Map Reduce - Data (Stream) - <key, value> pairs
Table of Contents
1 - About
The MapReduce framework operates exclusively on <key, value> pairs conceivably of different types.
The key and value classes have to implement:
- the Writable interface (to be serializable)
- the WritableComparable interface (to facilitate sorting)
See also MapReduce - InputFormat
2 - Articles Related
3 - Example of pipeline
Input and Output types of a MapReduce job forms a pipeline:
(input) <k1, v1> -> map -> <k2, v2> -> combine -> <k2, v2> -> reduce -> <k3, v3> (output)
where:
- MapReduce - Shuffling (Combine)
- MapReduce - Reducer