Choosing a Managed Provider
The notes app works on your laptop. For real users you need a cluster that stays up when you close the lid. Running a control plane yourself is serious work, so most teams rent one.
The rule
The provider runs the brain; you run what the brain runs.
In a managed Kubernetes service, the provider operates the control plane — API server, etcd, scheduler, controllers — including upgrades, backups and high availability. You still own your workloads: manifests, images, configuration and, usually, some choices about the worker nodes.
The big three:
- Amazon EKS (Elastic Kubernetes Service) on AWS.
- Google GKE (Google Kubernetes Engine) on Google Cloud. Its Autopilot mode manages the nodes too.
- Azure AKS (Azure Kubernetes Service) on Microsoft Azure.
Smaller clouds such as DigitalOcean, Linode (Akamai) and others offer simpler, often cheaper clusters.
Creating one is usually a single command or a few clicks, for example:
gcloud container clusters create-auto notes-prod --region asia-south1After that, the same kubectl and the same manifests you used locally work unchanged.
What to compare:
| Question | Why it matters |
|---|---|
| Where do our database, storage and users already live? | Staying in one cloud cuts latency and data-transfer bills |
| Is there a fee for the control plane? | Some providers charge per cluster-hour; others have a free tier |
| Who manages the nodes? | Node management modes remove patching work, at some cost in control |
| Which regions are available? | Pick one close to your users |
The shortcut
Follow your data. Choose the provider where your data and other services already are, then check cost and regions. Features across the big three are close; distance and bills are not.
Find the data
The team already stores backups in Google Cloud Storage and users are in India.
The trap is believing "managed" means "no work". Your app's configuration, security settings, resource sizes and upgrades of your own images are still yours.
Drills
Answer fast — trust the rule.
In managed Kubernetes, who operates etcd and the API server?
Which of these is still your job on a managed cluster?
Which managed service belongs to Microsoft Azure?
A team keeps all its data on AWS. By the shortcut, where should its first cluster go?
Remember
- The provider runs the brain (control plane); you run what the brain runs.
- Follow your data, then compare cost, node management and regions.
- Trap: "managed" does not mean "no work" — your workloads are still yours.