The cooldown period refers to the locking duration after one scaling activity (the addition or removal of CVMs) is executed in a scaling group. During which, the scaling group performs no scaling activities. You can optionally set the cooldown period to a value less than 999,999 seconds.
No. The Auto Scaling feature is completely free of charge. Please feel free to use it.
CVM instances that are automatically created through auto scaling are billed the same as pay-as-you-go CVMs. Manually added CVM instances are charged by their original billing policy, and immune to scaling activities.
One scaling group can have a maximum of 2,000 CVMs. For more information about the current quota of pay-as-you-go CVMs under an account in each availability zone, see “Purchase Limits for Pay-as-You-Go CVM Instances” under Purchase Limits. If you want to have more than 2,000 CVMs in one scaling group, or if you want to use more pay-as-you-go CVMs than the fixed quota, submit a ticket.
Applications that are deployed in the CVM instances in the scaling group should be stateless and replaceable. Because instances in the scaling group are likely to be repossessed by Tencent Cloud during scale-in, CVM instances for Auto Scaling cannot save the status information (such as sessions) and data (such as databases and logs) of the applications. If you need to save the status information in your application, you can consider saving it in a standalone CVM outside the scaling group.
A CVM that is manually added to the scaling group must meet the following requirements:
A CLB instance (in the case of a cross-region CLB instance, its backend instance VPC) that is associated with a scaling group must be in the same network environment (VPC or the classic network of the same region) as the scaling group.
No. Auto Scaling is a management service that elastically adjusts computing resources according to your business needs and policies. It can automatically add CVM instances when the business grows and remove CVM instances when the business declines. Currently, Auto Scaling does not support "scaling up", that is, Auto Scaling cannot automatically change the CPU, memory and bandwidth of CVMs.
No. Auto Scaling can independently scale CVM instances out and in. It can work with CLB or work alone.
You can configure two scheduled actions: one defines the desired capacity of scale-out, and the other defines the desired capacity of scale-in.
Tencent Cloud Auto Scaling provides two removal policies:
Take the maximum value as an example. It collects one sample per minute from each CVM in the period. If the sample value meets the configured rules for multiple periods consecutively (the number of periods can be customized), the alarm scaling action will be triggered.
For example: there are five CVMs in a scaling group, and the defined alarm scaling policy is “CPU utilization exceeds 50% for 3 consecutive periods”. The system takes one value per minute from each CVM, that is, it takes 25 CPU utilization values within one period (5 minutes). If there are values exceeding 50% for 3 consecutive periods, the scaling rule is satisfied and scaling actions are triggered.
The desired capacity is the reasonable number of instances in the current scaling group. Its value should be between the minimum and maximum capacity. You can manually adjust the desired capacity, and you can also use scheduled action and alarm scaling task to trigger adjustment. The scaling group automatically adjusts the actual number of instances to match the desired capacity.
If the launch configuration specifies a data disk snapshot, you must ensure that the data disk can be automatically mounted to ensure proper auto scale-out. Please configure the source instance of the snapshot as instructed below:
Mounting Cloud Disks.
After the scaling group is disabled, automatically triggered actions will not proceed, but restrictions on the scaling group are still effective.
After the scaling group is disabled, the following automatic actions will not proceed:
To ensure the normal running of the business, manual scaling is still available, with the following restrictions:
mincapacity, the removal will not be allowed.
maxcapacity, the addition will not be allowed.
maxcapacity manually will not trigger a scaling action.