• kube-apiserver: central access point for all operations
• etcd: persistent state store
• kube-scheduler: assigns Pods to nodes
• controller-manager: maintains cluster state (replicas, jobs, nodes)
• cloud-controller-manager: cloud provider integration

• kubelet: ensures Pods run as instructed
• kube-proxy: networking and load balancing rules
• container runtime: runs containers

• Each Pod gets a unique IP
• Pods can reach each other without NAT
• Services provide stable virtual IPs
• CNIs implement the network model

• Source of truth for the entire cluster
• Stores cluster configuration, secret data, and runtime state

AI Home