schlosna commented on PR #12782: URL: https://github.com/apache/kafka/pull/12782#issuecomment-2278960075
Added a JMH benchmark to exercise `org.apache.kafka.common.requests.MetadataResponse#toPartitionInfo` that calls `org.apache.kafka.common.requests.MetadataResponse#convertToNodeArray` 3 times per invocation -- for replica IDs, in-sync replica IDs, and offline replica IDs. This shows between ~20% and ~40% speedup varying by the number of nodes. VM version: JDK 17.0.11, OpenJDK 64-Bit Server VM, 17.0.11+9-LTS Before ``` Benchmark (nodes) Mode Cnt Score Error Units MetadataResponseBenchmark.toPartitionInfo 10 avgt 5 183.330 ± 75.806 ns/op MetadataResponseBenchmark.toPartitionInfo 500 avgt 5 5418.979 ± 154.885 ns/op MetadataResponseBenchmark.toPartitionInfo 1000 avgt 5 12207.096 ± 647.602 ns/op ``` After ``` Benchmark (nodes) Mode Cnt Score Error Units MetadataResponseBenchmark.toPartitionInfo 10 avgt 5 114.204 ± 3.927 ns/op MetadataResponseBenchmark.toPartitionInfo 500 avgt 5 3871.721 ± 176.063 ns/op MetadataResponseBenchmark.toPartitionInfo 1000 avgt 5 9625.442 ± 581.124 ns/op ``` -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
