Serverless ApsaraDB RDS for MySQL scales compute resources automatically within a user-defined range based on workload, and expands storage automatically as data grows. Unlike fixed-specification instances, it charges only for what you use—and when automatic start and stop is enabled, paused instances incur storage charges only, with no compute fees.
How it works
A serverless instance uses RDS Capacity Unit (RCU) as its compute unit. One RCU is approximately equivalent to one CPU core and 2 GB of memory. You define a minimum and maximum RCU range, and the instance adjusts its performance within that range in response to workload changes.
Automatic scaling covers two dimensions:
Compute resources
Compute: The instance scales RCUs up or down based on CPU utilization and memory usage. Scale-out triggers when either CPU utilization or memory usage reaches the scale-out threshold; scale-in triggers only when both fall to the scale-in threshold.
Storage resources
Storage: When available space is insufficient, the instance expands storage automatically with no manual intervention. Automatic reduction is not supported—manual reduction is available when needed.
When automatic start and stop is enabled, the instance pauses after 10 minutes with no active connections and resumes automatically upon receiving a connection request.
Use cases
Variable workloads: Traffic that spikes during promotions, seasonal events, or unpredictable bursts. Without serverless, you either over-provision for peak demand and waste money at off-peak times, or under-provision and risk performance degradation. A serverless instance scales up to meet peak demand and scales back down automatically—no manual resizing required.
Development and testing: Staging environments and test databases that sit idle for extended periods. With automatic start and stop enabled, you pay only for storage while the instance is paused, eliminating the cost of keeping a fixed-specification instance running around the clock for occasional use.
Intermittent scheduled tasks: Workloads such as educational labs, batch jobs, or reporting pipelines that run at set times and are completely idle between runs. The instance pauses between sessions and starts automatically when a connection request arrives.
Unpredictable workloads: Internet of Things (IoT) and edge computing applications where usage patterns are difficult to forecast. The instance handles unexpected load spikes without manual intervention and scales back down when demand drops.
Limits
| Constraint | Details |
|---|---|
| Supported regions | China (Hangzhou), China (Shanghai), China (Qingdao), China (Beijing), China (Zhangjiakou), China (Hohhot), China (Ulanqab), China (Shenzhen), China (Heyuan), China (Guangzhou), China (Chengdu), China (Hong Kong), Singapore, US (Virginia), US (Silicon Valley), Germany (Frankfurt), UK (London), Thailand (Bangkok), South Korea (Seoul), Philippines (Manila), Japan (Tokyo), Malaysia (Kuala Lumpur), Indonesia (Jakarta) |
| Supported series | Basic series and High-availability series only |
| Storage type | PL1 ESSD only |
| RCU range | 0.5–32 RCU; scaling granularity of 0.5 RCU |
| Storage capacity | 20 GB–32 TB |
| Storage reduction | Manual only. After reduction, capacity must meet min{Data volume × 1.3, Data volume + 400 GB} and cannot fall below the instance type minimum |
| Manual pause | Not supported. Instances can only be paused via automatic start and stop |
| MySQL versions | 8.0 and 5.7 only |
Billing
Serverless instances are billed hourly. View billing details on the Expenses and Costs page.
Cost breakdown
Total fee = Fee per node × Number of nodesNumber of nodes: Basic series has one node; High-availability series has two nodes.
Compute resource fee
Compute fee per node:
Unit price of compute resources × Hourly RCU usageHourly RCU usage is the average RCUs consumed per second over the hour.
Unit price of compute resources: Pricing varies by region. For example, in Singapore, the RCU unit price is
USD 0.07455/RCU-hour. For other regions, see the ApsaraDB RDS purchase page.Singapore compute unit price: USD 0.07455/RCU-hour. For other regions, see the ApsaraDB RDS purchase page.
Storage resource fee
Storage fee per node:
Unit price of storage resources × Storage capacityStorage capacity is shown under Basic Information > Instance Resources > Storage Capacity32 TB.
Unit price of storage resources: Pricing varies by region. For example, in Singapore, the storage unit price is
USD 0.0004/GB-hour. For other regions, see the ApsaraDB RDS purchase page.Singapore storage unit price: USD 0.0004/GB-hour. For other regions, see the ApsaraDB RDS purchase page.
How start and stop affects billing
When automatic start and stop is enabled, billing changes as the instance transitions between states:
| Instance status | Compute fee | Storage fee |
|---|---|---|
| Running | Billed | Billed |
| Pausing | Billed | Billed |
| Paused | Not billed | Billed |
| Starting | Not billed | Billed |
The full cycle: the instance runs normally → no connections for 10 minutes → begins pausing → fully paused → a connection request arrives → begins starting → returns to running.
Billing example
Setup: Basic series (single-node) instance in Singapore, with a 0.5–8 RCU range and 20 GB storage. Over 24 hours: 1 hour at 8 RCUs (peak) and 23 hours at 1 RCU.
Unit prices (Singapore):
| Billable item | Unit price |
|---|---|
| RCU | USD 0.07455/RCU-hour |
| Storage | USD 0.0004/GB-hour |
Calculation:
RCU fee: (0.07455 × 8 × 1) + (0.07455 × 1 × 23) = USD 2.31105
Storage fee: 0.0004 × 20 × 24 = USD 0.192
Total: USD 2.50305
Choose an RCU range
The RCU range defines the minimum and maximum compute capacity of your instance. The right range reduces both cost and performance risk.
Choose the minimum RCU
Setting the minimum RCU to 0.5 minimizes idle costs, but there are trade-offs to consider:
Scale-up speed depends on current capacity. The higher the current RCU, the faster the instance can scale to handle a spike. If your workload can spike suddenly and requires rapid scale-out, set a higher minimum so the instance reaches peak capacity faster.
Use 0.5 RCU when idle cost is the priority. This is best for development environments, intermittent batch jobs, or workloads where gradual scaling is acceptable.
Choose the maximum RCU
Run a stress test to measure the peak CPU cores—or RCUs—required at maximum load. One RCU is approximately one CPU core and 2 GB of memory.
Set the maximum high enough to handle peak demand without throttling.
Adjust after launch
Monitor actual RCU usage over the past 30 days (see View RCU usage and storage changes) and adjust the range as your workload evolves.
Create a serverless instance
Before creating a new instance, note that you can also convert an existing pay-as-you-go instance to serverless. Subscription instances must first be converted to pay-as-you-go before conversion to serverless.
To create a new serverless instance:
Go to the ApsaraDB RDS purchase page and set Billing Method to Serverless.
Configure basic instance information:
Select a Region. Region selection is final after purchase.
To connect from an Elastic Compute Service (ECS) instance, select the same region.
To connect from other devices, select a region close to the devices to reduce latency. You can then access the instance over the Internet.
Select MySQL as the Engine. Only versions 8.0 and 5.7 are supported.
Configure instance architecture and storage:
Select a Series:
Basic series: Single-node deployment. Cost-effective; suitable for development, testing, or scenarios that do not require high availability.
High-availability series: Primary/secondary dual-node deployment with automatic failover. Suitable for production environments.
Select PL1 ESSD as the Storage Type.
Configure automatic scaling:
Set the Resource Scaling Range (RCU). The instance scales within this range based on workload.
Set the Elastic Policy for cases where smooth scaling is not possible: In most cases, RCU scaling takes effect immediately without affecting running workloads. Force Execution is needed only in rare situations such as cross-host scaling.
Force Execution: Forces RCU scaling to maintain performance. May trigger a switchover. Make sure your application supports automatic reconnection.
Do Not Force Execution: Defers RCU scaling temporarily to preserve active connections.
Set Automatic Start/Stop. When enabled, the instance pauses after 10 minutes of inactivity and resumes on the next connection request. You are charged only for storage while paused.
Configure network and zone:
Select a VPC (Virtual Private Cloud). If connecting from an ECS instance, select the same VPC.
Select whether to Add to Whitelist. When enabled, ECS instances in the same VPC can directly access the RDS instance.
Select a Primary Zone and Network. For High-availability series, configure the secondary zone and deployment method (multi-zone or single-zone).
Set the Storage Capacity. The minimum adjustment step size is 5 GB.
(Optional) Configure More Settings. Keep the default values unless you have specific requirements.
Parameter Description Parameter Template Select high-performance, asynchronous, or default. See Use a parameter template. Time Zone Set the instance time zone. Table Name Case Sensitivity Default is Case-insensitive. Change to Case-sensitive if your source database is case-sensitive. Release Protection Prevents accidental deletion. See Enable or disable release protection. Instance Name A description to help identify the instance. Tags Attach tags for classification. See Filter instances by tag. Resource Group Assign to a default or custom resource group. Privileged Account Select Configure Now to create a privileged account, or Configure Later to do it after instance creation. Review the configuration and click Confirm Order. To create multiple instances with identical configurations, set the quantity (up to 20) in the top-right corner before confirming.
After payment, click Back to Console or go to the ApsaraDB RDS instance list. Select the region at the top to locate your instance. Instance creation takes about 10 minutes. The instance is ready when its status shows Running.
There may be a short delay between successful payment and the instance appearing in the console. If the instance is not visible immediately, wait a moment and refresh.
View RCU usage and storage changes
Review RCU and storage changes over the past 30 days to determine whether to adjust your elastic scaling configuration.
Go to the ApsaraDB RDS instance list, select a region, and click the instance ID.
In the left navigation pane, click Monitoring and Alerts.
On the Standard Monitoring tab, click Old Version in the top-right corner.
On the Resource Monitoring tab, view the Database Capacity (RCUs) and Disk Space metrics.
Adjust elastic scaling settings
Adjust the RCU range and elastic policy
Go to the ApsaraDB RDS instance list, select a region, and click the instance ID.
In the Configuration Information section, click Adjust Elastic Settings.
In the panel that appears:
RCU scaling range: Adjust the minimum and maximum RCU values.
Elastic policy: Choose Force Execution or Do Not Force Execution for rare cases where smooth scaling is not possible.
Click OK.
Manually adjust storage capacity
Storage expands automatically when space runs low. To reduce storage manually:
Go to the ApsaraDB RDS instance list, select a region, and click the instance ID.
On the Basic Information page, in the Instance Resources section, click Modify next to Storage Capacity.
Adjust the storage value, click OK, and confirm in the dialog box.
Storage reduction requires data copying and takes slightly longer than expansion—typically a few minutes. During expansion or reduction, the instance status shows Upgrading/Downgrading. It returns to Running after the operation completes.
Manage instance start and stop
Enable or disable automatic start and stop
When enabled, the instance pauses after 10 minutes with no active connections and resumes on the next connection request. Compute charges stop while the instance is paused.
Go to the ApsaraDB RDS instance list, select a region, and click the instance ID.
In the Configuration Information section, click Modify Scaling Settings.
Click Advanced Settings, select Enable or Disable from the Automatic Stop and Restart drop-down list, and click OK.
Startup takes approximately 6–40 seconds depending on the state of the instance's databases and tables. The instance is unavailable during startup.
Application design considerations: During startup, connection attempts may fail if many requests arrive before the instance finishes resuming. Implement retry logic in your application to handle these transient connection failures.
Manually start an instance
Paused instances can be started manually. Running instances cannot be paused manually—they can only be paused through the automatic start and stop feature.
From the instance list: Go to the ApsaraDB RDS instance list, find the instance, and in the Actions column click More > Start Instance.
From the instance details page: Go to the ApsaraDB RDS instance list and click the instance ID. On the Basic Information page, in the Running Status section, click Start Instance.
View start and stop history
Go to the RDS console. In the left navigation pane, click Tasks to view automatic and manual start and stop records.
Upgrade from Basic series to High-availability series
When you need enhanced disaster recovery or additional High-availability features, upgrade a Basic series instance to High-availability series.
Go to the ApsaraDB RDS instance list, select a region, and click the instance ID.
In the Configuration Information section, click Adjust Elastic Settings.
In the panel that appears, select High-availability series next to Series, and click OK.
Review the pre- and post-upgrade information in the confirmation dialog, and click Confirm. The instance status shows Upgrading/Downgrading during the upgrade and returns to Running when complete.
Schedule RCU configuration
RCU scaling typically completes in seconds. In rare cases—such as cross-host scaling—it may take 3–5 minutes. To guarantee stability during a specific time window (for example, before a planned traffic surge), schedule RCU configuration to proactively increase capacity in advance. For details, see Schedule the RCU configuration of a serverless instance.
FAQ
Why didn't my instance pause automatically?
The automatic start and stop feature is disabled by default. Enable it under Configuration Information > Modify Scaling Settings > Advanced Settings. If it is already enabled, check whether active connections are preventing the pause:
On the instance details page, go to Monitoring and Alerts and check the Session Connections metric.
To identify which connections are open, run
SHOW PROCESSLISTon the instance.
If you have a public IP address configured with an open whitelist, external connection scans may keep the instance active. Restrict the whitelist or remove the public endpoint to allow automatic pausing.
How long does RCU scaling take, and what triggers it?
In most cases, RCU scaling takes effect immediately. In rare cases—such as cross-host migration—it may take 3–5 minutes. Cross-host migration is less likely for High-availability series instances because of the secondary node.
Scaling triggers:
Scale-out: CPU utilization OR memory usage reaches the scale-out threshold. Default CPU scale-out threshold: 60%; default memory scale-out threshold: 90%.
Scale-in: Both CPU utilization AND memory usage fall to the scale-in threshold. Default CPU scale-in threshold: 40%; default memory scale-in threshold: 80%.
These thresholds may be adjusted automatically to maintain availability.
Will I be charged when my serverless instance is idle?
It depends on whether automatic start and stop is enabled:
Enabled: The instance pauses after 10 minutes of inactivity. Only storage is billed while paused; compute is not.
Disabled: The instance stays running regardless of activity. Both compute and storage are billed. Compute charges are typically based on your configured minimum RCU.
Can a High-availability series instance be deployed across multiple zones?
Yes. A High-availability series instance can be deployed in a single zone or across multiple zones, based on your requirements.