MapReduce was one of the seminal features of Hadoop that was arguably the most instrumental in bringing it to prominence. MapReduce works on the principle of dividing larger tasks into smaller subtasks. Instead of delegating a single machine to compute a large task, a network of smaller machines can instead be used to complete the smaller subtasks. By distributing the work in this manner, the task can be completed much more efficiently relative to using a single-machine architecture.

This is not dissimilar to how we go about completing work in our day-to-day lives. An example will help to make this clearer.