Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-21008

20% Single thread performance regression on Insert.EmptyCapped and Insert.CappedSeqIntID on wiredTiger

    • Fully Compatible
    • ALL
    • 0

      20% regression in single thread performance compared to baseline. This is using a recent compare to 3.1.7.

      To recreate the empty capped in the shell:

      var setup = function( collection ) {
                        collection.drop();
                        collection.runCommand( "create", { capped : true,
                                                           size : 32 * 1024 } );
      }              
      use test0
      setup(db.Insert_EmptyCapped0)
      benchRun({"ops":[{"op":"insert","doc":{},"ns":"test0.Insert_EmptyCapped0","safe":false,"w":0,"j":false,"writeCmd":true}],"seconds":30,"host":"127.0.0.1:27017","parallel":1})
      

      Original failure message:

      [2015/10/08 12:10:34.838] Override in references for test Insert.EmptyCapped.SeqIntID
      [2015/10/08 12:10:34.838] regression found on 1: drop from 12386.47 ops/sec (commit 7d7f4) to 9807.34 ops/sec for comparison Baseline Comparison 3.0.6-Baseline. Diff is 2579.13 ops/sec (20.82%), noise level is 91.96 ops/sec and multiple is 2.00
      [2015/10/08 12:10:34.838] regression found on 2: drop from 25065.53 ops/sec (commit 7d7f4) to 20346.30 ops/sec for comparison Baseline Comparison 3.0.6-Baseline. Diff is 4719.23 ops/sec (18.83%), noise level is 234.05 ops/sec and multiple is 2.00

      [2015/10/08 12:10:34.839] Override in references for test Insert.EmptyCapped
      [2015/10/08 12:10:34.839] regression found on 1: drop from 12445.05 ops/sec (commit 7d7f4) to 9495.62 ops/sec for comparison Baseline Comparison 3.0.6-Baseline. Diff is 2949.43 ops/sec (23.70%), noise level is 154.85 ops/sec and multiple is 2.00
      [2015/10/08 12:10:34.839] regression found on 2: drop from 25188.91 ops/sec (commit 7d7f4) to 20805.12 ops/sec for comparison Baseline Comparison 3.0.6-Baseline. Diff is 4383.79 ops/sec (17.40%), noise level is 400.21 ops/sec and multiple is 2.00

            Assignee:
            martin.bligh Martin Bligh
            Reporter:
            david.daly@mongodb.com David Daly (Inactive)
            Votes:
            0 Vote for this issue
            Watchers:
            7 Start watching this issue

              Created:
              Updated:
              Resolved: