[arin-ppml] How hard is it to transition to IPv6?

[Ted Mittelstaedt]

> -----Original Message-----
> From: arin-ppml-bounces at arin.net 
> [mailto:arin-ppml-bounces at arin.net] On Behalf Of Stephen Sprunk
> Sent: Friday, March 27, 2009 5:43 PM
> To: michael.dillon at bt.com
> Cc: arin-ppml at arin.net
> Subject: Re: [arin-ppml] How hard is it to transition to IPv6?
> 
> michael.dillon at bt.com wrote:
> > And read my email again, carefully this time. In effect, I 
> told the customers of all ISPs, not just one ISP, that they 
> SHOULD complain about their ISP supplier's negligence if that 
> ISP supplier does not provide satisfactory timelines for IPv6 
> support.  If you are signing two year contracts, then you 
> might get one more renegotiation cycle before IPv4 runs out. 
> If you are signing on for three years, you might not get a 
> chance until it is too late, which means that NOW IS THE TIME 
> TO TALK TO ALL OF YOUR SUPPLIERS of network services and 
> network equipment and network software. This is simple 
> reasoning based on the projected runout dates.
> >   
> 
> Note that "complaints" and "discussions" are, in most cases, 
> not enough.  You must make it clear that support for your 
> requested feature
> (IPv6 or anything else) is mandatory or you'll take your 
> business to someone else who does offer it -- and follow 
> through on that threat if they fail to comply.  Lost 
> business, particularly from existing customers, is the _only_ 
> thing that reliably motivates large companies.
> 

I must disagree with this approach.

We've had a handful of ultimatims from customers in our history.
I'm sure that every network provider has.  All of them I can
recall are outrageous.  For example, one time a customer with
an infected server (which we had identified was infected and was
spamming millions of pieces of mail a day as well as attacking
servers across the Internet) in our colo 
demanded that we allow them to continue running this webserver
for a week because they were too busy to get around to fixing
it - and that if we didn't let them do it, they would pull out
their server.  I can tell you that made a difficult decision
(cutting off their service, designing filters, etc.) very very easy.

Almost always, issuing a simple ultimatim is counterproductive.  We
have also discovered with experience that many times, responding to
an ultimatim is a waste of time - because by the time the customer
issues one, they already have made an internal decision to leave
anyway, and they are just mad at you for "forcing" them to leave,
so many times these are "spite" ultimatims.

When pushed, large companies look at everything from a cost/benefit ratio.
Meaning, how much money will it take to satisfy a current customer
demand vs how much money it will take to write off a current customer
and get a new one.  Right now, IPv6 cost to implement for a lot of these
large companies is higher than the cost to write off a customer demanding
IPv6 and get a new one not demanding IPv6 to replace it  In other words,
in a large ISP that isn't offering IPv6 right now, the chief of network
operations isn't going to get in trouble from his CEO if he tells a
customer issuing an ultimatim to screw themselves, because he can easily
show that answering that will cost more money than the customer is
worth.

Even the US Government, which you wouild think is the 600 pound gorilla
on this, discovered that.  They mandated IPv6 for last year - and still
a lot of their vendors weren't in compliance by the deadline.  Some likely
still aren't.  I'm sure all are working on it, though.

If every IPv6-demanding customer simply writes off a vendor and gets a
new one that is supplying IPv6, it is going to
end up creating several large networks who have all the IPv6 business
while the rest of the world is IPv4 and then everything will grind to
a halt with regards to IPv6 penetration.

Obviously, if your being pressed by a customer to sell them IPv6 and
your business is structured to where you're the flea on that customers
back (like if they are 90% of your business) you don't have a choice
if your feeds or peers don't supply IPv6.

But, for most other people, I think just being persistent and continuing
to bother your vendors, asking for it, would eventually work.  After all,
vendors know they have to switch over to it eventually and I think all
the large ones are working on it.

Ted