[Technical data comms stuff in the middle. Extropian conclusions
at the end.]
This announcement is really hard to evaluate. As far as I can tell
after carefully reading their extensive press release,
http://www.sprint.com
they are announcing an amalgamation of several different technologies.
I can't tell for sure, but it looks like a "spraypaint integration"
(i.e., don't really change anything in several separate products,
just repaint then all the same color.) The internal hi-speed Sprint
network will apparently be IP over ATM and voice over ATM. The
high-speed lines to big busnesses will apparantly be ATM over SONET,
with IP over the ATM and voice over ATM(?) The high-speed local loops
for small business and consumers will apparently be xDSL, which is
currently specified to use ATM. The local loops are an interesting story.
They run on copper wire that belong to the local phone company, but
recent regulations require the phone company to permit competitors to
install gear at the telco central offices to interface the local loops.
A few small companies have succeeded in getting past the telcos' delaying
tactics and actually installing gear, but Sprint is likely to get tied
up forever in arcane disputes, so don't expect to see it in your house
any time soon. Among other things, many new local loops don't go all the
way from your house to the central office, but instead terminate in a
pylon in your neighborhood, with the voice converted to PCM on a fiber.
Thus, a competitor's switch in the CO cannot connect directly to your local
loop anyway.
OK then, what's the REAL meaning of this announcement? I think the real
meaning is that Sprint has invested heavily in ATM and is now panicing
because ATM may become obsolete before it really gets used much. In
particular, AT&T recently announced that their vision of the future is
to run naitve IP directly on the fiber, removing the SONET and ATM layers.
voice will then run over IP. This is a data net with a voice overlay,
instead of the current voice-oriented (i.e., TDM, constant bit rate) net
with a data overlay.) One piece of evidence that Sprint paniced is that
their entire multi-page announcement never once mentions ATM.
What does all this mean? Well, it means that the future is arriving
at a rate that large companies have trouble dealing with. The distributed
IP protocol, with its distributed quasi-informal standards organization,
has triumphed over the more centralized, top-down standards promulgations
of the ITU and even over the industry-driven ATM forum. Furthermore, the
fundamental driving assumption of ATM (that IP-style switching is too
CPU-intensive to compete.) has been decisively refuted, even though the
analysis was forward-looking and relatively recent. It also means that
we may have to add AT&T and cisco to the short list of companies (Microsoft,
Intel) that utterly dominate a major high-tech area.