Skip to main content

Remediate Idle Redshift Cluster

Overview​

This tutorial demonstrates how to remediate an idle Amazon Redshift cluster that has had no database connections for 30 or more days. An idle Redshift cluster continues to incur full compute and storage costs regardless of usage. Pausing or deleting the cluster eliminates this waste.

Cost Impact: Redshift pricing is based on node type and count. For example, a single dc2.large node costs ~$180/month, while a 2-node ra3.xlplus cluster costs ~$782/month. A cluster with zero connections for 30+ days represents 100% waste — the full monthly cost can be recovered by pausing or deleting it.

Prerequisites​

  • Access to the AWS Console
  • Appropriate IAM permissions to manage Redshift clusters (redshift:PauseCluster, redshift:DeleteCluster, redshift:DescribeClusters)
  • Confirmation that the cluster is genuinely unused (check with application owners)

Step 1: Navigate to the Redshift Cluster​

Navigate to the Redshift Clusters page in the AWS Console at: https://us-east-1.console.aws.amazon.com/redshiftv2/home?region=us-east-1#clusters

Step 2: Select the Idle Cluster​

Click on the cluster name to open its detail page. Review the cluster summary to confirm the node type, number of nodes, and current status.

Step 3: Verify the Cluster Is Truly Idle​

Before taking action, confirm the cluster is not in use:

  • Check the Query monitoring tab: Look for any recent query activity.
  • Check the Database connections metric: In the Monitoring tab, verify that DatabaseConnections has been at or near zero for the past 30 days.
  • Check with stakeholders: Confirm with application teams or data analysts that no workloads depend on this cluster.

Step 4: Choose a Remediation Path​

Based on whether the cluster may be needed again, choose one of the following:

Pausing stops compute charges while retaining the cluster configuration and data. You only pay for storage while paused.

  1. From the cluster detail page, click Actions → Pause cluster.
  2. Confirm the pause operation.
  3. The cluster status will change to Pausing and then Paused.

Note: Paused clusters automatically resume after 24 hours if you have scheduled actions. Ensure no scheduled actions will trigger an unintended resume.

Deleting the cluster eliminates all costs. A final snapshot preserves your data for future restoration.

  1. From the cluster detail page, click Actions → Delete cluster.
  2. When prompted, select Create final snapshot and provide a snapshot name (e.g., cluster-name-final-YYYYMMDD).
  3. Confirm deletion.

Note: If you skip the final snapshot, all data in the cluster is permanently lost.

Step 5: Verify the Change​

  • If paused: Confirm the cluster status shows Paused on the clusters list.
  • If deleted: Confirm the cluster no longer appears in the active clusters list. If you created a final snapshot, verify it appears under Snapshots.

Alternative Approaches​

Resize to a Smaller Cluster​

If the cluster is occasionally used but over-provisioned, consider resizing to fewer or smaller nodes. Use Actions → Resize to scale down while keeping the cluster available.

Schedule Pause/Resume​

If the cluster is only used during business hours or specific periods, configure a scheduled pause/resume cycle. Under Actions → Create scheduled action, set a recurring pause for off-hours and a resume for working hours. This can cut costs by 60-70%.

Migrate to Redshift Serverless​

For intermittent workloads, Redshift Serverless automatically scales compute to zero when idle — you only pay for storage and actual query processing. This eliminates the need to manage pause/resume schedules.

Summary​

  • Remediation: Pause or delete the idle Redshift cluster
  • Savings: 100% of compute costs when paused (storage still charged) or 100% of all costs when deleted
  • Risk: Low if you create a final snapshot before deletion; pausing is fully reversible
  • Rollback: Resume a paused cluster instantly; restore a deleted cluster from its final snapshot

Regularly monitoring Redshift cluster connections ensures you're not paying for idle compute resources, especially in development or staging environments where clusters are often forgotten.