Fast, Safe, and Continuous -- Picking all 3 at Netflix
summary
Netflix turns a biweekly API deploy cadence into nightly canaries and daily deploys, about 14× faster deploys, by using Spinnaker pipelines, Kayenta automated canary analysis, and standardized build/test artifacts. Teams move smaller code sets, use single-action rollbacks, and record all production changes so they can recover locally. Netflix builds Chaos Kong regional evacuation across three AWS regions to shift traffic instantly, practices managed delivery to decouple app and infra, and sustains thousands of daily production changes without customer-visible impact.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:53Breaking the value triangle
- 6:10Fast: Developer velocity
- 9:48Microservices and distributed monolith
- 13:02Spinnaker for faster deployments
- 14:29Safe: Reliability tactics
- 18:43Recovery greater than prevention
- 20:50Regional evacuation with Chaos Kong
- 23:44Distributed reliability at Netflix
- 24:26Continuous deployments challenge
- 26:31Managed delivery for consistency
- 28:18Picking all three at Netflix
- 29:00Closing remarks
related talks
58mBuilding a Culture of Continuous Delivery
29mHow Netflix Thinks About DevOps
31mHow Netflix Thinks About DevOps
27mBehind the Scenes: Running Critical Services @ Netflix
36mSpinnaker - Land of a 1000 Builds
52mContinuous Deployment to the Cloud using Spinnaker
50mFast Delivery
53mContinuous Deployment with Spinnaker — Automate All Your Deployments!