----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/52669/#review160224 -----------------------------------------------------------
I have just run this patch on top of https://reviews.apache.org/r/55071/ (Disable H2 logging via SLF4J). Interestingly, in this case, the off-heap DB does no longer yield any improvements in our micro benchmarks. I assume that the H2 logging patch has reduced memory pressure significantly so that we are no longer profit from off-heap store in our benchmarks. However, off-heap storage could still be beneficial for production workloads. - Stephan Erb On Oct. 11, 2016, 8:17 p.m., Stephan Erb wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/52669/ > ----------------------------------------------------------- > > (Updated Oct. 11, 2016, 8:17 p.m.) > > > Review request for Aurora, David McLaughlin, John Sirois, and Zameer Manji. > > > Repository: aurora > > > Description > ------- > > This experiment is inspired by David's comment: "I don’t think the > storage engine matters. We just need to be able to offload it from > the Scheduler JVM. The problem with H2 isn’t SQL or anything else, > it’s the GC pressure." > > Basic idea is to switch to another storage backend: "nioMemFS stores > data outside of the VM's heap - useful for large memory DBs without > incurring GC costs" (http://www.h2database.com/html/advanced.html) > > Our micro-benchmarks look promising > > Current Master (on-heap db with latest versions): > > TaskStoreBenchmarks.DBFetchTasksBenchmark.run N/A > N/A N/A 10000 thrpt 5 72851.249 ± 15794.210 ops/s > TaskStoreBenchmarks.DBFetchTasksBenchmark.run N/A > N/A N/A 50000 thrpt 5 31626.929 ± 17326.988 ops/s > TaskStoreBenchmarks.DBFetchTasksBenchmark.run N/A > N/A N/A 100000 thrpt 5 0.078 ± 0.013 ops/s > TaskStoreBenchmarks.MemFetchTasksBenchmark.run N/A > N/A N/A 10000 thrpt 5 414.135 ± 315.838 ops/s > TaskStoreBenchmarks.MemFetchTasksBenchmark.run N/A > N/A N/A 50000 thrpt 5 68.643 ± 24.303 ops/s > TaskStoreBenchmarks.MemFetchTasksBenchmark.run N/A > N/A N/A 100000 thrpt 5 32.032 ± 13.870 ops/s > UpdateStoreBenchmarks.JobDetailsBenchmark.run N/A > 1000 N/A N/A thrpt 5 143.981 ± 78.985 ops/s > UpdateStoreBenchmarks.JobDetailsBenchmark.run N/A > 5000 N/A N/A thrpt 5 35.224 ± 25.593 ops/s > UpdateStoreBenchmarks.JobDetailsBenchmark.run N/A > 10000 N/A N/A thrpt 5 18.869 ± 3.318 ops/s > UpdateStoreBenchmarks.JobInstructionsBenchmark.run 1 > N/A N/A N/A thrpt 5 36.013 ± 19.743 ops/s > UpdateStoreBenchmarks.JobInstructionsBenchmark.run 10 > N/A N/A N/A thrpt 5 33.813 ± 11.216 ops/s > UpdateStoreBenchmarks.JobInstructionsBenchmark.run 100 > N/A N/A N/A thrpt 5 20.516 ± 10.526 ops/s > UpdateStoreBenchmarks.JobInstructionsBenchmark.run 1000 > N/A N/A N/A thrpt 5 16.564 ± 2.993 ops/s > UpdateStoreBenchmarks.JobUpdateMetadataBenchmark.run N/A > N/A 10 N/A thrpt 5 32.399 ± 21.310 ops/s > UpdateStoreBenchmarks.JobUpdateMetadataBenchmark.run N/A > N/A 100 N/A thrpt 5 35.518 ± 7.468 ops/s > UpdateStoreBenchmarks.JobUpdateMetadataBenchmark.run N/A > N/A 1000 N/A thrpt 5 19.757 ± 10.035 ops/s > UpdateStoreBenchmarks.JobUpdateMetadataBenchmark.run N/A > N/A 10000 N/A thrpt 5 10.849 ± 10.660 ops/s > > This patch (off-heap db): > > Benchmark (instanceOverrides) > (instances) (metadata) (numTasks) Mode Cnt Score Error Units > TaskStoreBenchmarks.DBFetchTasksBenchmark.run N/A > N/A N/A 10000 thrpt 5 77746.436 ± 47191.240 ops/s > TaskStoreBenchmarks.DBFetchTasksBenchmark.run N/A > N/A N/A 50000 thrpt 5 70099.087 ± 37223.642 ops/s > TaskStoreBenchmarks.DBFetchTasksBenchmark.run N/A > N/A N/A 100000 thrpt 5 30461.428 ± 22964.261 ops/s > TaskStoreBenchmarks.MemFetchTasksBenchmark.run N/A > N/A N/A 10000 thrpt 5 335.302 ± 229.328 ops/s > TaskStoreBenchmarks.MemFetchTasksBenchmark.run N/A > N/A N/A 50000 thrpt 5 61.443 ± 24.280 ops/s > TaskStoreBenchmarks.MemFetchTasksBenchmark.run N/A > N/A N/A 100000 thrpt 5 32.129 ± 13.067 ops/s > UpdateStoreBenchmarks.JobDetailsBenchmark.run N/A > 1000 N/A N/A thrpt 5 411.866 ± 396.308 ops/s > UpdateStoreBenchmarks.JobDetailsBenchmark.run N/A > 5000 N/A N/A thrpt 5 92.208 ± 65.238 ops/s > UpdateStoreBenchmarks.JobDetailsBenchmark.run N/A > 10000 N/A N/A thrpt 5 49.870 ± 44.519 ops/s > UpdateStoreBenchmarks.JobInstructionsBenchmark.run 1 > N/A N/A N/A thrpt 5 90.074 ± 31.360 ops/s > UpdateStoreBenchmarks.JobInstructionsBenchmark.run 10 > N/A N/A N/A thrpt 5 114.224 ± 48.576 ops/s > UpdateStoreBenchmarks.JobInstructionsBenchmark.run 100 > N/A N/A N/A thrpt 5 108.767 ± 25.154 ops/s > UpdateStoreBenchmarks.JobInstructionsBenchmark.run 1000 > N/A N/A N/A thrpt 5 45.995 ± 14.795 ops/s > UpdateStoreBenchmarks.JobUpdateMetadataBenchmark.run N/A > N/A 10 N/A thrpt 5 72.343 ± 77.696 ops/s > UpdateStoreBenchmarks.JobUpdateMetadataBenchmark.run N/A > N/A 100 N/A thrpt 5 84.984 ± 27.971 ops/s > UpdateStoreBenchmarks.JobUpdateMetadataBenchmark.run N/A > N/A 1000 N/A thrpt 5 89.819 ± 44.923 ops/s > UpdateStoreBenchmarks.JobUpdateMetadataBenchmark.run N/A > N/A 10000 N/A thrpt 5 31.670 ± 15.236 ops/s > > > Please be aware: The values seen here will not really carry over to > real world usage. It would therefore be awesome if one of you could > test this on a larger cluster! > > > Diffs > ----- > > src/main/java/org/apache/aurora/scheduler/storage/db/DbModule.java > e7287cec28e7b8ca978c506bfe821f261bc0ac26 > > Diff: https://reviews.apache.org/r/52669/diff/ > > > Testing > ------- > > `./gradlew build` > `./gradlew jmh -Pbenchmarks='UpdateStoreBenchmarks.*|TaskStoreBenchmarks.*'` > > > Thanks, > > Stephan Erb > >
