Average latency is up 50ms, but p99 is flat. How do you diagnose this?
Tests your grasp of latency metrics. A rising average with flat p99 means the *bulk* of requests (p50-p90) slowed, not the tail. Hypothesize a common bottleneck and segment data by endpoint/user to find it. Red flag: blaming new, slow outliers.
This tests your ability to reason from metrics to system behavior, specifically the difference between average and percentile latency. A rising average with a flat p99 indicates a widespread, moderate slowdown in the *body* of the distribution (e.g., p50-p75), not new extreme outliers. A great answer hypothesizes a cause affecting a common code path (e.g., a new auth check) and outlines a plan to segment latency data by endpoint, customer, or release to isolate the change.
Read the original → aerospike.com
- #analytics
- #metrics
- #debugging
- #latency
- #systems-design
Get five bites like this every day.
Tezvyn delivers a daily feed of 60-second tech bites with quizzes to lock in what you learn.