Metrics: numerical (CPU (Central Processing Unit), latency). Logs: text events. Traces: a request's journey across services.
Alert on symptoms users experience, not internal causes. Page humans only for things that need human action right now.
Lesson 4 of 412 min
Observability — metrics, logs, traces
Three pillars of observability. Dashboards. Alerts that page humans only when needed.
Key takeaways
- ▸Metrics + logs + traces = observability
- ▸Alert on user-facing symptoms
- ▸Pages should be rare and actionable