franz1981 edited a comment on issue #2514: ARTEMIS-2236 Address Latency Impact caused by ARTEMIS-1451 URL: https://github.com/apache/activemq-artemis/pull/2514#issuecomment-456711246 Suggestion: i would use something more precise (with an higher spectrum of latencies) if possible to make things comparable: https://github.com/franz1981/artemis-load-generator Change the core version you need on the pom and run it with: ```java -jar destination-bench.jar --bytes 100 --protocol artemis --url tcp://localhost:61616 ---out /tmp/latencies.txt --name q --iterations 100000 --runs 1 --warmup 100000 --wait 2``` It will give you the latencies over an all out throughout: then you need to choose a sustainable throughput eg 100000 and re-run the same test adding `--target 100000`, then you will have comparable latencies. The ideal would be to use a profiler (async-profiler) to measure the samples taken by awaking the idle threads: the point is, if you have too much threads sitting idle into the thread-pool you'll get a lower utilization. Just sizing the max thread pool with a reasonable value would be enough eg 1/2 of the real core count, given that netty is already stealing the rest (probably more).
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
