LinuxWorld: A software approach to clustering – Massive results from little pieces

“You’ve probably heard of Linux’s use in highly visible
clustering projects: for example, Google’s 4,000-machine cluster,
used to process millions of Internet search queries a day, or
Beowulf scientific clusters that provide supercomputing platforms
to government and university researchers at very low costs.”

“A recent example of Linux’s role in scientific clustering is
the work done at the Sanger Centre in Cambridge, England, where a
cluster of more than 300 Unix systems and 60 Linux machines have
been hard at work decoding one of the great puzzles of our time:
the human genome.”

“After the PolyServe/SuSE alliance was announced, I spoke with
PolyServe cofounders Michael Callahan, currently its CTO, and Roy
Friedman, currently a technical advisor for PolyServe and a
full-time professor of computer science (distributed processing and
networking), about Linux clustering, in particular the PolyServe
entry….”

“Understudy 1.3, the current shipping version, is a front-end
clustering tool. Unlike a similar product, TurboLinux’s
TurboCluster, Understudy provides a distributed solution. While
TurboCluster runs on a dedicated machine that sits between the
network and the servers, Understudy runs on each server and
controls the cluster in a collaborative fashion. With Understudy,
cluster administration can be done from either server or another
machine on the network.“

Complete Story