There's another trend in workstation computing. You
can see it from distributed.net. Basically, if you have a partitionable
computing space, you can cut up the computing and have it execute over
some number of low end machines using their 'extra' cycles. Distributed.net
is using approximately 14k volunteers to try and crack RC-64 by brute force.
You see this approach in closed house companies too. For instance my friend
works at is doing digital movie effects. Movie houses traditionally
bought a large number of high end SGI workstations running Irix, but now they
simple lease a couple of thousand high end PC boxes, run linux, and
allocate jobs based on utilization. After the movie, they return all the machines
and the leasing company sells them off as surplus. Again, it's the scalability
that is the key selling point.
Anyways, thanks for the article.
Greg
Barry Walker wrote:
>
> Here's an interesting quote --
>
> >> By the way, the cost of this Sun E10000 system is around $10 million,
> and was essentially running >> just simple database queries, each run
> serially, one after another for this particular TPC-D test,
> >> which makes it a very expensive personal computer. The fundamental
> point: There is no
> >> concurrency of queries from multiple users demonstrated in this
> particular test.
>
> You've got to be kidding. Someone would pay $10 million for some SUN
> server that can only do 1 I/O at a time?
>
> Here's a link to the entire article.
>
> http://www.s390.ibm.com/marketing/gf225122.html
>
> Barry
-- Greg Bolcer email: gbolcer@endtech.com web: http://www.endtech.com work: 714.505.4970 cell: 714.928.5476 fax: 603.994.0516