Instance class and I/O drive Aurora spend

Amazon Aurora cost optimization

Aurora bills for instances, storage, and I/O, and each has a lever most teams never pull. Right-sizing, the move from Serverless v1 to v2, and the I/O-Optimized decision are where the savings hide.

Why the waste happens

  • Provisioned instances are oversized for the real workload, or left running for environments that are idle.
  • Clusters still run Aurora Serverless v1, which is end-of-life, instead of the cheaper and more elastic v2.
  • High-I/O clusters stay on standard storage when Aurora I/O-Optimized would be cheaper past a breakeven point.
  • Instances run non-Graviton classes (db.r5) instead of Graviton (db.r6g) at a discount.

How to spot it

  • Instances with low CPU and few connections over a representative period.
  • Clusters still on Aurora Serverless v1.
  • Clusters where I/O charges are a large share of the bill (often the I/O-Optimized breakeven around 25 percent).
  • Non-Graviton writer and reader instances.

The fix

  • Right-size provisioned instances and scale down or pause non-production clusters.
  • Migrate Aurora Serverless v1 clusters to v2 for finer-grained scaling and lower cost.
  • Switch high-I/O clusters to Aurora I/O-Optimized once I/O passes the breakeven share of spend.
  • Move instances to Graviton classes for a same-performance discount.

Find every Aurora issue in one scan

Cloud Cost Analyzer evaluates Aurora clusters for over-provisioning, Serverless v1 that should move to v2, the I/O-Optimized breakeven, and non-Graviton classes, and estimates the monthly saving for each change.

$ cca scan --provider aws

Read-only, runs in about 5 minutes. Free tier, no credit card.

Aurora cost FAQ

When should I use Aurora I/O-Optimized?+

Aurora I/O-Optimized removes per-request I/O charges in exchange for higher instance and storage rates. It becomes cheaper once I/O is roughly a quarter or more of your Aurora bill. For I/O-heavy clusters it can cut costs and make spend predictable.

Should I migrate Aurora Serverless v1 to v2?+

Yes. Serverless v1 is end-of-life. v2 scales in finer increments, scales faster, and supports more features, and it is generally more cost-effective for variable workloads. Plan the migration before v1 support ends.

Does Aurora support Graviton?+

Yes. Aurora offers Graviton-based instance classes (for example db.r6g) that are typically cheaper than the equivalent Intel classes for the same performance, usually a minor version and instance-class change.

See what you are overpaying for

Point a read-only CLI at your account and get a ranked fix list with dollar amounts. Free tier, no credit card.