Actionable Observability
summary
Argues that observability plus monitoring by SLOs makes alerts actionable, with New York Times practice: pick prioritized user journeys, define composite SLIs (availability + load time), set targets, and use 250 ms leading indicator and 300 ms secondary threshold as an example. Chooses burn-rate alerts over consumption alerts for faster reaction, uses rolling error budgets (7/30/90 days), ties alerts to automated rollbacks, and cites ~73% of incidents originating from changes.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 2:52Defining observability and incident management
- 8:07Monitoring vs observability
- 10:56Metric categories
- 16:12Configuring monitors
- 19:43Prioritizing metrics
- 23:14Service level objectives
- 27:24Reliability vs availability
- 31:37Defining SLOs
- 34:06Debugging production
- 36:54Addressing system instability with automation
- 38:04Automation as an enabler not a replacement
related talks
36mThe road to world-class observability
29mObservability; a gentle introduction
41mA Theory and Practice of Alerting with Service Level Objectives
13mObservability that matters
33mTackling Alert Fatigue with SLOs
50mApplied Alerting Philosophy
36mHow to Include Latency in SLO-based Alerting
43mSLO Burn—Reducing Alert Fatigue and Maintenance Cost in Systems of Any Size