Posted On: Nov 9, 2022
Starting today, you can configure the set of instance types used when requesting EC2 capacity with attribute-based instance type selection.
Attribute-based instance type selection is a feature for Amazon EC2 Auto Scaling, EC2 Fleet, and Spot Fleet that makes it easy to create and manage instance type flexible capacity requests. With attribute-based instance type selection, you can define your instance requirements such as number of vCPUs and memory, and let EC2 Auto Scaling or Fleet pick the instances that fit the requirements you provided. Until now, attribute-based instance type selection allowed to exclude specific instance types from the selection through the excluded instance types list, but did not allow the reverse, i.e. only including specific instance types for consideration. For example, to allow using only M6i instance family, you would have had to exclude over 400 instance types. Now, you can just set M6i in the allowed instance types list to narrow down the selection set before applying other selection attributes, such as vCPU and memory. This is useful for workloads that have some extent of instance type flexibility but are still limited to specific instance types and want more granular control over which EC2 instance types to run on.
To get started, create or modify an Auto Scaling group or Fleet and specify allowed instance types list and other attributes such as the number of vCPUs and memory that you need. You can specify a list of instance types, families, or generations. Once you are done, EC2 Auto Scaling or Fleet will only consider the instance types that you explicitly included when selecting and launching instances.
This feature is now available in all public AWS Regions and AWS GovCloud (US). For more information, see the Amazon EC2 Auto Scaling attribute-based instance type selection documentation and blog post, EC2 Fleet attribute-based instance type selection documentation, and Spot Fleet attribute-based instance type selection documentation.