Examples to explain this:
p50 = 100ms
→ Half your users get responses under 100ms
p99 = 500ms
→ 99 out of 100 requests finish under 500ms
→ 1 out of 100 takes longer
p99.9 = 2s
→ 999 out of 1000 under 2 sec
→ 1 out of 1000 is slower
MAX = 30s
→ Your worst case scenario
Ben Dicken@BenjDicken
Numbers every engineer should know: - p50: Median request latency - p99: 1% of requests are slower / 99% are faster - p99.9: One in a thousand requests are slower - MAX: Slowest requests p50 is useful for assessing median, across-the-board performance. p99(.9) for studying long-tail issues. MAX for investigating the longest running / analytics requests. This goes for every component of a distributed system, not just databases.
English
