Opportunity Name
Aurora Hybrid Read Replica Optimization
AWS Resource Type (AWS service name)
Amazon Aurora (Amazon RDS)
Opportunity Description
This CloudFix cost optimization Finder identifies Aurora database clusters with underutilized provisioned read replicas and recommends converting those replicas to Aurora Serverless v2 with a minimal baseline capacity (0.5 ACU). This can deliver large savings for replicas that exist primarily for high availability/failover, while still allowing replicas to auto-scale if workload increases.
It supports both Aurora MySQL and Aurora PostgreSQL engines.
Criteria for identifying the opportunity
CloudFix flags an opportunity when all of the following are true:
Cluster requirements
-
Aurora engine is MySQL 5.7+ or PostgreSQL 10.14+
-
Cluster is not part of an Aurora Global Database
-
Cluster engine version is Aurora Serverless v2 compatible
Read replica requirements
-
Cluster has at least one read replica
-
Replica p95 CPU utilization < 20% over a 7-day lookback
-
Resource is not tagged with
ff_dont_fix_it
Cost threshold
-
Estimated minimum annual savings > $100
Potential Savings (if known)
-
Up to 90% cost reduction for idle / very low-utilization read replicas (particularly where the replica primarily exists for HA).
What happens when the Fixer is Executed?
This opportunity does not have an automatic Fixer. CloudFix provides the recommendation and supporting analysis, but the change must be implemented manually in AWS.
Is it possible to roll back once CloudFix implements the Fixer?
No — this is a Finder-only opportunity.
Can CloudFix implement the fix automatically once I accept the recommendation?
Not applicable — CloudFix does not apply an automated change for this opportunity. If you implement the change manually, rollback would also be manual (e.g., revert the read replica back to a provisioned instance configuration).
Does the fix require downtime?
No application changes are required, and this optimization is designed to be performed without application downtime. (As with any database change, CloudFix recommends using a normal change window and monitoring, especially for latency-sensitive workloads.)
Bill Gleeson
Comments