[arin-ppml] Portable address space vs. IPv6 auto-numbering

Ted Mittelstaedt tedm at ipinc.net
Fri Jun 13 14:56:53 EDT 2008

> -----Original Message-----
> From: arin-ppml-bounces at arin.net 
> [mailto:arin-ppml-bounces at arin.net] On Behalf Of Milton L Mueller
> Sent: Friday, June 13, 2008 7:43 AM
> To: ppml at arin.net
> Subject: Re: [arin-ppml] Portable address space vs. IPv6 
> auto-numbering
> > -----Original Message-----
> > From: arin-ppml-bounces at arin.net
> > 
> > As for the argument that a relatively small fraction of the IP 
> > addresses currently advertised are actually in use, well 
> whether "the 
> > market" coughs up those "3.7B which could be advertised" IPv4 
> > addresses, or whether it doesen't, POST IPv4 runout, is at 
> this point 
> > really just speculation.
> The degree to which that happens is highly uncertain, but 
> it's a bit more than speculation. We know that gray market IP 
> address transfers have already been going on, and we have 
> experience with secondary markets in other resources, such as 
> bandwidth and spectrum. 
> > We won't know until after
> > IPv4 runout.
> Two points. First, I don't understand why you need to wait 
> for what you call "IPv4 runout" (see point 2) to permit 
> transfers. You could do it tomorrow.

The ONLY legitimate need for IPv4 transfers at the current
time is to subvert the IP utilization policies.

Sicne the RIR's have free IPv4, there is no economic benefit
to an org to pay a transfer fee to a 3rd party then start
paying ARIN fees when they can merely go to ARIN and get
the IPv4 for free - UNLESS for some reason ARIN would deny
their application.  Such as their failure to meet utilization

Note that a transfer doesen't happen when org1 buys org2's network,
that is an acquisition.

> Second, and pardon me if 
> this sounds pedantic, but people who understand price systems 
> know that resources don't literally "run out" unless there is 
> something drastically wrong with the social systems used to 
> ration them. They become increasingly scarce and expensive. 
> We will never, for example, actually "run out" of oil. It may 
> eventually become so scarce that you only see it in tiny 
> vials displayed as jewelry around the necks of beautiful 
> models, but it won't "run out." 

Tell that to the elephant tusk ivory resource.  Tell that
to the passenger pigeon meat resource, I understand they
were pretty tasty.

Sure we won't ever "run out" of IPv4.  In the future, post
IPv4-runout, there will still be orgs who go bankrupt and
stop paying their bill, and the RIR can then assume their
subnets.  That is a marginal case.  The term "IPv4-runout"
is as good as any for a term to label the day in the future
that the last virgin IPv4 block is handed out from ARIN.

> More importantly, I am wondering how you view the impact of 
> v4 transfer markets on the v6 transition. This seems an 
> important issue. Let's suppose it is fantastically successful 
> at prolonging the life of v4 Internet by releasing many 
> unused address resources. Does it then discourage v6 
> migration?

It delays it.

> Or does it over the longer term improve the value 
> proposition for v6 by making the fees/costs associated with 
> v6 addresses look more favorable?

It actually makes things a lot worse.  Unless you can rework
IPv4 to make IPv4 "effectively unlimited" the way IPv6 is, 
every year you delay IPv6 adoption you put more IPv4 into
service and the cost to shift the Internet over goes up.

Until Windows Vista the argument that we aren't ready for IPv6
had a lot of merit.  Today, yes there's still a lot of XP.
But every year less and less and more and more Vista.  Eventually
the only people who can't deploy IPv6 will be people like my employer
who have older routers that don't have enough ram to update
to current code.  And IP numbering policy should not be
paying attention to this sort of problem.

More information about the ARIN-PPML mailing list