I measured the initial capacity performance of the Aurora Serverless v2 update "Up to 12 ACUs in under 1 second"
This page has been translated by machine translation. View original
Introduction
On August 5, 2026, an announcement was made regarding improved initial capacity during scale-up for Aurora Serverless v2.
In this article, we triggered a scale-up by changing the MinCapacity and measured the progression from 1 ACU to 30 ACU using CloudWatch metrics at 1-second intervals.
Verification Details
Verification Environment
| Item | Value |
|---|---|
| Engine | Aurora PostgreSQL Serverless v2 |
| Region | ap-northeast-1 (Tokyo) |
| PlatformVersion | 4 |
| Capacity settings before change | MinCapacity=1, MaxCapacity=30 |
| Capacity settings after change | MinCapacity=30, MaxCapacity=30 |
| Load | None (MinCapacity change only) |
The PlatformVersion was confirmed using ServerlessV2PlatformVersion in describe-db-clusters.
Measuring Scale-Up Speed
We raised the MinCapacity from 1 to 30 and measured how the capacity followed.
aws rds modify-db-cluster \
--db-cluster-identifier my-cluster \
--serverless-v2-scaling-configuration MinCapacity=30,MaxCapacity=30 \
--apply-immediately
The progression was obtained using the ServerlessDatabaseCapacity metric with period=1.
aws cloudwatch get-metric-statistics \
--namespace AWS/RDS \
--metric-name ServerlessDatabaseCapacity \
--dimensions Name=DBInstanceIdentifier,Value=my-instance \
--start-time 2026-08-06T03:29:00Z \
--end-time 2026-08-06T03:35:00Z \
--period 1 \
--statistics Maximum
The evaluation criteria are as follows.
- Change time: Execution time of
modify-db-cluster - Start time: First data point where ACU exceeded 1.0
- Completion time: First data point where ACU reached 30.0
Measurement Results
| Event | Elapsed time since change | ACU |
|---|---|---|
| MinCapacity change | 0 seconds | 1.0 |
| Scale-up started | 15 seconds | 1.0 |
| Initial expansion completed | 16 seconds | 13.5 |
| Reached 30 ACU | Approx. 3 min 45 sec | 30.0 |
Scale-up began 15 seconds after the change, and expansion of +12.5 ACU from 1.0 ACU to 13.5 ACU completed in 1 second. After 13.5 ACU, the value rose incrementally, reaching 30 ACU approximately 3 minutes and 45 seconds after the change.
Only the change points in the ACU progression are extracted below (intervals where the value remained the same are omitted).
| Elapsed time | ACU |
|---|---|
| 0 seconds | 1.0 |
| 16 seconds | 13.5 |
| 26 seconds | 14.0 |
| 33 seconds | 15.0 |
| 1 min 00 sec | 16.5 |
| 1 min 30 sec | 18.5 |
| 2 min 00 sec | 21.0 |
| 2 min 30 sec | 23.5 |
| 3 min 00 sec | 26.0 |
| 3 min 45 sec | 30.0 |
This measurement is a single measurement based on a MinCapacity configuration change. Behavior may differ with auto-scaling triggered by workload.
Summary
With this update, it is now possible to scale up to a maximum of 12 ACU within 1 second. In this simple measurement as well, expansion from 1 to 13.5 ACU completed in 1 second, and reaching 30 ACU was completed in approximately 3 minutes and 45 seconds.
On the latest platform (4), up to 30% performance improvement and doubling of auto-scale speed have been announced.
Combined with this update, there is a possibility that MinCapacity can be set lower than before. If you have set a high MinCapacity on an older platform and Aurora Serverless costs are a concern, we recommend updating to the latest platform (4) and reviewing your ACU settings.
