MapReduce

Computer Science

How It Works

MapReduce is a distributed computational programming framework designed for scalable, parallel processing of massive data clusters. Computations segregate across two core functions: the Map phase, in which independent worker nodes transform unstructured records into intermediate key-value tuples, and the Reduce phase, where workers summarize values grouped by common key. An intermediate Shuffle and Sort phase partitions and routes tuples across network fabrics.

Governing Equation
Map: (k_1, v_1) → list(k_2, v_2) , Reduce: (k_2, list(v_2)) → list(k_3, v_3)