This document discusses MapReduce, a programming model and implementation developed by Google for processing large datasets across clusters of computers. It describes MapReduce as having two phases - a map phase where key-value pairs are input and output, and a reduce phase where the outputs of map are combined. An example is given of finding the highest CGPA for each year from a Makerere University dataset to illustrate how MapReduce works by breaking the processing into parallelized map and reduce steps.