Amazon EMR on EC2 Introduces New Allocation Strategies for Enhanced Flexibility and Control
Two new allocation algorithms for Amazon EMR on EC2 have been introduced by Amazon with the goal of improving control and flexibility when allocating instances for workloads. These new approaches give users more choices for managing On-Demand and Spot instances within their EMR https://aws.amazon.com/emr/ instance fleets and are accessible in all AWS Regions as well as AWS GovCloud (US) Regions.
Users can establish an order of preference for instance types with the first strategy, **prioritised allocation for On-Demand instances**. Amazon EMR will first try to employ the highest-priority instance type from the user-specified list in order to provision On-Demand capacity. EMR will then go to the next desired instance type if the demand is not fully satisfied, and so on, until the capacity is allotted.
The second technique, **capacity-optimized-prioritized for Spot instances**, respects the user's priority list on a best-effort basis and optimises for capacity. When minimising disturbances is important but particular instance types are still preferred, this approach works well.
Customers now have more control over their cloud computing resources thanks to these new allocation strategies, which are applicable to EMR 5.x releases (versions 5.0.0 and 5.0.3 excluded). This allows them to better customise the provisioning process to the unique requirements of their workloads. These strategies' configurations are easily managed via the CloudFormation, SDK, CLI, EMR API, and console.
Users can review Amazon's documentation on instance fleet allocation strategies for additional in-depth instructions on https://aws.amazon.com
