2 ms·
Often you're after tail latency, not throughput. It's pretty normal to target e.g. your web backend tier to not exceed 20% CPU load, and be ready to instantly
by nine_k 5y ago
Often you're after tail latency, not throughput.
It's pretty normal to target e.g. your web backend tier to not exceed 20% CPU load, and be ready to instantly react to incoming I/O.