DOP-C02 Question 171
Select 3You are managing a containerized application running on Amazon ECS with an Application Load Balancer (ALB). Your task is to configure auto-scaling for the ECS service based on appropriate metrics to ensure optimal performance and cost-efficiency. Which metrics should you use to configure the scaling policy?
- A
Average CPU utilization of ECS tasks
- B
Memory utilization of ECS tasks
- C
Number of ALB active connections
- D
ALB request count per target
- E
Disk I/O utilization of ECS tasks
Show answer and explanation
Correct answers: A, B, D
Explanation
The appropriate metrics for scaling ECS services should directly correlate with the workload on the service and the application's ability to handle traffic. CPU and memory utilization are essential for understanding resource usage, while ALB request count per target is critical for scaling based on traffic patterns. These metrics ensure that the application remains performant and cost-efficient.
- A. Correct.
CPU utilization is a critical metric for scaling ECS services as it directly reflects the workload on the containers. High CPU utilization can indicate the need for more container instances.
- B. Correct.
Memory utilization is another fundamental metric for scaling ECS services, particularly for memory-intensive applications. Scaling based on memory ensures that the application does not run out of resources.
- C. Incorrect.
Although the number of ALB active connections can be useful for monitoring, it is not typically used to trigger scaling in ECS services. It does not provide direct insight into ECS task performance.
- D. Correct.
The request count per target is a valid metric for scaling, as it reflects the load on individual ECS tasks behind the ALB. Scaling based on this metric helps maintain responsiveness during traffic spikes.
- E. Incorrect.
Disk I/O utilization is generally not a primary metric for scaling ECS services because containerized applications often focus on CPU, memory, and request-based metrics. Disk I/O might be more relevant for specific workloads like databases.