$linuxjunkies
>

etcd

also: etcd3

A distributed, strongly consistent key-value store used for shared configuration and service discovery in clusters, most commonly by Kubernetes to store all cluster state.

etcd is a lightweight, open-source database that stores data as key-value pairs and replicates them across multiple machines using the Raft consensus algorithm. This ensures all nodes agree on the data state even if some fail, making it reliable for critical cluster metadata.

In Kubernetes, etcd is the backing store for the entire cluster—it holds pod definitions, service information, secrets, and configuration state. Every kubectl command that reads or writes cluster data actually reads from or writes to etcd on the control plane.

Example: when you run kubectl create deployment myapp --image=nginx, Kubernetes writes the deployment object to etcd. If a control plane node crashes, etcd's replication ensures that data persists and other nodes can continue serving the cluster API.

etcd is language-agnostic and useful beyond Kubernetes for any distributed system needing reliable, consistent configuration management across a fleet of machines.

Related terms