There isn't going to be a great transition to IPv6 in the sense that
you seem to mean. IPv4 and IPv6 will coexist for a long time.
Yes, but I am afraid that underestimates the overhead of running
dual stack.
Actually, I can imagine that a lot of enterprise networks will try moving
to single stack, and will almost get there. I also imagine that tools
will emerge to reduce some of the overhead of dual-stack administration -
e.g. managing address administration and router filters with one set of
rules. I don't think we'll see that many apps run on both protocols in
practice - I think network admins will say "within our network, apps a, b, c
run on v4, apps x, y, z runs on v6".
To be clear, I do think dual stack is the best strategy for the
present early stage of _transition_.
I see it as a transition also. But I think there will be a long period
in which v6 is used mostly for new things, and only when v6 is more
ubiquitious than v4 will we see some of the core services migrate.