What is Kubernetes (K8s)?

How Kubernetes Works

Kubernetes runs as two logical layers: the control plane and the worker nodes. The control plane hosts the API server, the single entry point for all cluster operations. Administrators and CI/CD pipelines submit YAML manifests declaring desired state: replica counts, resource requirements, and node constraints. The API server persists that state to etcd, a distributed key-value store that serves as the cluster's source of truth.

The scheduler reads pending pod specifications from etcd and assigns each pod to a worker node based on available CPU, memory, and any affinity or taint rules. Once assigned, the kubelet on that node instructs the container runtime (containerd or CRI-O) to pull the image and start the container.

Controllers run continuously in the controller-manager process, comparing actual cluster state against desired state and giving teams continuous monitoring of running workloads. If a pod crashes, the ReplicaSet controller schedules a replacement. If a Deployment receives an updated image tag, the rolling update controller replaces pods incrementally to maintain availability. This reconciliation loop is how Kubernetes maintains declared state without manual intervention.

Why Kubernetes Matters

Running containers at scale without orchestration means manually tracking which host each container runs on, writing custom restart logic, and managing storage mounts by hand. That overhead compounds quickly as container counts grow.

Kubernetes eliminates that overhead through automated scheduling, self-healing, and declarative configuration. For platform and DevOps teams, it provides a consistent cloud infrastructure management model whether workloads run on AWS, Azure, GCP, or on-premises hardware. According to the CNCF Annual Cloud Native Survey, January 2026, 82% of container users now run Kubernetes in production, up from 66% in 2023. K8s proficiency has become a baseline infrastructure skill for platform and DevOps roles.

Kubernetes vs. Docker

Kubernetes and Docker operate at different layers: Docker produces the container images and runs them on a single host, while Kubernetes schedules and manages those containers across a distributed cluster at scale.

Common Misconceptions

Kubernetes is not a replacement for Docker. Kubernetes requires a container runtime (such as containerd or CRI-O) to execute containers. It handles orchestration above the runtime layer, not image building or local execution.

Kubernetes does not only run in public cloud environments. It runs on bare-metal servers, on-premises virtual machines, and across hybrid infrastructure. Managed services like EKS, GKE, and AKS reduce operational overhead but are not required for a functional cluster.