What is Azure Kubernetes Service (AKS)? A plain-English guide
Azure Kubernetes Service (AKS) is Microsoft’s managed Kubernetes offering on Azure — a way to run containerised applications at scale without having to build and operate a Kubernetes cluster yourself. Kubernetes is the open-source system that has become the standard way to orchestrate containers: deciding where they run, restarting them when they fail, and scaling them up or down. Running that system yourself is a genuine operational undertaking, and AKS exists to remove most of it. Azure hosts and manages the Kubernetes control plane, leaving you to focus on the applications running inside the cluster rather than the cluster itself. This article covers what AKS is, when you would reach for it, and how it fits alongside its counterparts on other clouds.
A quick word on Kubernetes
Kubernetes is a container orchestration system: it takes a set of containers — packaged, portable units of an application — and decides which machines they run on, keeps the right number of copies running, replaces ones that crash, and routes traffic to healthy instances. We cover the underlying concept properly in our what is Kubernetes guide and in our Docker vs Kubernetes comparison; this article assumes that background and focuses on what Azure adds on top of it.
Kubernetes itself is complex to run well. A cluster has a "control plane" — the brain that makes scheduling and scaling decisions — and a set of worker nodes that actually run your containers. Operating the control plane reliably, patching it, and keeping it available is a specialist job that most teams would rather not take on themselves.
What "managed" means with AKS
AKS takes the control plane off your hands entirely: Azure provisions it, patches it, upgrades it and keeps it highly available, and you are not billed for that management layer as a separate resource to run yourself. You still provide and pay for the worker nodes — the virtual machines that your containers actually run on — but even there AKS reduces the operational load, handling much of the node provisioning, health monitoring and upgrade orchestration.
The result is that a team can get a production-ready Kubernetes cluster running in a fraction of the time it would take to build one from scratch, and can keep it running without a dedicated Kubernetes platform team. You still write Kubernetes manifests and use standard Kubernetes tooling — AKS is not a different flavour of Kubernetes, it is the same open-source system, managed.
When you would use AKS
AKS suits teams that have already committed to containers and want the flexibility, portability and ecosystem that Kubernetes provides — as opposed to a simpler, more opinionated container service that hides Kubernetes entirely. Typical scenarios include:
- Microservices applications made up of many independently deployed services that need consistent scaling and networking rules.
- Teams migrating an existing on-premises or another-cloud Kubernetes workload into Azure, since Kubernetes manifests are largely portable.
- Workloads that need fine-grained control over scheduling, networking or scaling behaviour that simpler platform-as-a-service options do not expose.
- Organisations standardising on Kubernetes as their common deployment target across multiple clouds or environments.
How it fits: AWS EKS and Google GKE
AKS is Azure’s counterpart to Amazon EKS (Elastic Kubernetes Service) on AWS and Google Kubernetes Engine (GKE) on Google Cloud — all three are managed Kubernetes services solving the same problem for their respective platforms. Because Kubernetes itself is an open, portable standard, the workload definitions you write for one are broadly transferable to the others, which is part of why Kubernetes is popular for teams avoiding lock-in to a single cloud.
Within Azure, AKS sits alongside simpler container options such as Azure Container Instances for lighter, single-container workloads. Choosing between them is a question of how much orchestration control you need versus how much operational simplicity you want.
Pricing model, in one line
AKS follows a consumption model: there is no separate charge for the managed control plane on standard tiers, and you pay for the underlying compute, storage and networking resources — the worker nodes — that your cluster actually uses, so cost scales with what you run rather than a fixed fee. Exact rates, tiers and any premium control-plane options are on Azure’s pricing pages.
Where AKS shows up in certification study
AKS appears in Azure Administrator Associate (AZ-104) study as one of the compute services you are expected to know how to deploy and manage, at a conceptual level rather than deep Kubernetes engineering. Because Kubernetes itself is cloud-agnostic, it is also core material on the Kubernetes and Cloud Native Associate (KCNA), which tests Kubernetes concepts directly rather than any one vendor’s managed offering. Our AZ-104 and KCNA study material covers both in full depth — this article is the plain-English starting point.
Original practice questions, timed mock exams and revision notes. No card, nothing to pay.