Clusters are quite versatile in how they can be used, they can perform all of the usual mathematical tasks that you would expect a supercomputer to be able to do, and because of the standardization of libraries that are used on all commercial supercomputers, and so porting applications to a Linux cluster is fairly easy. One particular advantage of a cluster is that if more processing power is required it can be added “on the fly” while with a Supercomputer this would just not be possible.
File Servers
With a parallel Linux cluster, however, it is possible to construct a file server that is faster than most commercial high-performance, dedicated file servers. That can be done with a Linux cluster because of the multiple systems and multiple network connections that these clusters have. Such as server is useful for large applications like database systems where the performance bottlenecks are when data cannot be read from disk quick enough.
Web Servers
Linux clusters, because of their design tend to create very good redundant systems. Linux clusters can be designed so that they are all running duplicate copies of a particular application, and then have monitoring systems that detect whether a particular system has failed and then will share the load with the other systems in the cluster, this software will also share all tasks evenly across all the machines in the cluster so that no one machine is too heavily loaded, an example of a web server that use a Linux cluster, is the 4000 plus cluster that runs the google.com search engine.
Flight Simulators
Simulating a plane is a fairly complicated process, with so many things going on such as calculating the physics of the plane, the weather and the controls of the plane and so Linux clusters are ideal for this, as each node could have a separate task or a group of nodes could share a particular task. Companies such as Boeing and Aerobus Industries use clusters to make their flight simulators, they use it for the development of planes that they manufacturer and to test experimental concepts for future planes. It is a lot cheaper and safer for these companies to do their test with a computer system, as they can test a lot of situations such as a plane crash without actually doing it. Car manufacturers use simulators for the same reasons.
Virtual Worlds
This is where by a system is use to created a virtual world or room, very similar to a chat room but it contains robots as well as users, and also has other features such as being an environment with weather, scenery and other objects. People can logon to these environments and explore them, interacting with human and non-human characters, these communities evolve over time. Linux clusters allow these to develop these concepts as nodes in the cluster can process different parts of this environment.
Other uses
Distributed computing – Distributed computing projects such as genome@home seti@home, clusters are able to process units of data quicker that single computer systems.
Game servers – these require a lot of processing power to run multiplayer games and so clustered machines could coupe with more users.