Services & Networking

Every Pod gets its own routable IP address; all Pods can communicate with all other Pods on any node without NAT; and agents on a node (like the kubelet) can reach all Pods on that node. This flat network model is provided by the CNI plugin.

Pod IPs are ephemeral — they change when Pods are recreated, rescheduled, or scaled. A Service provides a stable virtual IP and DNS name that front a dynamic set of Pods, so clients do not have to track individual Pod IPs.

The CNI (Container Network Interface) plugin wires Pod networking, assigns Pod IPs from the Pod CIDR, and enables Pod-to-Pod routing across nodes. Examples include Calico, Cilium, and Flannel.

Until a Pod-network add-on (CNI) is applied, the kubelet cannot configure Pod networking, so nodes report NotReady. Applying a CNI manifest (kubectl apply -f <cni>.yaml) brings nodes to Ready.

An overlay network encapsulates Pod traffic (e.g. VXLAN) so it tunnels between nodes regardless of the underlying network. A routed/native approach programs the node and underlying network to route Pod CIDRs directly without encapsulation, typically with lower overhead.