I agree with Sam, I don't think that we are proposing to shut down the IPv4 Internet overnight so what is proposed here is more of the nature of a PR stunt, a proof of concept than a test of a transition strategy.
PR stunts can be good, but they can also have the opposite effect if you don't know what is going to happen. I have been involved in several 'public' interop events at conferences. There is no way I would be within a hundred feet of one, let alone participate if I did not have absolute certainty in advance of what the result was going to be (yes there is usually a pre-interop before hand).
So what is the message we want to project here and what is the result we expect from the test?
If the result is that IPv6 does not work on the Windows machines in the room there will be much glee on Slashdot but the effort will tend to demonstrate the opposite of what I would want to demonstrate.
There is no shortage of address space in net 10. There will always be plenty of IPV4 address space for people to run private networks to laptops and the like behind a NAT.
A better test would be:
1) Set up a transparent IPv4 to IPv6 proxy
2) Turn off the externally routable IPv4 address space in the room, giving machines a NATed address in net 10 as their leases expire
3) Demonstrate that people can perform basic Internet functions in this state (Web, mail, Jabber) without disruption
People are making far too much of this transition as if it is something that the end user should care about. If we want the transition to work it has to happen without the end-user needing to know or care just like they didn't need to know when fuel injection replaced carburetors in the mid 80s.
From: Sam Hartman [mailto:hartmans-ietf@xxxxxxx]
Sent: Mon 17/12/2007 10:47 AM
To: Norbert Bollow
Cc: ietf@xxxxxxxx
Subject: Re: IPv4 Outage Planned for IETF 71 Plenary
Sent: Mon 17/12/2007 10:47 AM
To: Norbert Bollow
Cc: ietf@xxxxxxxx
Subject: Re: IPv4 Outage Planned for IETF 71 Plenary
>>>>> "Norbert" == Norbert Bollow <nb@xxxxxxxxx> writes:
Norbert> Iljitsch van Beijnum <iljitsch@xxxxxxxxx> wrote:
>> But what about transition mechanisms, or would that be unfair?
Norbert> IMO it would be unfair on IPv6 to do the test without
We should provide transition mechanisms only if we believe they are a good idea.
So far I think we want to recommend dual stack. So, we need to think
about this outage more as a way to get ourselves experience with IPV6,
not as a network configuration anyone should be expected to use on a
regular basis.
_______________________________________________
Ietf mailing list
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf
_______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf