Kubernetes
An open-source platform that deploys, scales and keeps containerized applications running across a group of servers.
Pronunciation
What it is
Kubernetes is a container orchestrator: it takes applications packaged in containers (for example with Docker) and decides which server runs each one, how many copies exist, how they talk to each other and what to do when something fails.
Google open-sourced it in 2014, inspired by its internal Borg system. Today it is maintained by the Cloud Native Computing Foundation (CNCF), part of the Linux Foundation. The name comes from Greek and means “helmsman”. It is abbreviated K8s: the K, 8 letters and the s.
What it solves
Running one container on one server is easy. Running dozens of services, with several copies each, across several servers and without downtime is not. Kubernetes automates:
| Capability | What it does |
|---|---|
| Self-healing | If a container or a server fails, it starts another copy elsewhere |
| Scaling | Adds or removes copies according to load |
| Rolling updates | Updates a version gradually and allows rolling back |
| Discovery and load balancing | Each service gets a stable name and traffic is spread across its copies |
| Configuration and secrets | Keeps configuration separate from code |
Core concepts
- Cluster: the set of servers (nodes) Kubernetes manages.
- Pod: the smallest unit; one or more containers that run together.
- Deployment: describes how many copies of a pod should exist and which version.
- Service: a stable access point to a group of pods.
Kubernetes is declarative: you describe the state you want (“3 copies of the API, version 2.4”) and it works continuously to keep it.
Does your company need it?
✅ It makes sense when:
- There are several services (microservices) deployed independently.
- Load varies a lot and automatic scaling is needed.
- High availability and zero-downtime deployments are required.
❌ It is usually too much when:
- It is a single application or a small team: a managed container service or a platform as a service is simpler to operate.
In the cloud it is almost always used as a managed service (Amazon EKS, Azure AKS or Google GKE), where the provider runs the control plane.
Demo example: an online store runs its API with 3 copies. During a campaign, load triples and Kubernetes scales to 9 copies; when it ends, it goes back to 3.
Related terms
- Docker - The containers Kubernetes orchestrates
- Microservices - The architecture that benefits most
- DevOps - The practices needed to run it well
- CI/CD - How new versions reach the cluster
Official resources: