[AusNOG] My Predictions for the ISP Industry

[Mark Delany]

On 14Mar12, Leo Vegoda allegedly wrote:

> > You need to explain why a business would voluntarily stop listening to
> > IPv4 traffic and why ISPs would stop carrying it.
> 

> I can't tell you the decade but I would have thought the decision
> for a commercial organisation would be relatively simple. If it
> costs more money to maintain an IPv4 service than is made by its
> presence then commercial organisations would be motivated to remove
> the IPv4 service. After all, they're in business to make profits.

True of course. Though if folk have to run dual stack for a number of
years, would you expect the maintenance burden to be very high?

Quite possibly the opportunity cost of v4 addresses may become the
dominant economic factor in the earlier years. But at some point that
market will start to decline pretty rapidly.


One cost that isn't talked about much is the transition for
application s/w and databases. While there is plenty of evidence that
stacks and routers are good at v6 support these days, I wonder about
the state of network related applications written over the last 15
years or so.

E.g, how much running code blithely copies gethostbyname() responses
with an int32? Or assumes that the string representation in a database
is <= 15 bytes?  Or uses a hand-written dotted quad config parser?

Sure it's bad programming, but there are a lot of bad programmers.

What worries me most about these is that if their service provider or
IT dept. helpfully shields them behind 6to4s then they're going to get
a nasty shock very late in the day when they finally come out of
hiding.


Mark.