[arin-ppml] IPv6 Guarantee

[Owen DeLong]

On Apr 26, 2010, at 11:35 PM, James Hess wrote:

> Well, a few things...   (a) How do we measure the 20%?
> How can ARIN assure the would-be IPv6 applicant that a study will be done?
> 
I've got a couple of ideas on how this might be accomplished.  I know a network
that is in a pretty good position to take a representative sample of the current
IPv6 traffic load. ;-)  I'm investigating whether we'd be able to easily produce
these statistics or not.  The trick is identifying the residential traffic.

> Perhaps more importantly...  will ARIN be able to afford it  if the
> bet goes wrong?
> Isn't the per-reg price of an IPv6 registration connected to ARIN's
> costs of creating and  maintaining each registration?
> 
I think this would be a waiver of the annual fees (which presumably
people would end up still paying their IPv4 fees, so, it wouldn't likely
have a significant financial impact to ARIN).
> 
> And (b)...  does tunneled traffic count,  then where does packet count
> get measured?   For example, end hosts sending IPv6 packets  that get
> encapsujlated in IPv4 packets  and received by ISP as IPv4 packets,
> that are then transmitted to a tunnel endpoint for de-encapsulation.
> 
Tunneled packets will appear as IPv6 on the native IPv6 backbones.
They're only IPv4 from one end of the tunnel to the other.  Certainly,
I think they should count whether tunneled or not. Measurement should
be conducted from one or more points with native IPv6.

> Do those count towards the number of IPv6 packets or towards the
> number of IPv4 packets,  and  can we  expect a study to be able to
> identify/distinguish them
> 
They should only be counted once.  A v6 packet inside a v4 packet
should, IMHO, be counted as a v6 packet.

> IOW -- what type of packets does the study have to say there are 20% of...
> native end-to-end V6,   or    packets that are V6   at any point
> during their travel.
> ;)
> 
Packets where the destination and origin for the actual (innermost) payload
are IPv6 should count as IPv6. Packets where the destination and origin of
the actual (innermost) payload are IPv4 should count as IPv4.

The harder part is identifying residential packets vs. others.

Owen

> 
> On Mon, Apr 26, 2010 at 9:41 PM, William Herrin <bill at herrin.us> wrote:
>> All this rubbish about IPv6 as insurance popped an idea into my head:
>> 
>> The IPv6 guarantee:
>> 20% or it's free
>> 
>> Every January 1, ARIN will determine whether at least 20% of typical
>> residential packets on ARIN-territory Internet connections are carried
>> via IPv6, as reported by scientifically defensible studies. If not,
>> all registrants who both hold IPv6 registrations and route the
>> registered address blocks on the public Internet will receive a full
>> refund of any registration, maintenance or other ARIN fees incurred in
>> the prior year solely as a consequence of the IPv6 registration.
>> 
>> 
>> Do we -really- believe in IPv6? Lets prove it.
>> 
>> Regards,
>> Bill Herrin
> 
> 
> --
> -J
> _______________________________________________
> PPML
> You are receiving this message because you are subscribed to
> the ARIN Public Policy Mailing List (ARIN-PPML at arin.net).
> Unsubscribe or manage your mailing list subscription at:
> http://lists.arin.net/mailman/listinfo/arin-ppml
> Please contact info at arin.net if you experience any issues.