What is the name of Google's distributed computing framework?

Google’s distributed computing framework was MapReduce. It is a programming model and implementation for processing very large datasets across clusters of computers by dividing work into parallel stages.

A MapReduce job normally begins with a map function, which reads input records and emits intermediate key-value pairs. During the shuffle stage, records with the same key are grouped together. A reduce function then combines or summarizes each group—for example, adding the counts for every occurrence of a word.

Google engineers Jeffrey Dean and Sanjay Ghemawat described the system in a 2004 paper. Its runtime handled difficult infrastructure work such as partitioning data, scheduling tasks, transferring results, and recovering from machine failures. Google used MapReduce for tasks including rebuilding its web index. The model also inspired Apache Hadoop’s open-source implementation, although Google later moved much of its internal big-data processing toward newer systems.