AWS AutoScaling Policy on CPU Utilization on the entire group or on individual EC2

Question

We are looking for the threshold behaviours on Scaling Policies. When we setup a the thresholds like this for example

An AutoScaling group set under an ELB. And min and max instances are 4 and 8.

Average of CPU Utilization <65% for 2 consecutive 300 seconds spin down an instance.

Average of CPU Utilization >80% for 2 consecutive 300 seconds spin up an instance.

What is the behaviour here?

Is the CPU Utilization is checked over among all the instances in the group or the utilization individual instances?

Or is it the utilization of the ELB that is placed on the AutoScaling group?

Thanks in advance.

Regards, Atchut.

Answer 1

The CPU utilization calculated is Average CPU utilization for all the EC2 instances. So if there are say 5 EC2 instances attached, the Autoscaling group will check the average CPU utilization and trigger the scaling policy based on calculated average.