Map Reduce
Stc
Date: 2008-06-05
Time: 11:00
Room: BBL room 471
Speaker: Jeiel Schalkwijk
Title: MapReduce
Abstract
MapReduce is a framework developed and used by Google to run programs on
clusters of commodity PCs.
Programmers program the logic of their program in a map and reduce
function, and the framework automatically executes those programs on many
PCs.
All the work of distributing the computation is done by the framework.
This includes partitioning the input data, scheduling computation tasks
and handling malfunctioning nodes.
The abstraction that the MapReduce framework provides is similar to that
provided by the map and reduce (or fold in Haskell) functions found in
functional languages.
During the talk I will show you how to program using the MapReduce
programming model and tell you how MapReduce executes your programs.
And there is no need to despair if you do not work for Google. There are
many MapReduce implementations, including the open source Hadoop from
Apache.
