MLA-C01 Question 399
Select 4You are designing an ML inference pipeline using Amazon SageMaker for a high-traffic application. The application requires low latency, high availability, and the ability to handle unpredictable spikes in traffic. Which combination of metrics should you monitor and optimize to ensure the system meets these requirements?
- A
Latency of the endpoint
- B
CPU and GPU utilization of the hosting instances
- C
Throughput of requests per second processed by the endpoint
- D
Scalability of the endpoint to automatically scale instances
- E
Storage utilization of the hosting instances
Show answer and explanation
Correct answers: A, B, C, D
Explanation
To meet the requirements of low latency, high availability, and the ability to handle traffic spikes, it is essential to monitor latency, utilization, throughput, and scalability. These metrics collectively ensure that the system can provide fast responses, handle high traffic efficiently, and scale dynamically when needed. Storage utilization is less relevant in this scenario as it does not directly impact these key performance factors.
- A. Correct.
Monitoring latency is critical for ensuring that the application delivers low response times, which is especially important for high-traffic, real-time applications.
- B. Correct.
CPU and GPU utilization should be monitored to ensure the hosting instances are efficiently utilized without being over-provisioned or under-provisioned.
- C. Correct.
Throughput is a key metric for understanding how well the system handles the volume of requests, especially during traffic spikes.
- D. Correct.
Scalability ensures the system can handle unpredictable spikes in traffic by dynamically adding or removing instances as needed.
- E. Incorrect.
While storage utilization can be useful in some cases, it is not a primary concern for ensuring low latency, high availability, or handling traffic spikes in an inference pipeline.