Managing und Running Multiple Kubernetes Clusters
summary
Managed hosting runs hundreds of Kubernetes clusters with the control plane deployed as per-cluster containers inside namespaces, exposing an OpenVPN per-cluster endpoint and separate API ports. Worker nodes come from OpenStack, AWS or DigitalOcean and join via the machine deployments API. Etcd runs as a statefulset etcd with persistent volumes and automated external etcd snapshots; rolling API-server upgrades are supported, Prometheus monitors cluster health, and end-to-end upgrade tests take about 20–25 minutes.
AI-written summary · about
chapters (8)
- 0:00Introduction and speakers
- 1:45Challenges of managing clusters
- 7:05Managed Kubernetes service vision
- 13:13Cluster setup and architecture
- 19:22Live demo of cluster creation
- 25:41Lessons learned from implementation
- 28:29Future roadmap and improvements
- 32:27Q&A session
related talks
37mManaging and Running Multiple Kubernetes Clusters in Hybrid Setups
01h 01mDeploying Kubernetes, a Container Cluster Manager
37mKubernetesception: How to run a 1000 control planes
30mMission Critical Kubernetes
30mScaling Kubernetes to Thousands of Nodes Across Multiple Clusters, Calmly
03mFrom 5 to 1,300+ Clusters: Declarative Scaling for Private Kubernetes
53mManaging Containers at Scale with CoreOS and Kubernetes
31mDeclarative Underlays: Scaling Purpose-Built Infrastructure(Clusters)...