Details

    • Type: Sub-task
    • Status: Resolved
    • Priority: Major
    • Resolution: Invalid
    • Affects Version/s: None
    • Fix Version/s: None
    • Component/s: None
    • Labels:
      None

      Description

      Continuation of parent.
      We can also do this optimization for seek hints. This would allow filters and coprocessors to be more liberal with seek hints, as seeking when unnecessary is less of perf detriment.

      It's not quite as clear cut, since in order to check, we actually do need to create the seek hint Cell. Then when we actually seek, we need to create it again. Need to test carefully.

      1. 13109.txt
        2 kB
        Lars Hofhansl
      2. 14418-0.98.txt
        3 kB
        Lars Hofhansl

        Activity

        Hide
        lhofhansl Lars Hofhansl added a comment -

        W.I.P. patch. Need to perf test.

        Show
        lhofhansl Lars Hofhansl added a comment - W.I.P. patch. Need to perf test.
        Hide
        lhofhansl Lars Hofhansl added a comment -

        (also gets rid of the NEXT matchcode, which appears not be used anywhere)

        Show
        lhofhansl Lars Hofhansl added a comment - (also gets rid of the NEXT matchcode, which appears not be used anywhere)
        Hide
        lhofhansl Lars Hofhansl added a comment - - edited

        0.98 version. In my tests with Phoenix I did not see noticeable improvement, though. So might just close. Doing some more test until then.

        (turns out in my schema Phoenix would not use SKIP_SCANs)

        Show
        lhofhansl Lars Hofhansl added a comment - - edited 0.98 version. In my tests with Phoenix I did not see noticeable improvement, though. So might just close. Doing some more test until then. (turns out in my schema Phoenix would not use SKIP_SCANs)
        Hide
        lhofhansl Lars Hofhansl added a comment -

        Tried with some iterations of FuzzyRowFilter. This actually makes it slower. Probably gonna close this as invalid.

        Show
        lhofhansl Lars Hofhansl added a comment - Tried with some iterations of FuzzyRowFilter. This actually makes it slower. Probably gonna close this as invalid.
        Hide
        lhofhansl Lars Hofhansl added a comment -

        Yeah. Creating the Cell, just to return the hint is what's taking up more time than the actual seek. Closing for now. I might revisit this.

        James Taylor, FYI.

        Show
        lhofhansl Lars Hofhansl added a comment - Yeah. Creating the Cell, just to return the hint is what's taking up more time than the actual seek. Closing for now. I might revisit this. James Taylor , FYI.
        Hide
        lhofhansl Lars Hofhansl added a comment -

        (if I could push the comparison to the Filter, which would avoid materializing the Cell, that might work, but that we need a new Filter API)

        Show
        lhofhansl Lars Hofhansl added a comment - (if I could push the comparison to the Filter, which would avoid materializing the Cell, that might work, but that we need a new Filter API)

          People

          • Assignee:
            Unassigned
            Reporter:
            lhofhansl Lars Hofhansl
          • Votes:
            0 Vote for this issue
            Watchers:
            4 Start watching this issue

            Dates

            • Created:
              Updated:
              Resolved:

              Development