Course contentsShow
← Cloud and Infrastructure
Lesson 1491 of 2,98930. Managed Kubernetes ServicesFree lesson

AKS Architecture and Control Plane Management

Understanding how AKS manages the control plane and the separation between managed and user components.

AKS Architecture and Control Plane Management

You'll learn: How Azure manages the Kubernetes control plane in AKS and what that means for you as a cluster operator.

What Makes AKS Different

In Azure Kubernetes Service (AKS), Microsoft handles the control plane for you—completely. Unlike self-managed Kubernetes where you install and maintain the master nodes yourself, AKS runs the API server, scheduler, controller manager, and etcd as a fully-managed service. You never see these components as billable resources.

The Managed/User Split

Think of AKS like living in a managed apartment building: Azure maintains the building's infrastructure (elevators, heating, water supply), while you furnish and manage your individual unit.

Azure manages:

  • API server (your cluster's "front desk")
  • etcd (the cluster's database)
  • Scheduler and controller manager
  • Control plane upgrades and patches
  • High availability of control plane components

You manage:

  • Worker nodes (where your workloads run)
  • Node pools and scaling
  • Application deployments
  • Network policies within your cluster

No Direct Control Plane Access

You cannot SSH into control plane nodes or modify their configuration directly. You interact with the cluster only through the Kubernetes API using kubectl or other standard tools. This abstraction simplifies operations but means you rely on Azure's SLA for control plane availability.

Cost Implications

The control plane itself is free in standard AKS (you pay only for worker nodes). Azure absorbs the infrastructure cost and management overhead, making it economically attractive compared to running your own masters.

Key Takeaway: AKS separates responsibilities cleanly—Azure runs and maintains the control plane invisibly, while you focus solely on deploying and managing your workloads on the worker nodes.