LinkerD and Kubernetes
summary
A Kubernetes-based setup (Tectonic/CoreOS distro) runs prod, preprod and dev by namespace so a full preprod clone can spin up in about one minute. Helm templates manage deployments, Prometheus collects metrics, and Linkerd is used for dynamic traffic shifting for canary releases (percentage routing like 70/30) with rapid rollbacks in 2–3 seconds; a small Go daemon automates routing and orchestration. They push ~80–100 images per day and do not yet perform image security scans.
AI-written summary · about
chapters (10)
- 0:00Technical transformation
- 3:08Microservices goals
- 10:13Traffic deployments
- 17:52Dynamic routing & env
- 24:13Continuous deployment
- 28:47Live traffic demo
- 33:18Linkerd performance
- 37:16Open source partnerships
- 42:19Production management
- 45:57Q&A
related talks
22mAnatomy of a production grade Kubernetes cluster
37mManaging Containers at Scale with CoreOS and Kubernetes
53mManaging Containers at Scale with CoreOS and Kubernetes
42mDrupal, Kubernetes, Production
46mKubernetes Native Continuous Deployment with FluxCD, Flagger, and Linkerd
36mHow to make Kubernetes rhyme with prod readiness - Tiffany Jernigan and Matthias Haeussler
39mService Mesh: There and Back Again
46mKubernetes native CI/CD with Prow