9 Things You Should Do When Starting to Use SLOs
summary
Nine practical steps: pick user-facing SLIs, avoid pre-aggregation, and measure success rates not just errors; use raw event-level data so you can spot surprises like a daily cache refresh that creates a 30–45 minute latency spike. Back objectives with history — Fortress Bank sets an acknowledgement SLA of 99.9% within 2,000 ms (p50 ≈100 ms, p95 ≈1,500 ms, p99 ≈1,700 ms). Automate defenses, add annotations/badges, document SLOs in Markdown, and adopt standards like OpenTelemetry OpenSLO SLO DLC.
AI-written summary · about
chapters (8)
- 0:00SLOs introduction
- 3:15Why SLOs are hard
- 6:25Measuring user journeys
- 12:12Setting reliability goals
- 17:27SLOs to insights
- 23:28When systems fail
- 28:53SLO best practices
- 30:39Q&A
related talks
37mReal World SLOs and SLIs: A Deep Dive
38mImplementing SRE practices: SLI/SLO deep dive
20m9 SLIs; Oh My!
23mGoing from 30 to 30 Million SLOs
41mSLOs that Delight: Master Multi-Service SLO Calculations
43mOur Experience Tracking and Driving SLO Adoption at Goldman Sachs
34mAvoiding Goodhart's Law - Use SLO's as Tools Not Cudgels
33mHow and why we lowered our SLO