Which among the following is correct sequence of data flow in Apached hadoop parallel MapReduce ?
MH SET - 2020, SET Exam - 2020,
Answer: C
The correct sequence is split, map, shuffle, reduce.
In Hadoop MapReduce, the input is first divided into splits, each split is processed by a map task that emits key-value pairs, the framework then shuffles and sorts those pairs so that all values for a key reach the same reducer, and the reduce tasks combine them into the final output.
The other orders put the shuffle before the map (there is nothing yet to shuffle) or reduce before shuffle (reducers would not have their grouped inputs).
Please log in or register to report this question.