Skip to main content

Description

This policy identifies AWS Redshift Clusters that are consistently operating with low CPU utilization and disk I/O over the last 30 days, which may indicate that they are oversized for their current workload. A cluster is flagged as underutilized if it meets the following conditions:

  1. Average CPU utilization has been below 20%.
  2. Average disk read IOPS has been under 50.
  3. Average disk write IOPS has been under 50.

Rationale​

Over-provisioning Redshift clusters results in unnecessary cloud costs. Identifying underutilized clusters enables you to make informed decisions about downsizing, leading to cost savings and ensuring that resources are better aligned with actual demand.

Impact​

Resizing a cluster may require scheduling a maintenance window and could result in temporary downtime.

Audit​

This policy evaluates an AWS Redshift Cluster based on its 30-day performance metrics.

The Cluster is marked as INCOMPLIANT if all the following criteria are met:

  • CloudWatch: CPU, 30-Day metric is less than 20%.
  • CloudWatch: Disk Read IOPS, 30-Day metric is less than 50 IOPS.
  • CloudWatch: Disk Write IOPS, 30-Day metric is less than 50 IOPS.

The Cluster is marked as INAPPLICABLE if it is not in an available Status or it has been running for less than 30 days.

The Cluster is marked as UNDETERMINED if any of the evaluated metrics are empty, indicating insufficient data in the CMDB to assess whether the Cluster is underutilized.