Kubernetes Deconstructed: Understanding Kubernetes by Breaking It Down
summary
Starts from a real kubectl deploy of a namespace, Deployment, Pods, Service and Ingress and then traces where each object actually lands: the API server persists to etcd, kube-controller-manager creates ReplicaSets, service accounts and secrets, the scheduler picks nodes based on pod spec fields, and kube-proxy programs iptables entries to implement ClusterIP/NodePort load‑balancing. Kubelet talks to Docker (creates a tiny pause/infra container to share pod IPs), rkt behaves differently, ConfigMaps update live, the Downward API exposes metadata, and network policy needs a separate enforcement plugin. The demo deploys and serves pages successfully.
AI-written summary · about
chapters (13)
- 0:00Intro
- 2:35Deployment
- 4:36Basic User
- 6:53API Server
- 7:11CD Cluster
- 7:32Scheduler
- 10:59Master Components
- 12:47Pods
- 15:06Services
- 19:40Cloud Layer
- 20:09Components
- 24:16Linux
- 28:11Power User
related talks
23mBeyond Block Diagrams: Different Ways of Understanding K8s Architecture
37mKubernetes Wonderland: Adventures in Platform Building
36mUnderstanding Kubernetes Through Real-World Phenomena and Analogies
28mKubernetes Plane Aerodynamics: Breaking the Architectural...
32mKubernetes as a platform for platforms
15mFighting the Final Boss: Complex Multi-site Multi-cluster App Dep...
14mKeynote: Kubernetes: The Video Game
30mWho Let the Pods Out? Extending Kubernetes with Custom Controllers and CRDs