In Hadoop Mapreduce Framework, Hadoop tries its best to run a map task on a node where the input data resides. By this, it reduces the data transfer between nodes on a cluster.
This technique of moving computation to data but not bringing data to computation node is called as Data Locality Optimization.
