Computer clusters may sound like something only braniacs working for the government or some gene mapping program could use, but it has many applications. If you don't know why one would want to do such a thing, it may sound profoundly boring at first. However, when you learn that there are, in the works, development projects to make a single workstation or server link to dozens, perhaps hundreds or thousands of others to distribute the workload, thus providing compounded computer power - in other words, make your computer ten or an hundred times faster and on the cheap - then you may decide clustering sounds pretty cool.
Clustering is fairly geeky stuff right now. Very few programs are designed to capitalize on the capabilities of clusters, and the technology is mostly applied in military research and gene mapping. There are larger and more time-delayed clusters which are actually referred to as distributed computing, like the SETI project: a project where workloads are distributed to thousands of computers to process data used to search for extraterestrial intelligent life. Once clustering is seemless and transparent it can become a very powerful cost cutting feature for companies that have dozens, hundreds, or thousands of personal computers that don't presently do much most of the time. These otherwise dormant computers can be linked to produce amazingly fast data crunching for corporate or hobbiest desires.
Two plans in the works:
http://openssi.org/
http://openmosix.sourceforge.net/