Kubernetes Architecture
Understanding Kubernetes architecture is crucial for working with it effectively. A Kubernetes cluster consists of a control plane and one or more worker nodes.
Cluster Overview
┌──────────────────────────────────────────────────────────┐
│ Control Plane │
│ ┌──────────┐ ┌───────────┐ ┌──────────┐ ┌────────┐ │
│ │ API │ │ Scheduler │ │Controller│ │ etcd │ │
│ │ Server │ │ │ │ Manager │ │ │ │
│ └────┬─────┘ └─────┬─────┘ └────┬─────┘ └───┬────┘ │
└───────┼──────────────┼──────────────┼────────────┼──────┘
│ │ │ │
┌───────┼──────────────┼──────────────┼────────────┼──────┐
│ │ Worker Nodes │ │ │
│ ┌────┴─────┐ ┌─────┴─────┐ ┌────┴─────┐ │
│ │ kube │ │ kube │ │ kube │ │
│ │ -proxy │ │ -proxy │ │ -proxy │ │
│ └────┬─────┘ └─────┬─────┘ └────┬─────┘ │
│ ┌────┴─────┐ ┌─────┴─────┐ ┌────┴─────┐ │
│ │Container │ │Container │ │Container │ │
│ │ Runtime │ │ Runtime │ │ Runtime │ │
│ └──────────┘ └───────────┘ └──────────┘ │
└──────────────────────────────────────────────────────────┘
Control Plane Components
API Server (kube-apiserver) — The front door to the cluster. All communication between components goes through the API server. kubectl talks to the API server.
etcd — A distributed key-value store that holds all cluster state. Think of it as the cluster's database. Without etcd, Kubernetes has no memory.
Scheduler (kube-scheduler) — Watches for newly created pods with no assigned node and selects a node for them to run on based on resource requirements and constraints.
Controller Manager (kube-controller-manager) — Runs controller loops that regulate the state of the system. For example, the ReplicaSet controller ensures the desired number of pods are always running.
Worker Node Components
kubelet — An agent running on each node that ensures containers are running in a pod. It communicates with the container runtime to start, stop, and monitor containers.
kube-proxy — Maintains network rules on nodes, enabling Services to route traffic to the correct pods.
Container Runtime — The software responsible for running containers. Common runtimes include containerd, CRI-O, and Docker (deprecated).
How a Deployment Works
1. You run: kubectl apply -f deployment.yaml
│
▼
2. API Server stores the desired state in etcd
│
▼
3. Controller Manager notices the new Deployment
and creates a ReplicaSet
│
▼
4. Scheduler assigns the ReplicaSet's Pods to nodes
│
▼
5. Kubelet on each node tells the container runtime
to start the containers
│
▼
6. Everything is now running! K8s continuously
reconciles desired vs actual state