Migrate x86 workloads to Arm on GKE with Google Axion
Introduction
Explore the benefits of migrating microservices to Arm on GKE
Set up your environment
Create build-ready Dockerfiles for both architectures
Build and deploy multi-architecture images on GKE
Prepare manifests and deploy on GKE
Automate builds and rollout with Cloud Build and Skaffold
Next Steps
Migrate x86 workloads to Arm on GKE with Google Axion
Introduction
Explore the benefits of migrating microservices to Arm on GKE
Set up your environment
Create build-ready Dockerfiles for both architectures
Build and deploy multi-architecture images on GKE
Prepare manifests and deploy on GKE
Automate builds and rollout with Cloud Build and Skaffold
Next Steps
Review frequently asked questions
These FAQs were drafted with an approved AI-assisted workflow and reviewed by Arm contributors before publication. Human technical review remains part of the process so the final page reflects engineering rigor, accuracy, and Arm editorial standards.
For the native GKE Buildx workflow, list your GKE nodes and check the architecture labels to confirm both amd64 and arm64 node pools are present and ready. You can then run each BuildKit pod on the matching architecture. If you choose Cloud Build, use its own runner with QEMU instead.
Four services require small changes:
emailservice, recommendationservice, loadgenerator, and cartservice. The edits ensure that the correct compiler headers and runtime libraries are included for each architecture.Use the GKE-backed Buildx workflow. It runs separate BuildKit pods on the amd64 and arm64 node pools, so each platform builds natively. The alternative Cloud Build workflow enables QEMU in its runner for cross-architecture builds.
Use Kustomize overlays that select nodes by architecture and reference your Artifact Registry images. Apply the overlay for amd64 first, then apply the arm64 overlay to migrate the workload to Axion-based nodes.
GKE uses VPC-native (IP aliasing) and requires two secondary ranges on the subnet: one for Pods and one for Services. On the default VPC, these ranges are created automatically. For custom networks, verify both ranges exist before creating the cluster.