apart.
Which means now you will have spammers buying IPv4 blocks on the
black market that ARIN -cannot- revoke under Section 4 c ii, and iii.
and all kinds of nasty things.
 
> >
> >You have an ISP named "MCI Corp" that needs more IPv4 and 
> does not meet 
> >the utilization criteria to get it from ARIN
> >
> >SmallCo secretly sells it's block to MCI Corp.
> 
> How is this different than reassigning or reallocating addresses from 
> SmallCo to MCI? 

Because SmallCo still is responsible on a reallocation.  If ARIN goes
to SmallCo and says "you haven't paid your bill we are yanking your
allocation" then MCI can't do a damn thing about it. (except sue SmallCo)

> Instead of selling the addresses as capital, why 
> don't they just record a charge (= to an amortization "cost" - 
> remember, I'm not an economist), a "rental" of the addresses?
> 

That is what a SWIP does.

> >SmallCo sends a name-change in to ARIN claiming it's 
> spinning off it's 
> >ISP operations to a separate company named "NewCo Inc"  ARIN changes 
> >the record on the block.
> >
> >MCI Corp then registers the company name "NewCo Inc" as a 
> subsidiary of 
> >MCI Corp. in whatever state it's incorporated in.
> 
> This seems like a lot of trouble to accomplish what an 
> upstream/downstream relationship already is.  If I understand that 
> right.
> 
> >Another "black market" scenario would be for a large org 
> that did not 
> >meet utilization criteria to create an independent company, use that 
> >company to obtain a block from ARIN, then "buy" that independent 
> >company. Of course, that only works before IPv4 runout.
> 
> My problem is that I find it hard to imagine that a large 
> organization could not meet the utilization criteria (in a way 
> cheaper than having to incorporate and buy a new company).

There's been stories in the bad old days of large orgs with many
internal WAN circuits, each of which were numbered with a /24.  I
have gone through renumbering of a large network myself and it
takes a -lot- of work.  I can easily imagine a scenario where a
large org has thousands of hosts with STATIC IP addresses in 
application configuration files and they would regard renumbering
as more expensive than buying IP's from a black marketer.

Personally I think such logic is stupid because your passing up
an opportunity to correct a lot of old crappy configurations,
but I tend to think long term - most CEO's do not.

> I could 
> understand these scenarios after the free pool runs out and that is 
> the reason that the large organization failed to get space.
> 
> >I am not familiar enough with the existing language of the various 
> >RSA's that the RIR's use to know if such behaviour is explicitly 
> >prohibited but I think it is, certainly it violates the intent.
> 
> My read of the scenarios is that these seem like a lot of work to 
> circumvent the rules when there's gotta be something else that was 
> missed.  Maybe these scenarios are close to what's happening, or 
> could happen, but I don't see that the scenarios are realistic.
>
> >In any case, I would assume that ARIN would have the guts to revoke 
> >assignments on any blocks that were transferred in this 
> manner, if it 
> >discovered them.  Thus, I think the risks of block revocation would 
> >likely make an "IPv4 black market" a non-starter.
> 
> I can imagine that the one wanting the space might not get what they 
> want, but if someone has space available, that's space ARIN could 
> revoke, no?
> 

That is a good question.  Here is the RSA for ARIN:

http://www.arin.net/registration/agreements/rsa.pdf

Read it and decide for yourself.  Here is my $0.02: 

Yes.

Here's the operative section of the RSA:

4. CONDITIONS OF SERVICE
(c) Cooperation.

During the term of this Agreement, Applicant shall provide ARIN complete,
up-to-date and accurate information, assistance, and cooperation that ARIN
requests in ARIN's provision of the Services to Applicant, including,
without limitation, during its review of Applicant's utilization of
allocated number resources.

In other words, ARIN can any time during the term of the agreement, request
a review,
(see section 8)
and the registrant must comply - including providing ARIN with information
showing that
it is no longer using the IPv4 it was assigned (due to replacing it with
IPv6 let's assume).

> >The typical MO of the EPA when it discovers a polluter, is 
> to tell the 
> >polluter that if they spend the money to install pollution 
> control that 
> >they will only get a token fine and nobody will go to jail.
> 
> The limitation of this analogy is that there's no widespread cost to 
> clean up when a block of addresses has been "hoarded."  Unless you 
> stretch this to include router slots or something.  Addresses held 
> back represent growth that isn't happening but once they are 
> reclaimed, the damage is instantly undone.
> 

That wasn't the point of the analogy, actually.  I was illustrating that
when penalties are light enough more people will do the wrong thing,
nothing more.  Thus, the claim that an "IPv4 black market" will form
by itself post-IPv4 runout is untrue.  An "IPv4 black market" will
only form if the penalties for violating the RSA are light, and
there's little effort on ARIN's part to root out the participants in
such a black market.

Ted