> On May 11, 2015, 5:59 p.m., Kevin Sweeney wrote: > > src/main/java/org/apache/aurora/scheduler/storage/db/DbModule.java, line 84 > > <https://reviews.apache.org/r/34046/diff/1/?file=955485#file955485line84> > > > > Link to upstream ticket would be good here.
AFAICT there is not an issue tracking this. H2 operates in a very single-developer sense, so there's not much in the way of formal project management for big milestones. - Bill ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/34046/#review83242 ----------------------------------------------------------- On May 11, 2015, 5:58 p.m., Bill Farner wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/34046/ > ----------------------------------------------------------- > > (Updated May 11, 2015, 5:58 p.m.) > > > Review request for Aurora, Kevin Sweeney and Maxim Khutornenko. > > > Bugs: AURORA-1311 > https://issues.apache.org/jira/browse/AURORA-1311 > > > Repository: aurora > > > Description > ------- > > Upgade h2 to 1.4.187. > > > Diffs > ----- > > build.gradle 829b81392d6eaf9f1eb7ae1ee89ff60b4fb32731 > src/main/java/org/apache/aurora/scheduler/storage/db/DbModule.java > c8df88b77b9a2017c48bdc0c9a0477927ba0b179 > > Diff: https://reviews.apache.org/r/34046/diff/ > > > Testing > ------- > > I ran the benchmarks before and after this change. In nearly every case, > performance is marginally better (though the difference is generally within 1 > stddev). Note, however, that i needed to bump the heap size on the benchmark > suite. This is because my first run encountered an OOM on one microbenchmark > after the change. I have no evidence whether this is because the new H2 is > more memory-hungry, or if that happens occasionally and it was poor luck. > > Benchmark results before (old H2 version): > ``` > SchedulingBenchmarks.ClusterFullUtilizationBenchmark.runBenchmark 551189.355 > ± 14341.947 ops/s > SchedulingBenchmarks.InsufficientResourcesSchedulingBenchmark.runBenchmark > 51502.506 ± 975.321 ops/s > SchedulingBenchmarks.LimitConstraintMismatchSchedulingBenchmark.runBenchmark > 4427.202 ± 162.303 ops/s > SchedulingBenchmarks.PreemptorSlotSearchBenchmark.runBenchmark 52.291 ± 1.402 > ops/s > SchedulingBenchmarks.PreemptorSlotSearchBenchmark.runBenchmark 50.159 ± 1.907 > ops/s > SchedulingBenchmarks.PreemptorSlotSearchBenchmark.runBenchmark 52.891 ± 0.643 > ops/s > SchedulingBenchmarks.PreemptorSlotSearchBenchmark.runBenchmark 50.845 ± 0.689 > ops/s > SchedulingBenchmarks.ValueConstraintMismatchSchedulingBenchmark.runBenchmark > 51955.901 ± 1420.778 ops/s > StatusUpdateBenchmark.runBenchmark 0.176 ± 0.019 ops/s > StatusUpdateBenchmark.runBenchmark 0.038 ± 0.001 ops/s > StatusUpdateBenchmark.runBenchmark 0.010 ± 0.001 ops/s > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 2025.121 ± 30.865 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 115.166 ± 10.361 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 9.490 ± 4.451 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 0.241 ± 0.277 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 2010.467 ± 161.806 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 2009.343 ± 172.806 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 2006.227 ± 86.458 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 2008.623 ± 14.901 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 4368.957 ± 53.245 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 3932.150 ± 319.144 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 2069.126 ± 33.426 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 224.580 ± 10.336 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 19.286 ± 0.713 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 6287.064 ± 156.134 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 533.524 ± 101.918 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 12.988 ± 0.856 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 2.309 ± 0.105 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 5581.662 ± 366.345 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 5487.590 ± 179.228 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 5337.478 ± 113.176 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 5499.892 ± 166.960 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 13288.963 ± 661.267 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 11349.901 ± 422.637 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 5984.420 ± 454.136 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 995.175 ± 84.046 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 23.697 ± 1.261 > UpdateStoreBenchmarks.JobDetailsBenchmark.run 325.116 ± 48.465 ops/s > UpdateStoreBenchmarks.JobDetailsBenchmark.run 71.977 ± 3.475 ops/s > UpdateStoreBenchmarks.JobDetailsBenchmark.run 37.514 ± 2.345 ops/s > ``` > > Benchmark results after (new H2 version): > ``` > SchedulingBenchmarks.ClusterFullUtilizationBenchmark.runBenchmark 551152.813 > ± 14762.509 ops/s > SchedulingBenchmarks.InsufficientResourcesSchedulingBenchmark.runBenchmark > 52547.639 ± 1618.220 ops/s > SchedulingBenchmarks.LimitConstraintMismatchSchedulingBenchmark.runBenchmark > 4577.951 ± 90.308 ops/s > SchedulingBenchmarks.PreemptorSlotSearchBenchmark.runBenchmark 53.949 ± 1.366 > ops/s > SchedulingBenchmarks.PreemptorSlotSearchBenchmark.runBenchmark 52.915 ± 0.563 > ops/s > SchedulingBenchmarks.PreemptorSlotSearchBenchmark.runBenchmark 53.662 ± 0.522 > ops/s > SchedulingBenchmarks.PreemptorSlotSearchBenchmark.runBenchmark 49.356 ± 1.253 > ops/s > SchedulingBenchmarks.ValueConstraintMismatchSchedulingBenchmark.runBenchmark > 52106.306 ± 1465.181 ops/s > StatusUpdateBenchmark.runBenchmark 0.175 ± 0.017 ops/s > StatusUpdateBenchmark.runBenchmark 0.038 ± 0.001 ops/s > StatusUpdateBenchmark.runBenchmark 0.010 ± 0.001 ops/s > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 2084.183 ± 45.789 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 118.781 ± 3.092 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 10.318 ± 1.474 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 0.469 ± 0.459 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 2063.066 ± 40.586 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 2052.789 ± 41.723 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 2037.179 ± 23.099 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 2030.504 ± 51.156 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 4436.401 ± 85.899 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 4053.142 ± 170.530 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 2089.764 ± 27.751 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 228.982 ± 5.989 > ThriftApiBenchmarks.GetAllTasksBenchmark.run 5 20.398 ± 0.588 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 6321.673 ± 34.080 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 501.385 ± 20.985 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 12.746 ± 0.829 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 2.284 ± 0.134 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 5603.220 ± 139.780 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 5798.978 ± 35.320 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 5365.175 ± 65.432 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 5454.763 ± 117.124 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 13285.499 ± 480.019 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 11852.183 ± 432.408 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 5581.147 ± 203.938 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 1042.513 ± 16.771 > ThriftApiBenchmarks.GetRoleSummaryBenchmark.run 5 24.174 ± 2.080 > UpdateStoreBenchmarks.JobDetailsBenchmark.run 355.051 ± 39.341 ops/s > UpdateStoreBenchmarks.JobDetailsBenchmark.run 79.105 ± 1.991 ops/s > UpdateStoreBenchmarks.JobDetailsBenchmark.run 39.225 ± 1.028 ops/s > ``` > > > Thanks, > > Bill Farner > >
