I don't think Grid did much to make money, though there was a lot of talk
about how Grid would bring utility computing - as easy as plugging into a
wall socket.
I think the problems with that model are more to do with non-computing
issues than with "making it work". Things like assuring information
security are a big deal for anyone who has serious money to spend on the
computation.
I guess I don't really see why security is non-computing issue - to me,
Grid is basically irrelevant to HPC because computing is not a fungible
commodity. that is, there are vastly different, incompatible kinds of
computation, and it really matters where the computing is located and how
it's connected. if Java had conquered the world and if the network had
become too cheap to meter and all running at a gazillabit, maybe Grid might
have worked. (in spite of the fact that latency isn't really subject to
technical fixes, and even within a city, nodes will be milliseconds apart -
that is millions of clock cycles apart.)
_______________________________________________
Beowulf mailing list, [email protected]
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
