If you think of IPv6 as an end to end technology that can gracefully ride on
top of the global IPv4 ISP provided infrastructure, you don't have to have the
"Internet Infrastructure" transition to IPv6 for a long, long time.
Peer to Peer applications are going to be sufficient to get people to upgrade
their "NAT infrastructures" to support IPv6 over IPv4 tunneling. This is all
aided by the plummeting cost of NAT replacement (e.g. < $50, which is a lot
cheaper than the early Internet upgrade days of modem replacement for v.34,
v.90, cable modem, etc.). Thus, the tunnel phase of IPv6 is quite viable,
and has the FEATURE that you don't need massive shifts/upgrades in public
Internet infrastructure. This is relatively simple to promote to end users as
well: your old and new can work together ... ( I have observed: not too many
end users are into the "management" of their networks.)
IPv6 only networks can be created to support closed IPv6-only device networks
such as the cell phone networks. Thus, HUGE networks with application
relays/proxies between ipv6 and ipv4 infrastructures will be created where
global interop is essential, such as email, web browsing, etc. The incentives
to the cell phone operators are obvious. One might also observe: the cell
phone operators and their suppliers are not keen on building open host
platforms for deployment of arbitrary applications and protocols. This will
turn out to be a feature for IPv6 deployment. Thus, a new breed of ISPs will
drive IPv6 deployment in ISP space.
Enterprises will be last to go. Why? Because in fact they run relatively
small networks, and the core value of IPv6 is large addressability. The
extent to which better security is available for IPv6 may be a sufficient draw
for many companies, but not all. Thus, the current "revenue champion" for
networking will not be the early revenue driver for IPv6 as we have seen for
routers and ethernet switches. Since enterprise also drives a huge amount of
ISP deployments, I would not expect the current ISPs to be the first to
transition to, or augment with, IPv6.
The transition will occur based on technologies and deployments as discussed
and planned in the IETF oh so long ago. Much like it took the world many,
many years to realize the value of transitioning to IPv4 in the first place,
transition to a new infrastructure protocol is going to take a long time.
IPv6 was created for a world with tons and tons of hosts and devices. That
time is almost here, so it is only now that IPv6 deployment is going to be
economically relevant and viable. Pushing in front of this time horizon is
good for getting the deployment issues worked out, but people can't expect that
just because the technology is good to go, that it will just go.
regards,
peterf
P.S. The IETF might want to spend more time working on topics germane to the
cell phone industry. Not that IPv6 hasn't taken a lot of time ... :-)