Overview
In a scaling group, you can specify an instance that is not scaled-in during a scaling-in activity. When a scaling-in activity occurs, AS selects a sub-machine to be scaled-in from other machines.
You can enable Instance Protection for one or more scaling group instances. You can change the scaling group or instance protection settings at any time.
If all remaining instances in the scaling group are protected by scaling-in protection and a scaling-in event occurs at the same time, AS reduces the required capacity instead of removing instances.
Applicable Scenario
Normally, the machines in a scaling group are stateless, and all machines can be removed at any time. However, in actual practice, the following situations apply to exempting a specific instance from being scaled-in:
- One Machine for Multiple Uses: Based on cost considerations, some machines, besides doing things in the cluster, also have other purposes, such as storing data generated in the cluster, so this machine is actually stateful.
- Avoid Misoperation: If you are worried that incorrect policy settings affect your business, you can set some machines to be exempt from being scaled-in. In this way, AS will never scale-in the machine, and the request-LB-instance path can remain unobstructed.
How to Set Up?
In the instance list of the scaling group, you can directly set: