Improve SBE perf for bestbuy_query and bestbuy_agg when sharded

XMLWordPrintableJSON

    • Type: Improvement
    • Resolution: Unresolved
    • Priority: Major - P3
    • None
    • Affects Version/s: None
    • Component/s: None
    • None
    • Query Execution
    • None
    • None
    • None
    • None
    • None
    • None
    • None

      On bestbuy_query and bestbuy_agg workloads SBE engine experiences worse performance changes than classic engine when comparing sharded vs non-sharded collections.

      Example:

      Query Engine Non-sharded ops/sec Sharded ops/sec Diff
      count_no_predicate-useAgg {thread_level: 1} Classic 0.75 0.50 33%
      count_no_predicate-useAgg {thread_level: 1} SBE 1.12 0.57 49%
      find_project_6_sort_indexed_skip_limit_1-noAgg {thread_level: 1} Classic 1472 378 74%
      find_project_6_sort_indexed_skip_limit_1-noAgg {thread_level: 1} SBE 2035 412 79%

       

       

            Assignee:
            [DO NOT USE] Backlog - Query Execution
            Reporter:
            Ivan Fefer
            Votes:
            0 Vote for this issue
            Watchers:
            4 Start watching this issue

              Created:
              Updated: