@jamey@toot.cat I'll never forget that one of my very first public technical talks on observability/performance monitoring of distributed systems was attended by Brendan Gregg, a legend of the field. He demonstrated that simply shouting at even well mounted hard drives could introduce significant variance in command service times.