What is Operator?

A Kubernetes Operator is a software extension that uses custom resources and controllers to manage complex stateful applications on Kubernetes. It encapsulates the operational knowledge of a human site reliability engineer (SRE) into code, automating tasks such as deployment, scaling, backup, recovery, and upgrades. The Operator pattern was first introduced by CoreOS in 2016 and has since become a standard approach for managing applications like databases, message queues, and monitoring systems on Kubernetes.

The Operator works by defining one or more custom resource definitions (CRDs) that represent the application's desired state. A controller, typically written in Go, runs inside the Kubernetes cluster and watches for changes to these CRDs. When a user creates or updates a custom resource, the controller reconciles the actual state of the application with the desired state, executing the necessary steps to achieve and maintain that state. This reconciliation loop is the core mechanism, allowing the Operator to handle failures, scale events, and configuration changes automatically.

Operators sit above the core Kubernetes controllers (like Deployments and StatefulSets) in the stack. While built-in controllers handle generic pod lifecycle, Operators provide application-specific logic. The Operator Lifecycle Manager (OLM) and OperatorHub.io provide a registry and management framework for distributing Operators. The community maintains a large catalog of Operators for popular software such as etcd, Prometheus, PostgreSQL, and Elasticsearch. The pattern is formalized in the Operator Framework, an open source toolkit from Red Hat.