Posted On: Jan 5, 2022
Customers often have applications with different compute architecture requirements. For example, customers may want to run machine learning workloads on AWS Graviton ARM-based instances that offer compelling price-performance benefits, while running daily data transformation jobs on non-Graviton-based instances. Previously, Amazon EMR on EKS users could only customize container images of non-Graviton-based EC2 instances. With this release, users can customize images for both Graviton-based and non-Graviton-based EC2 instances. Building, deploying, and using container images for different architectures from the same image repository in Amazon EMR on EKS is now more simple.
To learn more about how to build and use a custom image, please visit our documentation. Custom image functionality for Graviton instances is supported for all supported Amazon EMR on EKS releases and is available in all regions where Amazon EMR on EKS is currently available.