Beowulf cluster

From Citizendium
Revision as of 08:39, 11 April 2007 by imported>Eric M Gearhart (→‎Beowulf Development)
Jump to navigation Jump to search
In a Beowulf cluster, the network cables (shown in blue) that connect the eight nodes to the master serve as a bus

A Beowulf cluster is a group of personal computers linked together so that they act as one. This is known as parallel computing. This type of cluster is composed of a 'master' (which coordinates the processing power of the cluster) and usually many 'nodes' (computers that actually perform the calculations). The 'master' is typically a server, and "has more horsepower" than the individual nodes. The nodes in the cluster don't have to be identical, although to simplify deployment this is usually the case.

Usually the Beowulf 'nodes' are running Linux,[1] however this is not required, as both Mac OS X and FreeBSD clusters have been created.[2][3]

The nodes in a Beowulf cluster are networked into a small TCP/IP LAN, and have libraries and programs installed which allow processing to be shared among them.

The name Beowulf is derived from the main character in the Old English epic Beowulf.

Beowulf Development

In early 1993, NASA scientists Donald Becker and Thomas Sterling began sketching out the details of what would become a revolutionary way to build a cheap supercomputer: link low-cost desktops together with commodity, off the shelf (COTS) hardware and combine their performance.[4]

By 1994, under the sponsorship of the "High Performance Computing & Communications for Earth & Space Sciences" (HPCC/ESS)[5] project, the Beowulf Parallel Workstation project at NASA's Goddard Space Flight Center had begun.[6][7]

Popularity in High-Performance Computing

Today Beowulf systems are deployed worldwide as both as "cheap supercomputers" and as more traditional high-performance projects, chiefly in support of number crunching and scientific computing.

It should be noted that more than 50 percent of the machines on the Top 500 List of supercomputers [8] are clusters of this sort. Many on the list are built on Becker and Sterling’s own specific architecture, the Beowulf cluster.[4]

Common Beowulf clusters

There is no particular piece of software that defines a cluster as a Beowulf. Commonly used parallel processing libraries include MPI (Message Passing Interface) and PVM (Parallel Virtual Machine). Both of these permit the programmer to divide a task among a group of networked computers, and recollect the results of processing. It is a common misconception that any software will run faster on a Beowulf. The software must be re-written to take advantage of the cluster, and specifically have multiple non-dependent parallel computations involved in its execution.

References