MapReduce: a technique for massively distributing computation

Olivier Dagenais

Abstract

Scaling up to deal with increasing amounts of data and demand used to mean moving to bigger and more powerful servers, but not anymore. The use of MapReduce, a technique based on functional programming concepts, coupled with many, less expensive commodity servers can provide performance that scales with the number of computers participating in the cluster, minus the exorbitant costs. Come learn this technique pioneered by Google and see how you can write and test your programs on one computer so they can run as easily on one thousand computers.