Skip to main content

OverProvisioned Recommendations

The OverProvisioned Recommendations section identifies resources that are larger than necessary for their workloads.
By right-sizing these resources, you can significantly reduce AWS costs while maintaining optimal performance.


Overview of OverProvisioned Recommendations

This section groups recommendations by service and highlights:

  • Potential Savings from downsizing overprovisioned resources.
  • Number of Recommendations for each service.
  • Actions to Downsize instances or Snooze for later review.
OverProvisioned recommendations overview with potential savings and resource counts

EC2 OverProvisioned Recommendations

For EC2 instances, CloudKeeper Tuner analyzes usage patterns over several months to detect underutilized resources.
The dashboard shows:

  • Account ID, Resource ID, and Region for easy identification.
  • CPU and Memory Utilization statistics over the last several months.
  • Suggested downsizing actions (e.g., "Downsize instance from r5a.8xlarge to r5a.4xlarge").
  • Estimated Potential Savings from right-sizing.
Detailed EC2 overprovisioned recommendations with usage stats and downsizing actions

Taking Action

Unlike Cleaner recommendations, actions for overprovisioned resources are manual to ensure a human review before modifying potentially production workloads. For recommendations like rightsizing, we want customers to first review the workloads before making changes, so nothing critical gets removed or any data loss occurs. This way, your workloads are always protected, and only informed decisions are taken.

Downsizing an instance is a more invasive change, and the final decision should be made by your engineering team after reviewing the provided utilization data.

CloudKeeper's support team is available to assist and guide you through implementing these changes if needed.


The OverProvisioned Recommendations help ensure that your AWS resources are right-sized to match their workload, eliminating unnecessary expenses while keeping performance stable.