[ppml] Policy Proposal: 2007-12 IPv4 Countdown Policy Proposal

Ted Mittelstaedt tedm at ipinc.net
Wed Mar 21 19:56:38 EDT 2007

>-----Original Message-----
>From: James Jun [mailto:james at towardex.com]
>Sent: Wednesday, March 21, 2007 3:45 PM
>To: 'Ted Mittelstaedt'; 'Rich Emmings'; ppml at arin.net
>Subject: RE: [ppml] Policy Proposal: 2007-12 IPv4 Countdown Policy
>> Thus the organizations that are rich and can
>> afford to twist things to get IPv4, will over time collect the higher
>> paying customers.  In other words, the rich get richer and the poor
>> get poorer.
>My concern here is with the merit behind your rationale that, instead of
>moving IPv6 FORWARD, that we should artificially regulate, where
>people will
>need to be forced and imprisoned to transition to IPv6 and create a
>situation where IPv4 is still what they want.  So instead of moving IPv6
>forward, the plan is to create a cynicism around IPv6 where people
>will WANT
>to go back to IPv4 unless they are forced not to by an unneeded regulation.
>This is not necessary.
>The 'rich gets rich', 'poorer gets poorer' would not happen when everyone
>wants IPv6, not be forced into it as you propose; therefore such clause of
>argument makes no sense.
>And you too, can help to spark interest in IPv6 by consider enabling IPv6
>inside your own network today, for your customers, as opposed to
>proposing a
>forced regulation schedule.
>ar2.bos>show bgp ipv6 unicast reg _10248_

But then if I did that I would be not sparking interest in IPv4
reclamation.  Which I think is important and should be done before
IPv6 switchover.  You have a point, though. :-)

Simply publicizing a T-date can simply be nothing more than publishing
a T-date.  Since there does not seem to be a way to prevent people from
advertising IPv4, a T-date could be considered to be nothing more than
symbolic - but symbols are powerful.

If they were not, people wouldn't be arguing over the idea of the
RIR's getting together and agreeing on a single T-date.

>> It appears to me that the only way to prove need of IPv6 is to be
>> able to state with authority that "All usable IPv4 addresses are
>> currently in use"  Anything else is a prediction of when it will
>> happen.
>Your notion that the only way to adapt to IPv6 is to set a timeline is
>wrong.  Ask carriers why they are not supporting IPv6 today, and their
>response is "show us the money and we will."  How do these carriers make
>money, by moving bits.  When there is traffic in IPv6, market will adopt,
>plain and simple.  And to create traffic for IPv6 is to get application
>writers and rest of the IT industry to realize, through mass marketing
>effort, that shows IPv4 is ending soon and IPv6 is the preferred industry

That is nothing more than a statement of a catch-22.

A timeline is one way out of this catch-22.  Can you suggest others?
I can't think of any.

>Setting artificial timeline as a matter of policy within IP allocation
>authority only exacerbates the problem for ISPs and carriers,

No, it merely creates this publicity and advertising that you say
is needed to get people to switch to IPv6.

>and further
>confines the boundary of the problem to carrier industries, as opposed to
>all participants of the Internet.

Yes, and there is a good reason why.

The users of the Internet will do what they are told.  If their ISP says
that they have to spend money to switch to IPv6 and, not liking this, they
go to their ISP's competitor, and also hear that they have to spend money
to switch to IPv6, they will have no choice but to either disconnect from
Internet, or spend the money to switch to IPv6.

The carriers by contrast don't have a group of ISPs telling them what to do.
You can tell them to switch to IPv6 by regulation, or their customers can
tell them to switch by customer demand.

If you want to use customer demand, then fine.  THe historical way this is
done is to design a "killer app".  For example if you could download, for
free, all the pirated music and videos you wanted as long as you were on
the IPv6 network, why then there would be huge customer demand for it and
the carriers would switch immediately.

But, there is no "killer app" out there that requires IPv6.  Thus you will
not get customer demand for IPv6.  Thus the only way left to get carriers
to switch is regulation.

If we do nothing with policy, then the regulators who distribute IPv4
allocations will eventually run out, and will deny new IPv4 reguests.  That
is a regulatory approach to switching, thus the carriers will be forced
to switch.

But, it's going to be done in a very haphazard way because the runout will
not occur all at the same time.

So, some carriers will switch before others.  That has in fact already
happened as there is a lot of deployed IPv6 in Asia.  But, customers sitting
in Japan cannot very well run to the United States for Internet service
when they were told they had to spend money to switchover to IPv6.

When ARIN runs out of allocations then you will have a situation where
some carriers in the US will run out before others, and the problem will
propagate downwards to their ISP customers.  And eventually you will in
fact have situations were multiple ISP's in the same city, some will be
able to delay IPv6 switchover.  Very likely it will be the deep pocket
ISPs.  So a day will come when the poorer ISPs start telling their
customers they must switch to IPv6, and their customers, not wanting to
spend money, will simply drop service and go to the deep pocket ISPs
who have been able to delay IPv6 switchover, and get a few more years of
IPv4 service before they will finally be backed into a corner and will
have to spend money to switch over.

If you want all Interent service concentrated in the AOL/Comcast/RoadRunners
of the world, then I guess you might be in favor of this happening.

A published and advertised T-date can help to equalize things because
those who run short of IPv4 first, can use the published T-date to
help educate their customers that IPv6 is inevitable, and moving to
their competitor down the street that will not require them to switchover
to IPv6 yet, will not in the long run save them any money.

> We do not need a policy to
>educate people
>that IPv4 is running out; this is best solved by mass marketing effort
>coordinated with journalists and industry efforts, not a
>regulation approach
>that oversteps the boundary of RIR's role as acting in stewardship.

Then the RIRs can simply set a T-date that is nothing more than a T-date
without any cooresponding policy to do anything about it.  What is your
objection to this?


More information about the ARIN-PPML mailing list