Reducing Trauma in Production
summary
Defines SLOs as measurable SLIs (example: 99% SLO = 1% error budget), uses concrete math (100,000 events → 1,000 allowed failures), and ties chaos engineering to validation across environments. Shows tools and flows: use Gremlin in staging and ramp blast radius, telemetry plus tracing and synthetic monitors, Datadog detects instance downtime and PagerDuty fires alerts. Describes defensive coding, dependency testing, and running injected latency to find surprising critical paths.
AI-written summary · about
chapters (15)
- 0:00Reducing trauma introduction
- 1:17Why SLOs and chaos matter
- 3:05Downtime costs and SLA vs SLO
- 6:55SRE pyramid and post-incident
- 7:17What users truly care about
- 11:25Chaos to validate metrics
- 12:48SLO prerequisites for maturity
- 16:56SLO math and error budgets
- 18:43Buffer between SLA and SLO
- 20:51SLOs as team goals
- 22:15Chaos beyond production
- 23:17Upstream dependencies and SLAs
- 25:46SLOs in post-mortem and sprints
- 26:26Lifecycle of SLOs and user experience focus
- 27:48Resources and closing
related talks
05mReducing Trauma in Production with SLOs and Chaos Engineering
47mCultivating Production Excellence
25mHow to detangle and understand production
51mShrinking Production Incidents
58mCultivating Production Excellence
29mCultivating Production Excellence
35mFrom Monitoring to Observability: Left Shift your SLOs with Chaos
30mFrom Monitoring to Observability: Left Shift your SLOs with Chaos