MLA-C01 Question 299
Select 2You are designing an AWS SageMaker training job that processes highly variable workloads with fluctuating compute requirements. You plan to use Auto Scaling policies for the underlying instances to optimize cost and performance. Which scaling policy or policies should you choose to handle dynamic workloads while ensuring performance and cost efficiency?
- A
Target tracking scaling policy
- B
Step scaling policy
- C
Scheduled scaling policy
- D
Predictive scaling policy
- E
Instance refresh scaling policy
Show answer and explanation
Correct answers: A, B
Explanation
For highly variable workloads, using a combination of target tracking and step scaling policies ensures both reactive and controlled scaling. Target tracking scaling automatically adjusts resources based on a target metric, while step scaling provides fine-grained control for specific scenarios. Scheduled and predictive scaling are better suited for predictable patterns, and instance refresh is unrelated to dynamic scaling.
- A. Correct.
Target tracking scaling policy adjusts resources automatically to maintain a specific target metric, such as CPU utilization, making it ideal for dynamically fluctuating workloads.
- B. Correct.
Step scaling policy allows for predefined scaling actions based on specific threshold breaches, providing more granular control during dynamic workload changes.
- C. Incorrect.
Scheduled scaling policy is used for workloads with predictable changes at specific times, but it is not suitable for dynamic or unpredictable workloads.
- D. Incorrect.
Predictive scaling policy uses machine learning to forecast demand, which is effective for workloads with predictable patterns but less reliable for highly variable workloads.
- E. Incorrect.
Instance refresh scaling policy is used to update or refresh instances within an Auto Scaling group but is unrelated to handling dynamic workload scaling.