2004
Dean and Ghemawat publish a programming model for processing large datasets across clusters of machines.
Automatic scheduling, data distribution and failure handling help make large-scale data processing practical. This paper describes Google’s implementation; it is not an open-source release.