At Adevinta we use AWS Graviton instances built on the ARM CPU architecture to reduce computation costs. These run alongside x86 instances depending on the processing task. But a combination of CPU architectures can create problems when auto-deploying instances – especially if a Kubernetes container is deployed to the wrong architecture.
With 5000 applications to consider, we needed an effective way to update more than 5000 deployment manifests automatically. Read our latest article to discover:
- How to publish container images for ARM.
- How to configure the Kubernetes node selector to use ARM nodes
- Optimising multi-architecture builds.
- Introducing Noe, the Open Source component that enables progressive and transparent switching between ARM and x86 architectures within our Kubernetes clusters.
Ready to learn more about Noe and how to use it in your own projects? Read the full article on the Adevinta tech blog.