Vinit

• Sep 26th, 2017

 

1. In multi files, there can be skew spillage due to imbalance in amount of CPU time
2. If data is not evenly partition, some of the partition can contain zero records and which will consume memory

Jatin

• Apr 30th, 2018

 

1. When Graph starts there is time taken for start and shut down, so if data in multifile is not much use serial processing as over all time taken will not have any significant difference.

2. If multi file contains uneven no of key data and requirement is to use key components then skew will be high and degrades performance.