Docs

Auto Scaling

Product Introduction

Concepts

Concepts

When using the Auto Scaling (AS) service, you may encounter the following terms. This topic describes common terms.

Term	Description
Scaling group	A scaling group is used to manage a group of instances for the same business scenario. The configuration of a scaling group is predefined, including the VPC to which the instances belong, and the maximum and minimum numbers of instances.
Scaling configuration source	Used to define the instance configurations for scaling groups, including those related to compute, memory, network, storage, and security. When the scaling group scales up, new instances are created based on the preset instance configuration.
Scaling rule	Defines the conditions for adjustment jobs: the timing and method for triggering scaling-up or scaling-down operations.
Adjustment type	Indicates the type of condition that triggers an adjustment, such as scheduled scaling, periodic scaling, or manual scaling.
Adjustment	Indicates the number of instances to be modified when the trigger conditions are met, for example, increasing the instance count, reducing it, or setting it to a fixed number.
Instance in a scaling group	An instance refers to an actually available cloud resource within a scaling group, such as an ECS instance.
Expected number of instances	Used to limit the number of instances actually running in the scaling group to a value between the minimum and maximum instance counts.
Instance protection	Protected instances are not deleted or moved out during adjustments and health checks.
Instance health check	Determines whether an instance is running normally and whether related business is available. This prevents an instance with exceptions from affecting normal business access, improving overall service availability.
Scaling log	Records the results after the scaling group executes adjustment jobs.
Lifecycle hook	After a scaling rule of the scaling group and the health check are triggered, the lifecycle hook takes effect, and target instances enter the Adding/removal suspended state. For instances in waiting states, you can perform custom actions until the lifecycle hook times out, or you manually call back to complete the wait.
Vertical scaling	Vertical scaling refers to adjusting the instance type configuration. This feature allows you to scale up or down the configurations (including vCPU and memory) of pay-as-you-go instances in a scaling group.
Rolling upgrade	Automatically updates instances in a scaling group in batches, based on instance image updates, custom script executions within instances, or other methods. This feature offers one-click automatic upgrades for multiple instances. Currently, the rolling upgrade feature is released only in China (Beijing), Asia Pacific (Johor), and China (Shanghai), and is in Beta in other regions. If you need a trial, please contact your account manager to apply.

Last updated: 2026.06.16 15:56:42