This document explains how to run Arm workloads on Google Kubernetes Engine (GKE).You can run Arm workloads in GKEAutopilot clusters using thePerformance orScale-Outcomputeclasses,or in GKE Standardclusters using theC4A,N4A, orTauT2A machine series.

You can run single-architecture Arm images or multi-architecture (multi-arch)images compatible with both x86 and Arm processors. To learn about the benefitsof Arm, seeArm VMs on Compute.

See the following for more information about choosing workloads to deploy on Armand preparing those workloads for deployment:

• Choosing workloads to run on Arm: Consider the benefits of the followingmachine types when choosing workloads to run on Arm. For more informationabout what types of workloads work well with each of thesemachine series, see the table inGeneral-purpose machine family forCompute Engine:

  • C4A nodes provide Arm-based compute which achieves consistently highperformance for your most performance-sensitive Arm-based workloads.
  • N4A nodes provide Arm-based compute that balances price and performance.
  • T2A nodes are appropriate for more-flexible workloads, or workloads whichrely on horizontal scale-out.

• Deploying across architectures: With GKE, you can usemulti-arch images to deploy one image manifest across nodes with differentarchitectures, including Arm.

  • To ensure that your container image is Arm-compatible and can run onyour targeted architectures, seeBuild multi-architecture images forArm workloads.
  • To follow a tutorial for using multi-arch images to deploy acrossarchitectures, seeMigrate x86 application on GKE tomulti-arch withArm.

• Preparing Arm workloads for deployment: Once you have an Arm-compatibleimage, usenodeaffinityrules andnodeselectorsto make sure your workload is scheduled to nodes with a compatiblearchitecture type.

  • Autopilot clusters: seeDeploy Autopilotworkloads on Armarchitecture.
  • Standard clusters: seePrepare an Arm workload fordeployment.

Requirements and limitations

• Arm nodes are available in Google Cloud locations that support Armarchitecture. For details, seeAvailable regions andzones.
• Config Connector andConfig Controllerare not supported on clusters with Arm node pools.

• See the following requirements and limitations for C4A:

  • To create a cluster with C4A nodes that usesAutopilotmode,clusterautoscaling,ornodeauto-provisioning,you must use the following versions or later:

    • 1.28.15-gke.1344000
    • 1.29.11-gke.1012000
    • 1.30.7-gke.1136000
    • 1.31.3-gke.1056000

  • To create a Standard cluster with C4A nodes, you must use one of thefollowing versions or later:

    • 1.28.13-gke.1024000
    • 1.29.8-gke.1057000
    • 1.30.4-gke.1213000

  • You can useLocalSSDs with C4Anodes with the following versions or later:

    • 1.29.15-gke.1325000
    • 1.30.12-gke.1033000
    • 1.31.8-gke.1045000
    • 1.32.1-gke.1357000

  • GKE doesn't support the following features with C4A nodes:

    • Confidential GKE Nodes
    • Compact placement
    • Simultaneous multi-threading (SMT)
    • Persistent disks(useHyperdiskinstead, seeSupported disk types forC4A)
    • Nested virtualization
    • GPUs

• See the following requirements and limitations for N4A:

  • To create a cluster with N4A nodes that usesAutopilotmode, use GKE version 1.34.1-gke.3403001 or later.

  • GKE doesn't support the following features with N4A nodes:

    • Local SSDs
    • Confidential GKE Nodes
    • GPUs
    • Compact placement
    • Simultaneous multi-threading (SMT)
    • Persistent disks(useHyperdiskinstead, seeSupported disk types forN4A)
    • Nested virtualization
    • 1 GBhugepages(only 2 MB hugepages supported)

• See the following requirements and limitations for T2A:

  • GKE doesn't support the following features with T2Anodes:

    • Confidential GKE Nodes
    • GPUs
    • GKE Windows
    • Local SSDs
    • Policy Controller,Config Sync, andConfig Controller

What's next

• Create clusters and node pools with Arm nodes
• Build multi-architecture images for Arm workloads
• Prepare an Arm workload for deployment
• Migrate x86 application on GKE to multi-arch with Arm