Blaming in a Blameless World
summary
Lyft demonstrates that Pagerduty on-call data is far more current than a service catalog for assigning responsibility, and that standardized Envoy metrics let teams auto-generate error-rate SLOs (failures ÷ total requests) with a strict time window. Mapping the on-call email up the org chart to a director creates director-level rollups; the program takes 2 years to full coverage, yields ~2/3 of SLOs customized by developers, and correlates with fewer hard outages and faster incident resolution.
AI-written summary · about
chapters (12)
- 0:00Blameless incident response intro
- 1:29SLO problems at Lyft
- 3:11Understanding SLOs and SLIs
- 5:39Envoy metrics for SLOs
- 8:27Setting SLO thresholds
- 10:10Tracking SLO violations
- 12:19Finding the right person
- 15:06Directors and SLOs
- 17:10Locating officers in org
- 19:17Lyft’s SLO journey
- 20:44Future improvements and maintenance of SLOs
- 21:47SLOs and business performance
related talks
30mReliability Engineering for Humans
34mBuilding a Culture of Reliability
23mMotivational Leadership Driving Change Through Empowered Eng
44mYou build it, you run it
50mYou build it, you run it (why developers should also be on call)
31mHow to Build a Healthy On-call Culture
16mHuman Centered On Call Practices
28mFixing On-Call When Nobody Thinks It's (Too) Broken