The mapper output (intermediate data) is stored on the Local file system (NOT HDFS) of each individual mapper nodes. This is typically a temporary directory location which can be setup in config by the Hadoop administrator. The intermediate data is cleaned up after the Hadoop Job completes.
