[ppml] Proposed Policy: IPv4 Countdown

Ted Mittelstaedt tedm at ipinc.net
Tue Mar 13 21:39:05 EDT 2007

>-----Original Message-----
>From: ppml-bounces at arin.net [mailto:ppml-bounces at arin.net]On Behalf Of
>Edward Lewis
>Sent: Tuesday, March 13, 2007 1:08 PM
>To: petition at arin.net
>Cc: ppml at arin.net
>Subject: Re: [ppml] Proposed Policy: IPv4 Countdown
>I would like to voice a qualified "I object" to this rejection.
>I would like to see some discussion on this, even though I comprehend
>the reasons given below for the rejection and acknowledge that they
>are valid.  My motivation for speaking up is to see if there might be
>a way that the spirit of the proposal can be pushed forward in ARIN
>even if the particular proposal has mechanics that are problematic.
>At this point, I don't have a specific recommendation, just wanted to
>say that there might be a reason to reconsider this, perhaps in
>another form.
>At 12:47 -0500 3/2/07, Member Services wrote:
>>On 1 March 2007 the ARIN Advisory Council (AC) concluded its review of
>>the proposed policy 'IPv4 Countdown' and did not accept it as a formal
>>policy proposal.

Objection methods are spelled out here:

"...In the event that the AC decides not to accept the proposed policy, then
the author may elect to use the petition process to advance the
proposal. For petition details see the section called "Petition
Process" in the ARIN Internet Resource Policy Evaluation Process which
can be found at:

The deadline for the author to initiate a petition per the ARIN Internet
Resource Policy Evaluation Process is 40 days prior to the meeting; the
petition deadline for the ARIN XIX Public Policy Meeting
is 14 March 2007. If the author chooses not to petition or the petition
is unsuccessful, then the proposed policy is closed. If a petition is
successful, then the proposal will be numbered and posted for discussion
and presented at ARIN's Public Policy Meeting...."

Frankly, it was a politically naieve proposal and it is unsurprising
that they killed it.  (the anti-trust excuse given is just hogwash,
of course)

The IP numbering registries are all political bodies, and the process
for IP address assignment is also political.  As everyone who knows
anything about politics knows, problems only get attention that are
about ready to burn down the house.

When all numbering registries have exhausted the pool of IPv4 assignments,
only at that time will there be the political will to start the garbage
collection process of reclaiming abandonded IP number blocks.  Look on
the Bogon list for a good place to start.  But more than that, of all the
number blocks assigned, it is clearly obvious that the vast majority of ones
assigned to corporations are NOT being used externally.

It is quite obvious, for example, that having something like
assigned to Microsoft makes it quite easy for Microsoft's administrators to
have this giant worlwide WAN that runs BGP internally and has many
points to the world.  Very good.  Then how come Looking Glass only shows
MS's AS's  (3598, and 8068-8075) interconnected to the rest of us
via Level 3?

And more importantly, is it reasonable to assume that Microsoft has anywhere
near sixty five thousand hosts directly accessible from the global Internet?

If they do not have sixty five thousand separate hosts then why do they
need 65536 routable IPv4 addresses?  Espically since they themselves sell
Small Business Server and they also sell MS IDS, both of which are firewalls
that force you to number your internal network privately.  Obviously what is
sauce for the goose is not sauce for the gander, here.

I also might point out that with a little effort you could find LOTS of IPv4

For the above example, for instance, Microsoft's administrators might argue
that they need dozens of /24's to be able to advertise at many different
Level 3
interconnection points because everyone filters anything below /24

And why is this?  It is because in the past, router technology has not
been able to deal with more than a few hundred thousand route entries.

Fine then.  ARIN can write an RFC introducing /29 global BGP routing.
Everyone on the Internet running core routers can replace their old crap and
buy new routers that can easily manage 20-30 million BGP route entries.
VISA can manage ten times that number of credit card numbers globally so
you know that companies could make that kind of router hardware if there
was demand for it.

Then Microsoft can replace all their /24 advertisements with /29
at their Level 3 interconnects.

Needless to say there will be more screaming than you can imagine by the
core that doesen't want to drop the cash into upgrading.  Thus it won't
happen until the alternative becomes more expensive (ie: shift the Internet
to IPv6
and renumber)

Clearly what is going to need to happen is for people like MS - who has no
real justification for that large an IPv4 assignment - to renumber and give
up most of their existing allocations.

But until the Internet is COMPLETELY OUT of IPv4 addresses, there will NOT
be the political will for ARIN to go to MS and force them to spend the money
to renumber.

The situation is the same and the US and soon China's dependence on oil
from the Mid East.

China and the US will never switch to alternative fuels until all of the oil
drained out of the Mid East oil reserves, and it is naieve to think

Thus, we will have to continue to pay attention to the idiots in Israel and
Palestine killing each other until this happens, and those people will have
incentive to change their ways.

Once the mid east oil runs out the world will ignore the mid east, and
5 years they will have run out of weapons and there will be no more war

Once the Internet "runs out" of IPv4 addresses, then these large
like Microsoft will have no choice but to give up their allocations and
renumber into something that is inline with the hosts they have on the
and suddenly there will be an oversupply of IPv4 numbers.

You need to understand the politics behind the IPv4-on-the-Internet-backbone

All the hardware and router vendors (ie Cisco) are solidly for going to IPv6
because they want to make $$$ selling hardware upgrades.

The telcos and ISP's are all solidly against going to IPv6 and are for
reclamation, because they have gear that is working perfectly well and they
don't want to scrap it.

>From the hardware people's POV they have won this war already.  But, they
have time on
their side.  They know that eventually IPv4 allocations will run out and
think at that time that they will be able to sell gear.  They do not want to
like greedy bastards so they will not publically support anything that will
hasten the day that IPv4 allocations run out.

>From the Telco's POV they know that there's lots of IPv4 hoarded out there
deep pocket companies like MS, and DoD and so on that got it ages ago when
it was plentiful.  They know that the more of this they can push the
registries to
cough up the more time they have to push off the day they have to spend the
for new hardware.  So they won't publically support anything either that
hastens the day IPv4 allocations run out.

And the standards bodies like ARIN don't want to get caught in the
They are going to pretend the problem is out of their control and there is
nothing that they can do.  Thus, they will be blameless when the day comes
the last IPv4 allocation is given out.

To politically naieve people like you I am sure all this sounds like a bunch
idiots that are just heading straight for a train wreck and you cannot
why nobody is jumping up and down and trying to slow the train down.  What
don't understand is that the cooler heads know all about this, but they also
know that if they jump up and down, that nobody is going to pay attention to
It is better to remain aloof, then when the train wreck does happen and you
a lot of stunned people running around panicing, then you can come sailing
on your white horse with your new IPv6 plan and "rescue" everyone. They will
so happy your picking up the wreck that they won't care your screwing them
up the ass doing it.

This is just how life and politics works.


More information about the ARIN-PPML mailing list