From Code to Customer
summary
Describes turning a twice-weekly manual process into true production continuous deployment across 15 teams (~60 people) using Kubernetes on AWS EKS, GitLab pipelines, deployment rotation to teach teams deployments, and GitLab feature flags (Unleash style) to decouple releases. Migrates flaky Selenium tests to Playwright, parallelizes tests, adds build nodes and caching, and drops manual steps. Result: pipeline lead time falls from ~80–90 minutes to ~10, peak 20 deployments/day, faster recovery and markedly improved DORA metrics and developer experience.
AI-written summary · about
chapters (21)
- 0:00Continuous deployments intro
- 6:04Real journey to CD
- 14:33Building vision and support
- 17:00Platform team for devX
- 18:47Deployment rotation intro
- 21:17Simplify automate deployments
- 24:49Decouple deployments releases
- 26:35Coaching teams tech
- 27:36Selenium to Playwright
- 29:21Refresh vision buy-in
- 31:48Overcoming pushbacks
- 34:36Deployment rotation concerns
- 36:41Friday deployments myths
- 38:28Results of continuous deployments
- 41:18DORA metrics with CD
- 43:45DevX and customer relations
- 45:53Key takeaways
- 47:19Start early with CD
- 48:01Overcommunicate sponsors
- 50:56Recommended books
- 52:21Closing remarks
related talks
30mTestOps: Transforming Chaos into Streamlined Story-Based Releases
01h 01mA team's transition to Continuous Delivery
24mReinvent your pipeline with GitLab, Kubernetes, and Amazon EKS
31mRelease Engineering as a Service - The Art of Integrating with K8s
07mLightning Talk: Continuous Delivery with Kubernetes: The Prequel
33mFrom 0 to DevOps in 80 days
28mClimbing the Mountain of CI/CD: A Tester’s Tale
53mTransforming testing methodologies: From manual to robust strategies