Monday, August 8, 2011

What Hadoop is Good For

Hadoop is ideal for the following situations:
  • Processing lots of unstructured data
    • Examples being books, journals, text files, audio and video files.
  • Situations where jobs can be easily parallel-ized.
  • Situations where you can get your hands on lots of cheap, commodity hardware.
It probably isn't best suited for:
  • Intense computing with small data sets.
    • In other words, not ideal for running a fibonacci sequence.
  • Jobs that can't be parallel-ized very easily.
  • Situations where one needs interactive results
    • Hadoop is designed for offline processing.
  • Probably not a good wholesale replacement for a relational database server.

    No comments:

    Post a Comment