SOA-C02 Question 19
Single answerYou are a SysOps Administrator managing an application running on Amazon EC2 instances. You want to be notified if the average CPU utilization across all instances in an Auto Scaling group exceeds 80% for 5 minutes. How should you configure an Amazon CloudWatch alarm to achieve this?
- A
Create an alarm on the Auto Scaling group’s CPU utilization metric, set the threshold to 80%, and the evaluation period to 5 minutes.
- B
Create an alarm on the Auto Scaling group’s average CPU utilization metric, set the threshold to 80%, and the evaluation period to 1 out of 5 data points.
- C
Create an alarm on the Auto Scaling group’s average CPU utilization metric, set the threshold to 80%, and the evaluation period to 5 out of 5 data points.
- D
Create a separate alarm for each EC2 instance in the Auto Scaling group monitoring CPU utilization, set the threshold to 80%, and the evaluation period to 5 minutes.
Show answer and explanation
Correct answer: C
Explanation
To monitor the average CPU utilization across an Auto Scaling group, you must use the 'Average' statistic for the group’s CPU utilization metric. Setting the evaluation period to 5 out of 5 data points ensures the condition is consistently evaluated over 5 minutes. Configuring alarms for individual EC2 instances or incorrect evaluation periods does not meet the requirement to monitor the group's overall resource usage.
- A. Incorrect.
This option is incorrect because the alarm must be configured on the average CPU utilization metric, not just the CPU utilization metric.
- B. Incorrect.
This option is incorrect because even though it uses the average CPU utilization metric, the evaluation period of 1 out of 5 data points does not ensure the CPU utilization is consistently above 80% for 5 minutes.
- C. Correct.
This option is correct because it uses the average CPU utilization metric, sets the threshold to 80%, and ensures that all 5 data points are evaluated to confirm the condition holds for 5 minutes.
- D. Incorrect.
This option is incorrect because creating separate alarms per instance is inefficient and does not provide a consolidated view of the average CPU utilization across the Auto Scaling group.