How to detangle and understand production
summary
Argues for concrete practices: define 99.9% SLO over 30 days on business events, measure SLIs like HTTP 200 under 300ms, and use error budget to decide ship versus fix. Build observability that lets you form and verify hypotheses without redeploying, instrument traces to find a bad machine or drain a data center, run blameless postmortems, and shrink blast radius instead of chasing frequency. Prioritize fixes by expected frequency × impact and eliminate systemic slowdowns in debugging.
AI-written summary · about
chapters (9)
- 0:00Start
- 0:21The Problem with Complex Systems
- 6:40The Art of Production Excellence
- 7:05Roadmap for Production Excellence
- 8:32Eliminate the Unnecessary Complexity
- 9:11Service Level Indicators
- 12:08Focus on User Outcomes
- 15:34Observability
- 19:33How Do We Reward Curiosity and Teamwork
related talks
58mCultivating Production Excellence
47mCultivating Production Excellence
29mReducing Trauma in Production
39mShifting Left on Production Excellence with Observability
29mShrinking the Impact of Production Incidents
29mCultivating Production Excellence
51mShrinking Production Incidents
38mImplementing SRE practices: SLI/SLO deep dive