Container orchestration in production. When K8s is the right tool — and when it is too big.
Category · DevOps & Infra
What Kubernetes does.
Kubernetes (K8s) is an orchestrator: it distributes containers across a pool of machines, restarts ones that have failed, scales up and down with load, and rolls out new versions without downtime.
You describe the desired target state declaratively, and K8s continuously makes sure reality matches it. Self-healing here is the norm, not the exception.
When it's the right tool.
K8s earns its keep when many services run in parallel, load swings hard and failover across several nodes is required — typical of the platforms we keep running for clients around the clock.
We use it where operations justify it, and lean on managed variants (AKS, EKS, GKE) so we don't have to maintain the cluster underpinnings ourselves.
When it's too big.
For a single app or a handful of services, Kubernetes is overkill — the complexity eats up the benefit. A managed container service or Docker Compose is faster, cheaper and more maintainable then.
K8s only pays off once the operational burden of a manual setup outweighs the cluster's learning curve. Recognising that threshold is the real decision.
