Feature Overview
|
Cloud Computing Unit (CCU) | Allows adjustment of the CCU elastic scaling range. The Serverless instance automatically increases or decreases CCU within this range based on actual business pressure. |
Elasticity policy | The Serverless instance continuously monitors the load conditions of nodes, such as CPU and memory, and triggers the automatic scaling policy based on rules. |
Automatic disk capacity expansion | Automatically expands disk capacity when disk utilization exceeds the threshold, requiring no manual intervention. |
CCU: Cloud Computing Unit
CCU Definition
CCU (Cloud Computing Unit) is the billing unit for computing resources in Serverless. One CCU is approximately equivalent to the computing resources of one CPU core and 2 GB of memory.
The CCU usage per billing cycle is calculated as the maximum value between the number of CPU cores used by the database and half of the memory size.
Calculation formula: CCU = max(CPU cores, Memory GB / 2)
Example:
For an actual configuration of 4 CPU cores and 6 GB of memory, CCU = max(4, 6/2) = 4 CCU.
For an actual configuration of 2 CPU cores and 10 GB of memory, CCU = max(2, 10/2) = 5 CCU.
Compute Power Range Configuration
An elastic range, which defines the upper and lower limits for the computing power scaling of a single node, must be configured for a Serverless instance:
|
Minimum Compute Power | The minimum computing resources that an instance always retains, billed at this level even under no load. | Starting from 1 CCU |
Maximum Compute Power | The upper limit of computing resources that an instance can elastically scale to, which will not be exceeded even under the highest load. | Up to 16 CCU |
Memory is automatically adjusted in a 1:2 ratio relative to the CPU. For example, if you set the minimum computing power = 1 and the maximum = 8, a single node automatically scales between 1C2G and 8C16G.
Note:
When setting the elastic range for the first time, set the minimum computing power to 1 CCU and select a higher value for the maximum to achieve greater elasticity.
Elastic Policy
A Serverless instance continuously monitors the CPU and memory utilization of its nodes and automatically triggers scaling based on the load:
Scaling-Out Policy
When CPU utilization consistently exceeds the scaling-out threshold (80% by default), the system automatically calculates the target specification required to meet the current load and performs the scaling-out operation:
One-step scaling: Scale out directly to the target specification based on the actual load, eliminating the need for incremental steps and avoiding the latency caused by multiple small-scale expansions.
Upper limit protection: The post-scaling specification does not exceed the maximum computing power limit set by the user.
When local resources are sufficient, it performs in-place configuration changes without interrupting connections or affecting business operations. When local resources are insufficient, it performs cross-machine configuration changes, during which connections may experience brief interruptions before automatically recovering.
Scaling-Down Policy
When CPU utilization consistently falls below the scaling-in threshold (65% by default), the system automatically scales in to an appropriate specification and releases excess resources:
Conservative scaling-in: After scaling in is performed, it ensures CPU utilization remains within a reasonable range (50%~65%) to avoid triggering an immediate scale-out.
Lower limit protection: The post-scaling-in specification does not fall below the minimum computing power limit set by the user.
Business transparency: The scaling-in process is performed through in-place adjustments, without interrupting connections, ensuring users experience no disruption.
Debouncing Mechanism
The system incorporates a built-in intelligent anti-jitter policy to prevent frequent scaling operations triggered by short-term load fluctuations:
Rapid scaling response: Upon detecting sustained high load, it scales out quickly to ensure business performance.
Slow scaling-in response: It performs scaling-in only after confirming a sustained decrease in load to avoid erroneous scaling operations.
A safety buffer is maintained between the scaling-out and scaling-in thresholds to ensure stable instance operation.
Effective Time
Vertical elasticity (CPU/memory scaling) takes effect within minutes. It is implemented by adjusting resource limits in place, without requiring node reconstruction.
Automatic disk capacity expansion
Serverless instances support automatic disk scaling:
When disk utilization exceeds the threshold, the system automatically expands the disk capacity.
The scaling-out process is transparent to the business and requires no downtime.
Disks support only automatic scaling-out and do not support scaling-in.
Use Limits
|
Horizontal Elasticity | Automatic horizontal scaling is not supported. |
Automatic Start/Stop | Automatic start/stop is not supported. The instance can run at the minimum compute capacity. |