Advantages
- By having many processors computing a portion of a large task, the computational power of a distributed system is far greater than that of an individual processor. Many small processors linked together can rival and supersede the computational power of a supercomputer
- Reliability is improved by the use of many smaller processors since if one breaks down, another computer can easily take its place. The impact of a crash of one of the many computers is far less than if one supercomputer crashes.
- The connection of many smaller computers is also cheaper to implement than trying to improve the processor speed of a supercomputer.
- Many computers sit idle for long periods of time and that wasted computer time can be used for computational tasks.
To show the power of distributed computing, consider that a typical computer will execute 100 million instructions per second, with 128 MB of memory, and 10 GB of disk storage. It will use 100 W of electricity with an approximate cost of $1000. Then, 100 million of these typical machines are connected to the internet. The price tag is $100 billion, all to be used for free with distributed computing, all during unproductive time. Impressive!
The interactive demonstration on the following page shows how multiple distributed computers can (on average) accomplish a linear search in less time than a single computer can.
You will notice that there is a list of numbers and a two selections you can make:
- # of computers
- search target
Choose either two or four computers and a number to find in the list. Click the start button and the computers at the top of the screen will search for the same number as the single computer at the bottom of the screen.
