Practical guide
RDS rightsizing: compare instance cost without losing headroom
compare RDS MySQL instance sizes and costs. Includes a worked example, calculator scope, and practical questions.
Updated · Sources checked
A cheaper instance is useful only if it serves the workload
RDS rightsizing starts with memory, CPU, connections, and storage behavior under representative demand. Compare the cost of supported candidate classes only after identifying which resources constrain the database. A lower hourly rate can be offset by slower queries or additional replicas.
The calculator provides the MySQL Single-AZ db.t3 instance subtotal for its supported US East choices. It does not benchmark performance, predict memory requirements, or price every RDS engine and deployment.
Worked example: isolate an hourly-rate change
Use hypothetical hourly rates of $0.08 for the current instance and $0.05 for a smaller candidate. At 730 hours, instance subtotals are $58.40 and $36.50 respectively. The potential instance-only difference is $21.90 per month.
These are teaching rates, not an AWS quote. If moving to the candidate requires a second identical instance for the same workload, the candidate’s total becomes $73.00. That would be $14.60 above the original single-instance subtotal, before additional storage or transfer.
The example shows why changing the class and architecture simultaneously needs a complete scenario comparison. In the real tool, use its displayed verified rate and supported class rather than expecting these hypothetical rates to appear.
Preserve the resource that matters
Review a peak business interval, scheduled reports, and maintenance windows. Average CPU can look comfortable while connection bursts or memory pressure cause poor response times. A burstable class also needs CPU-credit review when usage is sustained.
Keep storage assumptions unchanged only if the proposed deployment actually preserves them. The calculator omits storage, I/O-related charges, backup, transfer, and other components. It cannot establish that a smaller instance is safe or that a larger one will remove a query bottleneck.
A reversible trial with application latency and error checks gives stronger evidence than the cost difference alone. Record the workload and period so the comparison remains meaningful after traffic grows.
Frequently asked questions
Does low average CPU prove I can downsize? No. Examine memory, connections, burst behavior, and peak latency too.
Should I multiply the saving by 12? Only for a stable year-long scenario. Planned growth, seasonal load, or configuration changes require separate monthly estimates.
Calculate your scenario
Use the rds cost calculator. Keep the displayed region, pricing date, billing units, and exclusions alongside your result. The arithmetic examples above illustrate usage or explicitly hypothetical rates; they are not AWS quotes.
Source
Billing structure checked against official AWS documentation on September 6, 2026.