This document provides an introduction to MapReduce, including:
1) MapReduce is a programming model and framework that supports distributed processing of large datasets across clusters of machines. It handles parallelization, fault tolerance, and scheduling.
2) The MapReduce model consists of map and reduce functions that specify the computation. Mappers process input key-value pairs and produce intermediate outputs. Reducers combine intermediate outputs by key.
3) The execution framework handles scheduling, data/code distribution, synchronization, and fault handling to execute the user-specified map and reduce functions on large datasets in parallel across a cluster.