MapReduce
In plain terms
A model for processing huge datasets across many machines.
Going deeper
MapReduce splits work into a map phase (transform records in parallel) and a reduce phase (aggregate results), enabling massive distributed computation. It hides the parallelism details.
Under the hood
Pioneered by Google and implemented in Hadoop; handles fault tolerance and data locality, though newer engines like Spark improved on it.
Examples
Log analysis, word counting, Hadoop, large-scale ETL, indexing the web.
Think of it like...
Like many clerks each tallying a stack of receipts (map), then combining their subtotals (reduce).