Uploaded image for project: 'Hadoop Common'
  1. Hadoop Common
  2. HADOOP-3446

The reduce task should not flush the in memory file system before starting the reducer

    XMLWordPrintableJSON

Details

    • Improvement
    • Status: Closed
    • Critical
    • Resolution: Fixed
    • None
    • 0.19.0
    • None
    • None
    • Reviewed

    Description

      In the case where the entire reduce inputs fit in ram, we currently force the input to disk and re-read it before giving it to the reducer. It would be much better if we merged from the ramfs and any spills to feed the reducer its input.

      Attachments

        1. 3446-0.patch
          14 kB
          Christopher Douglas
        2. 3446-1.patch
          15 kB
          Christopher Douglas
        3. 3446-2.patch
          30 kB
          Christopher Douglas
        4. 3446-3.patch
          35 kB
          Christopher Douglas
        5. 3446-4.patch
          45 kB
          Christopher Douglas
        6. 3446-5.patch
          41 kB
          Christopher Douglas
        7. 3446-6.patch
          41 kB
          Christopher Douglas
        8. 3446-7.patch
          41 kB
          Christopher Douglas

        Issue Links

          is depended upon by

          Improvement - An improvement or enhancement to an existing feature or task. HADOOP-4063 Separate spill thresholds for serialization/accounting in MapTask

          • Minor - Minor loss of function, or other problem where easy workaround is present.
          • Closed

          Activity

            People

              cdouglas Christopher Douglas
              omalley Owen O'Malley
              Votes:
              0 Vote for this issue
              Watchers:
              5 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: