mainly for parallelism and charge, depend of the resources and context of your cluster, and size of input data for get compromise performance
Suppose there is data of 256 MB. There will be 4 blocks of 64 MB. So the number of mappers will be 4. This is automatically decided . So what will the option to set number of mappers do?If I set it to 5 does it make sense when it should be 256/64=4 and not 5?
Thanks