Gain more control and Achieve better performance with Preferred Instance types


You now have more control over the distribution of instance types within your cluster, letting you achieve better performance and greater cost efficiency by optimizing the makeup of your cluster for your specific workload.

When creating an Elastigroup, you select Spot instance types for Spotinst to take into consideration when applying the selection algorithm that provision instances in the most cost-effective and available Spot markets.

While maximizing the number of selected Spot instance types can improve availability and reduce costs, customers often seek to scope down the number of instance types selected to those with compute resources which optimally fit their workloads for example M3, C3, C3, M4.

Now, you can get the best of both worlds with preferred instance types. Simply broaden your selection of Spot instance types to achieve the best availability and optimize for cost savings, while also marking your ideal instance types as “preferred”.
Spotinst’s algorithms will take over from there to ensure that your workload runs on preferred instance types when possible, while sufficiently spreading your capacity across different Spot markets to ensure high availability. For example, it allows you to run your workload on General Purpose instances such as M{3,4} \ C{3,4,5} when Spot capacity is available and use other High-Performance Spot instances such as I3, I2 etc, Instead of using On-Demand instances.

Preferred instance types feature is now available through the Spotinst console and APIs. Click here to learn more about preferred instance types.