1.4 Linux, open source, and clusters

< Day Day Up >

Through the efforts of thousands of people around the globe, Linux is being developed at a faster pace than any operating system in history. The basic idea of open source is very simple: when programmers can read, redistribute, and modify the source code for a piece of software, the software evolves. People improve it, people adapt it, and people fix bugs. This process can happen at a speed that, if one is used to the slow pace of conventional software development, seems astonishing.

Operating systems such as Linux, which can be obtained virtually for free, provide a very economical solution to operating system licensing on large numbers of nodes. In addition, with the popularity of Linux, there are many tools, utilities, and applications available to help build and manage a cluster. Many of these programs are available either for free or for a very reasonable cost.

Additional information on Open Source can be found at:

http://www.opensource.org/

Linux clusters (on Intel® processors) have became very popular in academia. They represent an inexpensive platform for computer science issues like parallel programming, development of parallel tools, and management of distributed systems. At the same time, Intel processor-based Linux clusters have begun to appear in government and industrial research installations, initially as proof of concept systems, and now as production machines for scientific applications.

Linux clusters represent a better price/performance opportunity that takes advantage of a familiar architecture and programming model. Linux is a well understood UNIX® variant that, because of its Open Source connection, will probably not be diverged into vendor-specific versions.

Using economical software such as Linux and Linux-based programs on Intel-based hardware such as IBM xSeries rack-mounted servers makes the overall cost of implementing a cluster much lower than it has been in the past. Therefore, many enterprises are starting to make use of such clusters in environments where they have previously not been justified.

In the following sections, we describe the different types of nodes one often finds in clusters and the types of functions that must be carried out to successfully deploy and operate a cluster.

< Day Day Up >