I've been performing some benchmarking tasks on freshly installed CentOS machines with SELinux enabled and disabled. The tasks include copying files of various sizes, sending variable amounts of data through a network socket and process creation (forking and exiting) load.
The majority of my tests are reporting that with SELinux enabled these tasks are being completed faster than with SELinux in permissive mode. It was my understanding that SELinux would give more overhead thus, resulting in a longer time period for task completion. I originally assumed that these are edge cases, but reproducing the experiment has reproduced the result.
Can anyone advise?
Thank you for your time.