Skip to main content

Recommendation types

Overview

Recommendations appear on Home → Recommendations and in the recommendations detail view. Each card includes type, affected resource, estimated savings, and suggested action.

Common recommendation types

TypeDescriptionTypical savings
RightsizingInstance size exceeds observed CPU/memory utilization20–30% on compute
Idle resourcesRunning instances with near-zero utilizationVaries
Unattached storageEBS volumes or disks not attached to instancesHigh per-volume
S3 duplicatesDuplicate objects across buckets30–50% on affected storage
RI/SP coverage gapOn-demand usage that could use commitments20–40%
TTL violationsResources exceeding configured lifetime policiesPrevents runaway cost
Power scheduleResources running outside business hours10–20% on dev/test

Prerequisites for recommendations

  • Connected data source with synced billing data
  • Resource discovery enabled (IAM permissions)
  • Sufficient usage history (typically 7–30 days depending on type)

Acting on recommendations

  1. Open the recommendation detail panel
  2. Validate impact with resource owners
  3. Apply fix in cloud console or via automation
  4. Archive the recommendation when complete