"Hadoop" Hadoop streaming specifies that the reducer process is not performed

Sometimes, we only need to do concurrent processing of files, and do not care about the relationship between records of the same key.

At this point, only the map function is required to process the input data.

If you do not specify the REDUCER option, the system will still execute the cat command one time by default.

How to go to an unnecessary sorting operation of the bucket.

Method One:

With Mapred.reduce.tasks set to zero, the Map/reduce framework does not create reducer tasks, and the output of the mapper task is the final output of the entire job.

Method Two:

The "-reduce None" option, which is equivalent to "-jobconf mapred.reduce.tasks=0".