MLA-C01 Question 302
Single answerYou are managing an Amazon SageMaker endpoint for a real-time machine learning model with unpredictable traffic patterns. You need to select a scaling policy to ensure optimal cost efficiency and performance. Which scaling policy should you choose, and why?
- A
Target tracking scaling policy
- B
Step scaling policy
- C
Scheduled scaling policy
- D
Manual scaling
Show answer and explanation
Correct answer: A
Explanation
For endpoints with unpredictable traffic patterns, target tracking scaling is the most effective policy. It dynamically adjusts based on a target metric, such as latency or CPU utilization, ensuring the system performs well while remaining cost-efficient. Other policies, like step scaling or scheduled scaling, are either too rigid or suited for predictable workloads, and manual scaling is impractical for real-time adjustments.
- A. Correct.
Target tracking scaling policy is ideal for unpredictable traffic because it automatically adjusts the capacity of the endpoint to maintain a target metric, such as latency or utilization. This eliminates the need for manual intervention and ensures a balance between cost and performance.
- B. Incorrect.
Step scaling policy is better suited for scenarios where specific, predefined thresholds trigger scaling actions. It is less effective for unpredictable traffic patterns because it requires detailed configuration and does not dynamically adjust to changing patterns.
- C. Incorrect.
Scheduled scaling policy is designed for predictable traffic patterns where scaling actions can be planned ahead of time based on known usage patterns. It is not suitable for traffic that is highly unpredictable.
- D. Incorrect.
Manual scaling requires human intervention and is not a scalable or cost-efficient solution for managing unpredictable traffic patterns. It does not provide the automation required for real-time adjustment.