Peter CzanikWhy "#syslog_ng can do 7 million EPS" was announced on #AprilFoolsDay? It is a completely valid result. However, it was measured not in a #production, but in a #laboratory environment.
https://peter.czanik.hu/posts/new-toy-1st-of-april-syslog-ng-performance-tests/
So, it's an absolutely stunning result, just impossible in production.
My new toy: April 1 syslog-ng performance tests
Almost 15 years ago, Balabit had a campaign, stating that syslog-ng could process 650k messages a second. Now I am happy to present 7 million EPS (events per second). Timing the announcement to April 1 is not a coincidence :-) While the 650k EPS measurement was true, it was misleading. This value was measured right after syslog-ng 3.2 introduced multi-threading, in lab environment, under optimal circumstances, using synthetic log messages. However, there was no fine print explaining this, just the statement that syslog-ng could process 650k EPS.