Automatic performance scaling is available for ApsaraDB RDS for MySQL instances that run RDS High-availability Edition with local disks and use general-purpose instance types. The feature lets Database Autonomy Service (DAS) adjust the compute specifications of an instance automatically as traffic changes.
Overview
Business growth, increasing data volumes, and newly launched services or features drive up database traffic. When compute resources run short, scaling up the compute specifications (CPU) of your database is an effective solution.
With automatic performance scaling, Database Autonomy Service (DAS) automatically scales up the compute specifications of a database instance so that the instance quickly adapts to peak traffic. When traffic drops, DAS scales the specifications back down to avoid wasted resources.
Effective date
July 5, 2024
Supported scope
Automatic performance scaling applies only to instances that meet all of the following conditions:
Billing method: subscription or pay-as-you-go.
If the billing method is serverless, instance performance scales in and out automatically and requires no configuration.
RDS edition: RDS High-availability Edition.
Storage type: local disk.
Product type: standard.
Instance type: general-purpose.
The instance resides in a region that supports the DAS abnormality check feature. For more information, see DAS abnormality check.
Your Alibaba Cloud account has a sufficient balance to support the scale-up.
Billing
For instances that run RDS High-availability Edition with local disks and use general-purpose instance types, automatic performance scaling is disabled by default and incurs no charges. If you enable the feature, you are charged on a pay-as-you-go basis. The following table describes the unit price in each region.
| Region | Unit price (USD per CPU core-hour) |
| China (Zhangjiakou) and China (Ulanqab) | 0.063 |
| China (Hong Kong) and South Korea (Seoul) | 0.134 |
| Japan (Tokyo) | 0.100 |
| Malaysia (Kuala Lumpur) | 0.102 |
| Singapore and Indonesia (Jakarta) | 0.155 |
| Germany (Frankfurt) and UK (London) | 0.078 |
| US (Virginia) and US (Silicon Valley) | 0.129 |
| UAE (Dubai) | 0.091 |
| Other regions (all regions that are not listed above) | 0.083 |
Automatic performance scaling is billed by the minute and settled once every hour on the hour. The billing formula is Unit price x Number of CPU cores added x Number of hours.
For example, an instance in the China (Hangzhou) region has 4 CPU cores and is scaled up to 8 CPU cores for 30 minutes. China (Hangzhou) is not listed in the table and is therefore billed at the Other regions price, USD 0.083 per CPU core-hour. The fee is 0.083 (unit price) x 4 (cores added) x 0.5 (hours) = USD 0.166.
Limits
Automatic performance scaling increases only the number of CPU cores and the IOPS. It does not increase the memory or the maximum number of connections of the instance.
If host resources are insufficient, automatic performance scaling does not proceed.
Automatic performance scaling is performed only on the primary instance. If a primary/secondary switchover occurs after the primary instance is scaled up:
When the new primary instance meets the conditions for a scale-up or scale-down, the system automatically scales the instance up or down. The new primary instance is the instance that served as the secondary instance before the switchover. It has not been scaled up.
When the new secondary instance meets the conditions for a scale-down, the system automatically scales the instance back down to the original specifications. The new secondary instance is the instance that served as the primary instance before the switchover. It is the instance that was scaled up.