Distributed Apple
2001.02.06
Steve Sarrica
Companies like Popular
Power, Parabon, United Devices, and
Distributed Science
are currently trying to entice Internet users who own PCs, Macs, and
Linux boxes to download their client software (a Java-based distributed
multiprocessing application wrapped up in a screensaver that fires up
when the user is not actively using the computer) by offering small
payments, gift certificates, or warm fuzzies.
The companies that pay for the use of that otherwise untapped
computing power appear to budget around $5-$15 per month per client.
What if Apple picked a player in this emerging industry, anointed it
the chosen one, created an Apple-only version of the client software,
and made it part of the default installation on some future revision of
the OS? Certainly that would be worth a few bucks a month to Apple per
client.
Companies in this space currently crow about having anywhere from
40,000 to 160,000 clients operating - it seems that Apple could pretty
quickly pump those numbers up into the realm reserved for the great
granddaddy of distributed computation, SETI@home, with over 2 million client machines.
Even at a lowly buck per month per client kicked back to Apple, the
dollars could quickly get fairly interesting. Mac users might like the
ability to subsidize the cost of their connectivity. Corporations,
universities, and school systems with lots of Macs might not mind the
unexpected revenue stream, either.
The Mac OS currently offers limited support for Program Linking and
other applications, primarily graphic rendering applications, already
allow Mac users to make use of several computers linked together to
solve large, complex problems. Imagine this mostly unfulfilled promise
scaled up, designed for TCP/IP, able to work through firewalls,
multiplied by a couple of million times - you start to get the idea
that some pretty big things might get accomplished much faster than
anyone thought possible.
Take it up a notch: what if Apple could reserve a portion of the
distributed computing capacity for its own use in lieu of a portion of
its income stream from the service? The company would have access to a
distributed supercomputer inconceivably more powerful than the Cray
supercomputer Steve Jobs bought back in the 80s during his first turn
at the helm.
Take it up another notch: what if Apple could reserve a portion of
that distributed computing capacity for its users? What if the OS did
sophisticated load balancing and was able to "call for help" and tap
into the distributed supercomputer on the fly? Mac users could be
helping fellow Mac users without even knowing it. It would make
benchmarking the Mac against other computers even more difficult and
would be another arrow through the heart of the "megahertz war" between
Macs and PCs.
Granted, the logistics of all of this are daunting, but one of the
toughest parts of the job, designing and implementing the distributed
computing system, has already been solved by several companies. Apple
should pick a partner among the players and, to paraphrase Jobs,
"prepare to blow another hole in the universe." Of course, if Apple
isn't the first major computer manufacturer to implement this concept,
someone else surely will.
Links
Share your perspective on the Mac by emailing with "My Turn" as your subject.
