Home Forums MapReduce Files are already split in data node in blocks while copy file from local to hdf

Tagged: 

This topic contains 1 reply, has 2 voices, and was last updated by  Steve Loughran 6 months, 2 weeks ago.

  • Creator
    Topic
  • #55418

    Hutashan Chandrakar
    Participant

    Files are already split in data node in blocks while copy file from local to hdfs then what is use of input splits in mapreduce framework

Viewing 1 replies (of 1 total)

You must be logged in to reply to this topic.

  • Author
    Replies
  • #55421

    Steve Loughran
    Participant

    Input splits let the mappers work on fractions of a larger file. As there’s usually three copies of each file in HDFS, you can run three mappers against different parts of the same block, giving you 3x the bandwidth, so hopefully 3x the performance

    Collapse
Viewing 1 replies (of 1 total)