Dressing to the 9s
summary
Claims concrete steps: switch focus from reliability to resiliency, automate automated rollback and watch sev zeros recover in minutes, enforce defaults like backups and alerting on by default and a linter that rejects single-replica deploys, provide a golden path of copy‑paste examples, build a big red button to pause deploys and autoscaling during provider outages, migrate Kubernetes by creating a new cluster and backup‑restore to shift traffic in 5–20 minutes, measure fewer pages and higher customer satisfaction.
AI-written summary · about
chapters (17)
- 0:00Reliability myths
- 1:47Rule of nines fails
- 3:33Resiliency versus reliability
- 5:23Economic incentives for resiliency
- 6:48User happiness and pager duty
- 8:14Team collaboration and service
- 9:18Resilient systems with automation
- 10:44Human laziness for defaults
- 12:49Linters to enforce quality
- 14:15Systems designed for failure
- 15:40Big red button for maintenance
- 17:03Uniformity at scale
- 18:45Learning from incidents
- 20:10Removing humans from reliability
- 21:35Management sign-off and reliability
- 22:40Flexibility in system design
- 23:00Conclusion
related talks
40mStrategies for Zero Down Time, Frequent Deployments
16mResilient by Design: Architectures for Crisis Response & Recovery
25mAutomating Disaster Recovery: The Ultimate Reliability Challenge
51mDesigning for Failure: Build Resilient Services
37mZero-downtime Deployment Strategies (...) [ENG]
46mSix small steps towards continuous delivery
31mHow We Went From Two Major Outages to 99.98% Reliability in 6 Months
41mHave You Tried Turning It off and Turning It on Again?