grid computing - technical definition

A parallel processing architecture in which CPU resources are shared across a network, and all machines function as one large supercomputer. It allows unused CPU capacity in all participating machines to be allocated to one application that is extremely computation intensive and programmed for parallel processing.

There Is a Lot of Idle Time
In a large enterprise, hundreds or thousands of desktop machines sit idle at any given moment. Even when a user is at the computer reading the screen and not typing or clicking, it constitutes idle time. These unused cycles can be put to use on large computational problems. Likewise, the millions of users on the Internet create a massive amount of wasted machine cycles that can be harnessed instead. This is precisely what the Search for Extraterrestrial Intelligence program does with Internet users all over the world (see SETI).

Naturally, grid computing over the Internet requires more extensive security than within a single enterprise, and robust authentication is employed in such applications.

Peer-to-Peer and Distributed Computing
Grid computing is also called "peer-to-peer computing" and "distributed computing," the latter term first coined in the 1970s, which had no relationship to this concept. Grid computing is also known as "utility computing," although that term is more widely used with third-party datacenters that supply raw computing power. See utility computing, distributed computing, PC philanthropy and anticiparallelism. See also peer-to-peer network and peer-to-peer.