Uploaded image for project: 'Spark'
  1. Spark
  2. SPARK-3280

Made sort-based shuffle the default implementation

    XMLWordPrintableJSON

Details

    • Improvement
    • Status: Resolved
    • Major
    • Resolution: Fixed
    • None
    • 1.2.0
    • Shuffle, Spark Core
    • None

    Description

      sort-based shuffle has lower memory usage and seems to outperform hash-based in almost all of our testing.

      Attachments

        1. hash-sort-comp.png
          40 kB
          Burak Yavuz

        Issue Links

          Activity

            People

              rxin Reynold Xin
              rxin Reynold Xin
              Votes:
              0 Vote for this issue
              Watchers:
              11 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: