Skip to content

BRK421 Breakout Wed 18 Nov 10:15 PST

Operate a global AKS fleet across regions, edge, and sovereign clouds​

Schumann Ge, Simon Waight, Ameen Riaz Sherali

Summary

From the recording / Microsoft AI summary

AKS aims to provide one Kubernetes experience across Azure regions, sovereign clouds, multicloud, partner hardware and edge. JetBlue described moving to smaller purpose-built clusters to fix IP exhaustion.

Session description

Microsoft's catalogue description

As Kubernetes deployments expand across regions, public cloud, virtualized and bare-metal environments, and sovereign clouds, operational complexity grows. Learn how Azure Kubernetes Service Fleet Manager and AKS Flex Nodes help you standardize operations, enforce compliance, and simplify day-to-day management across distributed environments. Walk away with practical strategies to keep large-scale Kubernetes estates secure, consistent, and resilient as they grow.

Demo timestamps

Microsoft's player does not support links to a specific time, so open the recording and skip to the time shown.

  1. Skip to 0:19:00

    Onboarding clusters to Fleet Manager

    Adds clusters to an existing Fleet Manager in the portal (CLI and Terraform also work) and edits labels used for targeting. Fleet Manager supports up to 1,000 clusters of any type, including Arc-enabled Kubernetes.

  2. Skip to 0:23:56

    Staged upgrades across the fleet

    Builds an update strategy so non-production clusters get Kubernetes and node image upgrades before production, with concurrency limits and failure tolerance. The fleet picks the most recent consistent node image.

  3. Skip to 0:32:49

    GitOps rollout with Flux

    Creates a managed namespace on the fleet hub, distributes Flux RBAC to member clusters, and rolls out a Helm application via Flux. Nothing reaches production until the test stage is approved.

  4. Skip to 0:41:31

    Flex Nodes adding an on-premises GPU server (public preview)

    In a cluster with nodes in Azure and on-premises, an Arc-enabled GPU server joins as a managed worker via a flex node pool created in the portal. An AI video analysis app is deployed onto it.

Watch on-demand (opens in a new tab)

← All sessions