How's it done?

Customers of E1 can install the EC2 Rightsizer system for free from AstroHub.

Readers not currently using Echelon One can still build their own automation using the following AWS services:

• AWS CloudWatch (with custom memory metrics)
• AWS Compute Optimizer Enabled

Confidence in a new instance type is much easier when you see less than 30% hardware utilization for months.

Dynamically rightsizing EC2s begins with analyzing historical CPU and memory utilization for each instance. AWS Compute Optimizer provides excellent recommendations for reducing over-provisioned instance types to save on cost, as well as up-sizing instances if you are more concerned with workload performance. Echelon One provides additional logic capabilities to drive value even further.

Bottom line, AWS Compute Optimizer relies on AWS CloudWatch. Install the CloudWatch agent ensure it can push metrics to enable accurate recommendations.

Case Study

One customer used the EC2 Rightsizer system to drive a FinOps project from analysis to execution. They rightsized thousands of EC2 instances to successfully avoid spending millions more in annual EC2 compute costs.

The compute spend rate was reduced by 62 percent!

This cost would have continued to accrue year-over-year and likely grow. By consistently reducing over-provisioned instances and converting to AMD instance types, they were equipped to purchase a long-term Compute Savings Plan on an optimized environment to dramatically drive down their spend rate.

How do I know what I could save?

The EC2 Rightsizer system provides an estimate of potential savings for all EC2 instances across your AWS Organization. Otherwise, aggregate your AWS Compute Optimizer data for all your AWS accounts & regions.

How do I know what to change the instance type to?

Choose from AWS Compute Optimizer's instance type recommendations based on how aggressive you want to start with. E1 brings it all together in one central view and can be easily tailored to your preferences such as only using low risk options, for an example.

How do I schedule the automation?

In E1 scheduling is done the same way as any system command - within a few clicks. If building your own, you may want to setup a cron job for a local script, or an AWS Eventbridge rule to trigger a AWS Lambda function.

How do I handle exceptions to the automation?

In E1, easily track exemptions using your fully customizable system data table, or create a new target as applicable. When building your own, reference a version controlled file that can be added to/removed from.