Modify the resource scaling range (RCU)
If your database has insufficient resources during peak workloads or you want to reduce costs during off-peak periods, you can modify the RDS SQL Server Serverless instance's Resource Scaling Range (RCU).
To ensure product stability and user experience, Alibaba Cloud will discontinue the sale of RDS for SQL Server serverless instances on November 3, 2025, and will stop providing related technical support on June 1, 2026. This includes, but is not limited to, new feature iterations and technical issue fixes. To ensure your product experience, convert your RDS for SQL Server serverless instances to pay-as-you-go instances as soon as possible.
How it works
The performance of an RDS SQL Server Serverless instance is measured in RDS Capacity Units (RCUs). A serverless instance automatically scales RCUs up or down based on workload, within the RCU range you define.
RCU auto-scaling depends on the instance's CPU utilization, memory pressure, and Average Active Sessions (AAS). When any of these metrics reaches the scaling-up threshold, the instance scales up within minutes. When all three metrics reach the scaling-down threshold, the instance scales down within minutes.
Use cases
-
If you anticipate a traffic spike that could cause performance jitter, you can raise the minimum RCU value to ensure resource availability.
-
To reduce costs, you can lower the maximum RCU value. However, proceed with caution, as this action may impact instance performance.
Impacts
-
Changes to the serverless instance's RCU range take effect in about one minute.
-
Modifying the maximum or minimum RCU value has the following effects:
-
Modify the maximum RCU value
-
If the current RCU value is lower than the new maximum value, the RCU value remains unchanged.
-
If the current RCU value is higher than the new maximum value, the RCU value immediately scales down to the new maximum value after the change takes effect.
-
If the instance is already constrained by the current maximum RCU, it immediately scales up to the new maximum value after you increase it.
For example, if a traffic spike occurs and the current maximum of 6 RCUs is insufficient, and you change the maximum to 8, the system immediately scales up to 8 RCUs after the configuration takes effect.
-
-
Modify the minimum RCU value
-
If the current RCU value is higher than the new minimum value, the RCU value remains unchanged.
-
If the current RCU value is lower than the new minimum value, the instance immediately scales up to the new minimum after the change takes effect.
-
-
Billing
Modifying the Resource Scaling Range (RCU) is free; RCU fees are based only on usage.
Procedure
Related topics
-
If your serverless instance's storage capacity is insufficient for your workload, you can manually increase the storage capacity.
-
To save costs, you can pause the serverless instance when it is idle. For more information, see Manually pause or start an instance and Billing for different instance statuses.
-
If you no longer need a serverless instance, you can release the instance.
-
RCU scaling typically completes in seconds. In rare cases, cross-machine scaling may take 3 to 5 minutes. If you require strict stability during specific periods, you can schedule RCU configuration changes to increase the RCU count proactively.
-
To modify the RCU range programmatically, call the ModifyDBInstanceSpec API.