What is distributed computing (DC)?
Computers in science are nowadays increasingly being used to help scientists with research. In some areas (for example, in Climate Research or gene analysis) very much computing power is required for these kind of simulation and other lengthy calculations. The common solution to this problem are so-called supercomputers. These are quite large and usually consists of many processors, which are connected by complicated electronics to be able to work together on this issue. However, such supercomputers costs several million euros, what makes their purchasing affordable only to institutions with a huge budget. Research projects without major grants can not keep up, unfortunatelly.
Why should only a single, still quite expensive computer work on the problem? It's much smarter to divide the task in work units and solve it on multiple computers in parallel. Then each of these computers don't need to have such a tremendous power.
And this is precisely the part of the chain at which you are needed. Even if they can use completely normal PCs, dividing the task in many work units, it is not possible for the most of the research institutions to buy several hundred or even thousands of PCs. On the other hand, your computer usually has processing power, which is not used. While you are using your computer to write letters, or when the screen saver is running, while you are at lunch, then your computer uses only a very small proportion of it's capacity. Isn't it then a great idea to make that unused capacity available to solve scientific problems?
With the widespread use of the Internet the technical basis is in place to carry out such projects. Therefore a research team provides a server on the internet and a small program called Client, to be downloaded to your own computer. If this program is activated on your computer, it connects to the server on the Internet, and downloads data packets (also called Work unit) with instructions. These are just a tiny part of the original problem. On this part the computer will then work (the time needed varies depending on the project from a few minutes to hours or days), sends the results back afterwards and receives new data immediately. The program uses only the idle resources of the computer, i.e. even while working with other programs. Usually no delays on the response time are produced by the research program. Some of the programs are coded in that way, that they run as a screen saver and only becomes active, if no one is working at the computer.
More names and techniques
Distributed computing is often equated with a particular concept, but this is really only one possible way of using DC. In fact, there are various techniques to enable DC:
- Cloud Computing
- Client-server architecture (almost all the projects listed here on Rechenkraft.net
I want to participate!
Now, if you are interested in such projects, then take a look at our list of active projects. You may have the projects sorted by categories. Once you've decided on one project to start with, you need only the Client program for your operating system. And here you go by beeing a member of the research community. Become even a member of one of the teams. With some projects you may even win prices. If you have any questions or would like to share your experiences with others, we recommend our forum. Please check also our FAQ.
- 20.10.2009: Volunteering Computers for Science - Users Make Their Home PCs Available to Chase Medical Breakthroughs