Look at Your Data
summary
Averages hide crucial detail: a single 295 ms point can mask two distinct populations (author simulates adding 300 ms to half the hits) or a tiny 10% shoulder of slow requests. Use histograms and a vertical heatmap of per-minute distributions, and when that points to outliers drill into raw logs — printed sheets revealed a small set of test machines that added redirects, a config change removed the redirects and eliminated the worst-performing hits.
AI-written summary · about
chapters (7)
- 0:00The limitation of averages
- 1:49Data loss in visualization
- 4:15Understanding via histograms
- 7:03Detecting anomalies
- 10:01Advanced data visualization
- 11:45The value of raw logs
- 14:12Solving complex issues
related talks
37mMonitoring the Health of Your App
37mLies, Damn Lies, and Metrics
10mTurning data into leverage
39mThe Many Ways Your Monitoring Is Lying to You
41mKnow your data: The stats behind your alerts
38mKnow your data: The stats behind your alerts
27mLatency SLOs Done Right
45mUnderstanding Microservice Performance