pnowojski commented on pull request #18702: URL: https://github.com/apache/flink/pull/18702#issuecomment-1035032817
The initial version was causing a small performance regression. After a couple of optimisations there is no visible regression when the watermark alignment is not used. ``` "Benchmark","Mode","Threads","Samples","Score","Score Error (99.9%)","Unit","Param: sourceType" "org.apache.flink.benchmark.InputBenchmark.mapRebalanceMapSink","thrpt",1,30,9936.451429,255.258997,"ops/ms",LEGACY "org.apache.flink.benchmark.InputBenchmark.mapRebalanceMapSink","thrpt",1,30,14651.620943,383.589770,"ops/ms",F27_UNBOUNDED "org.apache.flink.benchmark.InputBenchmark.mapSink","thrpt",1,30,15669.405419,199.914977,"ops/ms",LEGACY "org.apache.flink.benchmark.InputBenchmark.mapSink","thrpt",1,30,30679.668113,265.650771,"ops/ms",F27_UNBOUNDED ``` However with watermark alignment enabled, most likely due to the `WatermarkTrackingOutput` wrapper, there is a small (under 3%) extra overhead visible in the `mapSink.F27_UNBOUNDED` benchmark. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@flink.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
