Skip to main content
Cloud & AI Hub
Browse
Glossary AI Directory Playgrounds Models Prompts Explainers Strategy Matrix Benchmark Decoder

Kubernetes

An open-source platform for orchestrating containerized workloads at scale.

What is Kubernetes?

Kubernetes (often abbreviated K8s) automates deploying, scaling, and operating containers across a cluster of machines. It handles scheduling, health checks, rolling updates, and service discovery.

Key objects

ObjectRole
PodSmallest deployable unit; one or more containers
DeploymentDeclarative desired state for replicated pods
ServiceStable network endpoint for pods
IngressHTTP routing into the cluster
ConfigMap / SecretConfiguration and sensitive data

Why teams use Kubernetes

  • Self-healing β€” restart failed containers, replace unhealthy nodes
  • Horizontal scaling β€” add replicas based on CPU or custom metrics
  • Portability β€” run on EKS, GKE, AKS, or on-premises
  • Ecosystem β€” Helm, operators, service meshes, GitOps tools

Managed vs self-hosted

Managed Kubernetes (EKS, GKE, AKS) runs the control plane for you. Self-hosted gives maximum control but operational burden for etcd, upgrades, and security patches.

When Kubernetes may be overkill

A single monolith on one VM or a few containers on App Engine / Cloud Run may be simpler. Kubernetes shines when you have many services, frequent deploys, and team scale.

Cost reality

The cluster is rarely the big cost β€” the people are. Budget for real platform engineering time: upgrades every few months, ingress/network debugging, and RBAC hygiene. On the infra side, the silent waste is overprovisioned requests: pods reserving 4Γ— the CPU they use means paying for nodes that sit idle. Right-sizing requests/limits is routinely a 30–50% cluster cost cut.

What people get wrong

  • Skipping resource requests and limits. Unset requests break scheduling and autoscaling; unset limits let one pod starve the node.
  • Treating it as β€œset and forget.” Kubernetes versions EOL quickly; clusters more than a few versions behind become risky, all-at-once migration projects.
  • Running stateful databases on it too early. Managed databases exist so you don’t have to learn StatefulSet failure modes in production.

Historical figures and technical concepts for informational purposes only. Not technical, professional, legal, or financial advice. Sources: Official Documentation.