No reducers in mapreduce hadoop

Question

I have only one key emitted from the converter to the gearbox, and I have not installed any gearboxes 10. So, one gearbox will work on this key and what other 9 gearboxes will do?

Answer 1

The remaining 9 reducers will go through their life cycle, as usual, they simply will not have any keys / values ​​to process after they start, so they will stop quickly. Thus, you will spend resources while they irrevocably run.

You, as a rule (most of the output formats do this) will also find that in the end you will receive a part file for reduction, but you did not write anything. The part file will not contain any actual data only for file metadata, for example, gzip headers.

Answer 2

The remaining 9 gearboxes will work until their slot time runs out.

They will not have any K / V pairs for a quick and quick process.

You can use customizable dividers to evenly distribute card outputs across all gearboxes; - at least for the first level; and finalaly combine through one gearbox in the very last phase - thus reducing the computational load in most of the reduction phases.