Idle Redshift Cluster
CostLens flags Amazon Redshift clusters that are barely used — low CPU and almost no connections over 14 days — as candidates to pause, downsize, or move to Serverless.
An always-on Redshift cluster bills per node-hour regardless of usage. A consistently idle cluster is paying full price for capacity nobody is querying. CostLens flags these so you can pause, downsize, or move to Redshift Serverless.
How it works
CostLens lists Redshift clusters
Calls redshift:DescribeClusters and skips clusters that aren't available (e.g. paused).
Reads usage metrics
Queries CloudWatch over 14 days for average CPUUtilization and peak DatabaseConnections.
Flags idle clusters
Flags clusters with average CPU < 15% and max connections < 5. Estimated savings is the cluster's monthly node cost (node price × node count × 730).
What to do
- Open the recommendation to see the node type, count, and observed CPU/connection levels.
- Decide the right action: pause the cluster, downsize it, or migrate to Redshift Serverless for bursty usage.
- This check is advisory — apply the change in the AWS console once you've confirmed the cluster is truly idle.
Severity levels
| Severity | Estimated monthly savings |
|---|---|
| critical | >$500 |
| high | $100–$500 |
| medium | $20–$100 |
| low | <$20 |
Required IAM permissions
Read-only — covered by the CostLens ReadOnlyAccess policy from AWS Accounts:
redshift:DescribeClusters
cloudwatch:GetMetricStatistics
Limitations
- Advisory only — CostLens does not pause or resize clusters automatically.
- Clusters without CloudWatch data (e.g. newly created) are skipped until enough history exists.