Azure Kubernetes Service (AKS) continues to simplify Kubernetes operations with innovations that reduce platform management overhead. With Managed System Node Pools (GA), Azure takes responsibility for managing and updating critical system components, allowing platform teams to focus on application workloads instead of cluster maintenance.
Taking automation even further, AKS Automatic delivers an intelligent, hands-off Kubernetes experience by automating cluster operations, scaling, upgrades, and optimization. In this session, we'll explore how these capabilities help organizations improve reliability, reduce operational complexity, and accelerate their cloud-native journey while spending less time managing infrastructure.
- Managed System Node Pools — how Azure owns lifecycle management of the system node pool, including OS patching, Kubernetes upgrades, and node replacements, without disrupting application workloads
- AKS Automatic — a new SKU that bundles Node Auto-Provisioning (Karpenter), Cilium CNI, Azure Policy, KEDA, and rapid upgrade channels into a single, opinionated cluster configuration
- Node Auto-Provisioning (NAP) — how Karpenter right-sizes nodes per pod request instead of scaling within a fixed VM SKU, reducing waste and accelerating scale-out
- Operational comparison — standard AKS cluster autoscaler vs AKS Automatic NAP: when to use each, and what you trade off
Hands-on lab portal: https://aks.piquantlabs.com
| Exercise | Topic | Duration |
|---|---|---|
| E1 | Node Pools & Workload Separation | 10 min |
| E2 | Managed System Pools & Upgrade Channels | 10 min |
| E3 | AKS Automatic: Create & Compare | 20 min |
| E4 | Node Auto-Provisioning & Intelligent Scaling | 15 min |
- Azure CLI (
az) — install guide kubectl— install guide- Access code from the session instructor
┌──────────────────────────┐ ┌──────────────────────────┐
│ System Pool │ │ User Pool │
│ CoreDNS · kube-proxy │ │ Your apps · workers │
│ metrics-server · etc. │ │ batch jobs · APIs │
│ │ │ │
│ Taint: CriticalAddons │ │ No taint — open to all │
│ Only=true: │ │ application workloads │
│ NoSchedule │ │ │
└──────────────────────────┘ └──────────────────────────┘
Azure manages You manage
| Channel | Behavior |
|---|---|
none |
Manual upgrades only |
patch |
Auto-patch within current minor version |
stable |
Auto-upgrade to N-1 minor version |
rapid |
Auto-upgrade to latest GA release (used by AKS Automatic) |
node-image |
OS image updates only, Kubernetes version unchanged |
| Cluster Autoscaler | NAP (Karpenter) | |
|---|---|---|
| Node SKU | Fixed per node pool | Right-sized per pod request |
| Scale-down | Removes empty nodes only | Consolidates underutilised nodes |
| Configuration | Node pool per VM size | Single NodePool CRD |
| Available in | Standard AKS | AKS Automatic only |