Posted On: Jul 26, 2018
Amazon EC2 Fleet simplifies the provisioning of EC2 capacity across different instance types, Availability Zones (AZs) and across On-Demand, Reserved Instances (RIs) and Spot purchase models, optimizing scale, performance and cost. Allocation strategies lets you determine how EC2 Fleet should select from instances and AZs you have specified to fulfill the desired capacity. Until today, you had the option of selecting the single cheapest Instance Pool for Spot and On-Demand, or diversifying Spot Instances across multiple instance types and AZs in an EC2 Fleet.
Starting today, you can use Prioritized List to specifically determine the order in which EC2 Fleet attempts to fulfill your On-Demand capacity. EC2 Fleet will attempt to launch all capacity using the instance with the highest priority first, and if all your capacity cannot be fulfilled using your highest priority instance then EC2 Fleet will attempt launching capacity using the second priority instance type. You can define a priority for all instances you specify in your EC2 Fleet.
Starting today you can also balance desired cost and availability in the context of applications by directing EC2 Fleet to evenly deploy Spot capacity across the N lowest priced Instance Pools. For example if you are running batch processing, you may prefer to set N to just 2 to maximize savings while attempting to ensure your queue always has compute capacity. However, if you are running a web service you may set N to 10 to minimize the impact of any Spot Instance Pool becoming temporarily unavailable.
These new allocation strategies are also available in Spot Fleet. To learn more Prioritized List in EC2 Fleet visit this page, and to learn more about and Lowest*N strategy visit this page. Visit this blog to learn more about EC2 Fleet.