AWS DevOps Engineer - Professional

By Amazon · Question Mode

QuestionQ45

Monitoring and Logging

A company runs an application on current-generation Amazon EC2 instances in a VPC. The EC2 instances run Amazon Linux and launch in an Amazon EC2 Auto Scaling group. The application retrieves data from one Amazon S3 bucket, processes it, and uploads the processed data to another S3 bucket.

Recently, the application's performance degraded. A manual investigation found that outbound network bandwidth utilization was too high for the EC2 instance type. The company updated the EC2 instances to a larger instance size.

The company's DevOps team must receive a notification from an Amazon CloudWatch alarm if the application tries to use more outbound network bandwidth than the EC2 instances provide.

Which solution meets these requirements?

Explanation

On Linux EC2 instances using ENA, bw_out_allowance_exceeded records packets that are queued or dropped because outbound aggregate bandwidth exceeds the instance maximum. The unified CloudWatch agent, using its ethtool collection, publishes this metric to CloudWatch (with an ethtool_ prefix in CloudWatch). An Auto Scaling group creates instances dynamically, so creating a metric alarm for each instance provides per-instance notification coverage. Composite alarms only evaluate the states of other alarms; they do not directly aggregate all matching instance metrics.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!