The Etcd datastore
We explained that kube-apiserver is a stateless API that can be scaled horizontally. However, it is necessary for kube-apiserver to store the state of the cluster somewhere, such as the number of containers created, on which machines, the names of the pods, which Docker images they use, and more. To achieve that, it uses the Etcd database.
The role of the Etcd datastore
Etcd is part of the control plane. The kube-apiserver component relies on a distributed NoSQL database called Etcd. Strictly speaking, Etcd is not a component of the Kubernetes project. As you might have gathered, Etcd is not named according to the same nomenclature as the other components (kube*). This is because Etcd is not actually a Kubernetes project but a project completely independent of Kubernetes.
Instead of using a full-featured relational database such as MySQL or PostgreSQL, Kubernetes relies on this NoSQL distributed datastore called Etcd to store its state persistently. Etcd...
