Imagine that you've been assigned the task of pushing a very heavy
car up a hill. You're allowed to recruit people who aren't doing anything else to help you move the car. You've got two choices: You can look around for one person big and strong enough to do it all by him or herself, or you could grab several average people to push together. While you might eventually find someone large enough to push the car alone, most of the time it will be easier to just gather a group of average-sized people. It might sound strange, but
shared computer systems use the same principle.
When a computational problem is really complex, it can take a single
computer a long time to process it -- millions of days, in some cases. Even
supercomputers have processing limitations. They're also rare and expensive. Many research facilities require a lot of computational power, but don't have access to a supercomputer. For these organizations, shared computing is often an attractive alternative to supercomputers.
Shared computing is a kind of
high-performance computing. A shared computing system is a network of computers that work together to accomplish a specific task. Each computer donates part of its processing power -- and sometimes other resources -- to help achieve a goal. By networking thousands of computers together, a shared computing system can equal or even surpass the processing power of a supercomputer.
Most of the time, your computer isn't using all of its computational resources. There are other times when you might have your computer on, but aren't actually using it. A shared computing system takes advantage of these resources that otherwise would remain unused.
Shared computing systems are great for certain complex problems, but aren't useful for others. They can be complicated to design and administer. While several computer scientists are working on a way to standardize shared computing systems, many existing systems rely on unique hardware, software and architecture.
