MapReduce and Massive Data
At the current state of technology, it has become difficult to make individual computer CPU's faster; however, it has become cheap to make lots of CPU's. Networks allow fast communication between large numbers of cheap CPU's, each of which has substantial main memory and disk.
A significant challenge of modern CS is to perform large computations using networks of cheap computers operating in parallel.
Google specializes in processing massive amounts of data, particularly the billions of web pages now on the Internet. MapReduce makes it easy to write powerful programs over large data; these programs are mapped onto Google's network of hundreds of thousands of CPU's for execution.