MLA-C01 Question 300
Single answerYou are deploying a machine learning model on AWS SageMaker and need to ensure that the infrastructure automatically scales based on workload demand. You are considering two scaling policies for the endpoint: Target tracking scaling and Step scaling. Under which circumstances would you choose Target tracking scaling over Step scaling?
- A
When you want to maintain a specific target metric, such as CPU utilization, at a predefined level.
- B
When you need to scale up or down in response to sudden, unpredictable spikes in traffic.
- C
When you want granular control over the scaling behavior with precise step adjustments.
- D
When your workload has consistent and predictable traffic patterns.
Show answer and explanation
Correct answer: A
Explanation
Target tracking scaling is ideal when maintaining a specific target metric, such as CPU or memory utilization, is critical. It automatically adjusts resources to keep the specified metric at the desired target. Step scaling, on the other hand, is better suited for workloads that require predefined steps for scaling based on thresholds, such as sudden traffic spikes. Understanding the use case and workload characteristics is key to selecting the right scaling policy.
- A. Correct.
Correct: Target tracking scaling is preferred when you need to maintain a specific target metric, such as CPU utilization, at a defined level. It automatically adjusts capacity to keep the metric close to the target.
- B. Incorrect.
Incorrect: Target tracking scaling is not designed to respond to sudden spikes specifically; Step scaling is better suited for such scenarios as it uses predefined step adjustments.
- C. Incorrect.
Incorrect: Step scaling, not Target tracking scaling, provides granular control over scaling behavior through precise step adjustments.
- D. Incorrect.
Incorrect: While consistent and predictable traffic can work with Target tracking, it is not a unique reason to choose this policy over Step scaling.