0
votes

I have observed it takes 15 data points in 15 minutes to trigger an alarm and start scale in and I could not find an option to change or adjust this behavior.

Can we change this behavior or we will have to define another scaling policy to scale in?

It seems like its the default settings for Target Tracking scaling policy for cpu utilization that

  1. CPUUtilization > 50 for 3 datapoints within 3 minutes [will trigger high alarm]
  2. CPUUtilization < 35 for 15 datapoints within 15 minutes [will trigger low alarm]

Is this true?

1

1 Answers

3
votes

Yes, you are correct. These are default settings for the target tracking policy. However, you shouldn't be editing target tracking alarms outside of autoscaling settings.

The AWS specifically writes:

DO NOT EDIT OR DELETE.

enter image description here

If you are not happy with how target tracking policy works, you can use step or simple scaling policies instead.

For more advanced tuning of your scaling, you can use multiple scaling polices.