[arin-ppml] Draft Policy2009-1: TransferPolicy (UsingtheEmergencyPDP)

Ted Mittelstaedt tedm at ipinc.net
Thu Mar 26 15:31:53 EDT 2009


> -----Original Message-----
> From: arin-ppml-bounces at arin.net 
> [mailto:arin-ppml-bounces at arin.net] On Behalf Of Matthew Kaufman
> Sent: Thursday, March 26, 2009 11:01 AM
> To: Kevin Kargel
> Cc: arin-ppml at arin.net
> Subject: Re: [arin-ppml]Draft Policy2009-1: TransferPolicy 
> (UsingtheEmergencyPDP)
> Kevin Kargel wrote:
> >> You conveniently left out my "the IPv4 space it needs to 
> last through 
> >> a transition that is starting too late to be done before runout".
> >>     
> >
> > Then complete the transition *before* IPv4 runout.
> >
> >   
> Demonstrably not possible. Slowing IPv4 allocation 
> sufficiently to not run out before transition completion 
> would at this point have the same effect as immediate runout. 
> Wishing the transition could happen faster won't help, either.
> >> Where does a hosted service that needs to deploy more servers to 
> >> serve increased demand from existing IPv4 clients get IPv4 
> addresses 
> >> from after the runout?
> >>     
> >
> > Umm , did you just ask where to get more stuff when there 
> is no more stuff?
> > The answer is you don't.
> >
> >   
> No. You get it from people who have it but for whom it has 
> much lower value than the value to you. They give it to you 
> in exchange for enough money that the money is worth more to 
> them than the addresses are.
> Google isn't going to stop growing just because ARIN won't 
> give them more addresses. They will go back and scavenge all 
> the ones they have, they will get them from other RIRs until 
> those RIRs are also out, they will assign additional value to 
> potential acquisitions that have existing PI IPv4 space, 
> especially if that space is underutilized, and they will 
> acquire or lease address space as necessary from other 
> entities that, with money applied, can find spare addresses 
> to release.

www.google.com resolves to a grand total of 4 IP addresses in
a round robin.  FOUR that serve the -entire internet-

gmail.com has an MX that resolves to a total of 5 addresses that
are -different- than the ones www.google.com goes to.

Doncha think that if Google needed IP's they might consider that
SMTP and WWW run on DIFFERENT ports and simply use the SAME IP numbers
for their MX records and webservers?  YA THINK!!!

Granted this is simplistic to the point of rediculous - but it illustrates
the point that IPv4 utilization by Google is already inefficient from the
of conservation of IPv4.  Frankly I doubt you know anymore than
I do how Google's internal network is structured, but I'm sure
they are well aware of this upcoming problem and have a plan already
worked out and being followed.

> This is a lot like how Sprint and Clearwire leased ITFS (now 
> EBS) spectrum on 30-year leases from universities that didn't 
> need all the instructional television channels they were 
> licensed for. It gets the cash-strapped university a bunch of 
> money to do other interesting new things, and it gets Sprint 
> and Clearwire a bunch of spectrum that otherwise wasn't 
> available. Good for everyone except little guys who didn't 
> think to lease the spectrum first, or didn't have the cash to 
> outbid Sprint or Clearwire... but then that's business, at 
> least in this country, these days.

That is not an apples-to-apples comparison.  RF spectrum and
land are analogous.  IPv4 is NOT analogous to either spectrum
or land.  If it was possible to completely replace the RF spectrum
with something different, Sprint and Clearwire wouldn't have
paid a cent to those universities.

> >> Where does an innovative new service that needs to deploy 
> servers to 
> >> serve customers who are only on the existing IPv4 Internet 
> get IPv4 
> >> addresses from after the runout?
> >>     
> >
> > See above.  Shouldn't innovative new services be using ipv6?
> >
> >   
> Innovative new services probably will be using IPv6. But 
> they'll also want to serve the large pool of existing 
> potential customers who are on
> IPv4 and who don't have access to an IPv4 to IPv6 proxy yet. 
> So, at least for a while, there'll be a lot of value in 
> having IPv4 addresses to reach these new services. Perhaps 
> enough value to make it worthwhile to spend cash to convince 
> someone else to part with, or at least lease, address space 
> to the new entity.
> >> Where does a new ISP that wants to deploy IPv6 to customers with an
> >> IPv6-IPv4 translation gateway to reach the legacy IPv4 
> Internet get 
> >> IPv4 addresses for the IPv4 side of that translation 
> gateway after the runout?
> >>
> >> The answer is that if there is money to be made in having those 
> >> addresses then acquiring those addresses will be part of 
> the capital 
> >> expenditure required to start or grow one of those 
> businesses. Yes, 
> >> it will cost more than it does now, and so yes there will 
> be business 
> >> models for which that no longer makes sense... but there may very 
> >> well be business models where the cost of getting a few more IPv4 
> >> addresses to continue growing, high as it might be, is 
> lower than the 
> >> cost of
> >> *not* having the addresses.
> >>     
> >
> > I see lots of questions describing problems of how to 
> continue to use 
> > a technology that is unavailable.  Face it, IPv4 is a 
> finite resource, 
> > when it runs out it runs out.
> >
> >   
> There is a big difference between "new IPv4 address space is 
> not available from RIR" and "the IPv4 Internet is shut down".
> In between those two times, there is value to new entrants 
> and growing entities in acquiring more IPv4 space. There is 
> also value to failing entities and entities with more IPv4 
> space than they need for their own growth to NOT having IPv4 
> addresses... at least there would be, if for a reasonable 
> transaction cost the addresses could be moved to the entities 
> with need. As I point out, that can already happen with 
> existing policy... this policy would just reduce the transaction cost.
> So you're against reducing the transaction cost to those 
> entities which might need additional IPv4 address space? I 
> don't understand that position given your belief that "when 
> they run out, they run out". If they've actually "run out" 
> then new or growing entities won't have any reason to acquire 
> addresses from existing entities with excess and so there 
> won't be any transactions covered by this policy anyway!

Nothing is preventing all these "new businesses" that supposedly
will need IPv4 from forming their own Internet and using IPv4 on it.
They can do that and number how they want.  Just don't expect to
connect that to us.

In the case of IPv4->Ipv6 transition, it's better for the majority
to get the IPv4->IPv6 transition executed as quickly as possible.
The faster that we switch everyone over, and the shorter that this
transition period takes, the less money that everyone
will have to spend supporting legacy stuff.

I don't have a problem with schemes that generate more assignable
IPv4 BEFORE arin runs out - I would prefer this be run through ARIN
of course - but the fact that it exists "for free" from ARIN pre-IPv4 runout
means that nobody "selling" IPv4 is going to make any money at all,
and thus there's no incentive to do it, and every incentive to
return it to ARIN and continue the existing assignment mechanism.

But once IPv4 runout is hit, the IPv4->IPv6 transition really will
commence for most people.  That is when it's going to hit the
general newspapers and such.  If people are told that well, IPv4
MIGHT be available if you get in a line at ARIN and wait long
enough, or it MIGHT be available if you go buy a company that
has an assignment, to them that means IPv4 is unavailable and they
will have incentive to switch themselves to IPv6.

By contrast, if people are told that IPv4 is still available it
is just going to cost more, to most people that is going to translate
to "I can forget about this IPv4 crisis and just go back to
watching Lost on TV" in short, to most people, it will mean IPv4
runout has not happened.

> >
> >
> > So why can't people go back to ARIN and get more IP addresses?  If 
> > they are available for transfer they are available for 
> recycling.  If 
> > they cannot be returned then they can't be transferred either.
> >   
> In our society, people often are more willing to part with 
> things they are using if they are offered something else of 
> value in exchange.
> I could have put the motorcycle I've had in my garage out on 
> the curb with a "free" sign on it. But I didn't. Instead I 
> sold it on eBay for $3000. When you understand why, you'll 
> understand why address space that isn't available for 
> recycling might be available in trade for other things of 
> value, like money.
> (Among other things, as I've pointed out months ago, an IT 
> department may find it much easier to justify the equipment 
> needed for an IPv6 transition if they can tell their 
> management and/or shareholders that by reducing their need 
> for IPv4 space they can recoup some of the transition cost by 
> transferring those IPv4 addresses for entities which have 
> figured out that the monetary value of getting more IPv4 space is
> non-zero.)

This presupposes that having your motorcycle once more back in
the hands of someone who will ride it, instead of rotting away
in your garage never being ridden, is beneficial to society.  I
think you would find a lot of people that would disagree with
anything that puts one more bike back on the street.

Keep in mind also that when you sold your bike, that someone out
there who wanted a bike DIDN'T go down to the Harley stealership
and buy a new one (I'm assuming your bike was a hog as nobody would
buy any other kind of used bike than a hog for that much money)

That means HD sold one less bike this year, thus made less money,
thus someone on an assembly line at HD didn't have any work, and
was laid off as a result.

Your sale was an action beneficial to you, NOT beneficial to the
majority.  It was tolerated by the majority because there's not
enough bike riders out there to go to the trouble of banning them.
But make no mistake about it, when the minority of bike riders
gets to be a problem, the majority reacts with stuff like helmet
laws, superbike limiting legislation, etc.

The fact is that post-IPv4 runout, it's more beneficial to the
majority for those IT departments to simply sit on their IPv4
holdings and NOT sell them - since this forces IPv6 adoption faster.
Obviously it's most beneficial if they convert to IPv6 and
walk away from those holdings.

But even if they don't get money for selling those holdings,
they pay less money to ARIN for IP holdings if they return them,
so they still financially benefit.


More information about the ARIN-PPML mailing list