[arin-ppml] is NAT an inevitabile part of IPv4 / IPv6 transition

[Owen DeLong]

On Feb 8, 2011, at 5:21 PM, Ted Mittelstaedt wrote:

> On 2/8/2011 3:06 PM, John Curran wrote:
>> On Feb 8, 2011, at 6:38 PM, Tony Hain wrote:
>> 
>>> If the IANA pool had run dry in 2009, the media attention we are seeing this
>>> past week would have already occurred, and CIOs would have already started
>>> efforts that are just now getting underway. The point is that dual-stack
>>> requires sufficient time to keep the old one working, so waiting until that
>>> is no longer an option as the starting point is guaranteed to create failure
>>> modes.
>>> 
>>> There is no one place to assign blame here, and blame was never my intent.
>>> If I had not put out my graph in 2005, attention on the consumption rate
>>> from IANA might have been ignored until it was too late to have any
>>> significant effect on the date, because Geoff's graphs from that time said
>>> 2019. If the RIR's collectively had not changed the practice of when&  how
>>> much to acquire from IANA at one time, the pool would have clearly burned
>>> down at least 2 years ago.
>>> 
>>> The point is simply that an opportunity for a graceful transition was lost
>>> because high level attention to the issue was deferred to the point where it
>>> was too late.
>> 
>> Hah. High-level attention doesn't drive deployment (except in a central-
>> planning or heavily regulated environment), a successful business case
>> drives deployment.
>> 
> 
> No, I have to disagree with this statement.  A successful business case
> only insures that your going to snare a sucker or two to make an investment.  What drives deployment is customer demand.  Sure, you can build a business case on customer demand - but the demand comes first, not the business case.
> 
>> The opportunity for graceful transition was lost when we both failed
>> to include transparent interoperability and then further provided no
>> additional functionality to drive deployment.  Reference RFC 1669.
>> 
> 
> I think you both must have extremely overinflated opinions of your own
> importance to assume that anything that you or anyone else would do would have motivated more than a fraction of people to bother with
> graceful transition any earlier than they did. ;-)
> 
That wasn't "we both" as in two people. That was "we" as in the collective
community and "both failed to include transparent interoperability and
then further provided no additional functionality"... The community failed
in both actions, not two people failed in particular things.

> The situation with IPv4->IPv6 transition is a game of chicken between competing ISPs.  Anyone running an ISP knows perfectly well that the ISPs who are later into the IPv6 game will benefit from the burned fingers of the ones earlier into the game.  So it is perfectly logical that most ISPs will wait until the very last minute to go to IPv6.
> They will wait till the day that they get the first call from a customer saying that they must have IPv6 or they will quit service, then they will think about moving on it.  We see Comcast doing trials now, because they are so large that they are probably already getting those calls from at least a few customers.  But it is going to take a while for the rest of them to follow along.
> 
I know at least one ISP that is doing quite well having been early to the
game with IPv6. Did we burn our fingers a few times? Sure. Did our
getting our fingers singed help others? Maybe. It helped us at least
as much and we got the lessons first, so, it put us ahead.

The last minute is upon us and may have passed.

As to the customer calls, well, I'm not at liberty to disclose customer
communications, but, I do know we win business because we have
IPv6 and other contenders are losing customers to us on that basis.

I think Comcast is doing trials, not because customers are demanding
it (I know there was customer demand years ago), but, because they
aren't spectacularly bad at math and they recognize that there's no
way for them to continue to operate and grow an IPv4 network.

Anyone who looks at the numbers and isn't spectacularly bad at
math can see that an IPv4-only network is simply not sustainable
and the practical days of dual-stack without IPv6-only hosts are
limited.

> How many customers out there are demanding IPv6?  Not many.  Only the ones who have no IPv4 at all are demanding it.  For all the rest of them, IPv6 is still on the "nice to have" list.
> 
Many, actually. Most of the demand we see for IPv6 is from people
who also want IPv4. IPv6 came off the nice to have list and moved
to the must have soon list over a year ago for most carriers I know.
Now, it's starting to move to the don't even bother to submit a bid
without IPv6 support list.

Owen