Amazon EMR now supports using multiple custom AMIs when you mix AWS Graviton2-based instances with non-Graviton2 instances in a single EMR cluster
Amazon EMR now supports using multiple custom Amazon Machine Images (AMI) when you mix Arm-based AWS Graviton2-based instances with non-Graviton2 based instances in a single cluster. This capability allows you to diversify across more instance types when using custom AMIs, helping improve your access to EC2 capacity for large clusters. Prior to this release, you could still mix multiple instance types within a cluster, but could not do so when using Custom AMIs. Custom AMIs enable you to preload additional software and libraries required by your applications, customize cluster and node configurations, and encrypt the EBS root device volumes of EC2 instances in your cluster.
Amazon EMR now supports using multiple custom Amazon Machine Images (AMI) when you mix Arm-based AWS Graviton2-based instances with non-Graviton2 based instances in a single cluster. This capability allows you to diversify across more instance types when using custom AMIs, helping improve your access to EC2 capacity for large clusters. Prior to this release, you could still mix multiple instance types within a cluster, but could not do so when using Custom AMIs. Custom AMIs enable you to preload additional software and libraries required by your applications, customize cluster and node configurations, and encrypt the EBS root device volumes of EC2 instances in your cluster.
Previously, to use a custom AMI you needed to ensure that all the EC2 instances within the cluster had the same underlying architecture. Now, you can specify custom AMIs for each instance type in an EMR instance group or EMR instance fleet in your cluster. Therefore, you can mix EC2 instances with different architectures by using multiple custom AMIs in the same cluster.
This feature is supported on Amazon EMR release 5.7.0 and later, and in all regions where Amazon EMR is available. To learn more, read Using a custom AMI in Amazon EMR documentation.