[ppml] PPML Digest, Vol 22, Issue 9

[Ted Mittelstaedt]

>-----Original Message-----
>From: ppml-bounces at arin.net [mailto:ppml-bounces at arin.net]On Behalf Of
>Stephen J. Mahler
>Sent: Wednesday, April 04, 2007 12:24 PM
>To: ppml at arin.net
>Subject: Re: [ppml] PPML Digest, Vol 22, Issue 9
>
>
>Throwing caution to the wind, and becoming involved in this discussion ...
>
>I suggest that the problem is not getting people out of the IPV4 address
>space,  the problem is getting them into the IPv6 space.  If moving to IPV6
>is easy, interesting, has a clear benefit, and worked as more of a
>long term
>project than a flash cut... they will move.
>
>The existing IPV4 address space is a tiny fraction of the IPV6 space.
>Follow the lead of Madison Ave. ... give away samples.  GIVE each existing
>address space owner (OK, not really the owner) an upsized chunk of the new
>space.  Let the computers handle the application for the space, just copy
>the registration information.  Each IPV4 Class C is given a range equal in
>size to a IPV4 class B.   Each Class B is given a range equal to a current
>Class A.  Each Class A is given a SUPER-A.
>

I already suggested this last week and we had some discussion on it.  The
biggest problem is that there's large swaths of IPv4 that was assigned
pre-RIR and the holders never signed an RSA.  This kind of creates a
catch-22
situation; we need to move to IPv6 because large amounts of IPv4 was
assigned to legacy holders who we can't tell if they are wasting it
or not (but, compare the list to assigned addressing to what is in the BGP
table to get some idea) but the legacy
holders, having not signed an RSA, are not subject to utilization
requirements that would help prevent them from wasting it.

The key is to force the legacy holders to sign RSAs.  To do that we have
to make them come out of the woodwork and sign IPv6 allocation requests
that commit them to an RSA.  This is why they would never go for an
auto-allocation scheme.  Sure, we can auto-assign IPv6 but in order for them
to
use it they would have to sign an RSA for their existing IPv4 assignments
in order to get the IPv6 that is auto-assigned.

Consider ARIN's fee schedule.  Above a /14 is $18,000.00 USD a year.  Why
would a legacy /8 holder who is currently paying ARIN not a cent, because
they
have not signed an RSA, want to sign an RSA to get auto-allocated IPv6
for free so they can start paying $18,000 a year for their /8 and be
subject to utilization requirements as well?

A legacy holder of a /8 IPv4 that wants IPv6 would be better off simply
registering IPv6 by itself and paying much less money.  In fact, that is
one of the loopholes I would like to see closed in ARIN's policy.

A pre-RIR /8 holder who didn't sign an RSA for the /8 could I believe
go to ARIN and pay the yearly membership fee of $500 and then the minimum
$500 under the fee wavier and get all the IPv6 they wanted.  Of course they
would be subject to an RSA for the IPv6 but according to the published
policies I think they would only be required to demonstrate 80%
utilization on their existing /8.  In that case it would be $17,000
cheaper per year for them to NOT participate in an IPv6 auto-allocation
program.

>Make this the last time you have to worry about your existing address space
>between now and your retirement.  Make the advantage of switching to IPV6
>that you have room to operate and test for years to come.  Make it a reason
>to work thru the headaches of using IPV6.
>

I think an auto-allocation scheme would work for IPv4 holders who are
paying ARIN now.  What I would like to see is the separate fee schedules
for IPv4 and IPv6 to disappear and there to be only a single fee schedule
and both IPv4 and an equivalent amount of IPv6 be allocated on each
request - in other words, an IPv4 holder would no longer pay 1 fee for
IPv4 and 1 fee for IPv6, then would pay a single fee for "Amount of IP
addressing" they are using.

>You won't have to worry about closing down IPV4 if there are advantages to
>moving to IPV6.  If the only advantage to switching to IPV6 is that IPV4 is
>closing down tomorrow, everyone will wait till the day before.
>

Then everyone will wait till the day before, because right now there is no
"killer app" that is an advantage to switch to IPv6.

Ted