MLA-C01 Question 401
Select 3A team is deploying a real-time machine learning inference system on AWS to handle streaming data from IoT sensors. The system must process thousands of requests per second with minimal latency. Which combination of key performance metrics should the team prioritize to evaluate and optimize the system's performance?
- A
Throughput and latency
- B
Scalability and utilization
- C
Availability and fault tolerance
- D
Model accuracy and precision
- E
Cost optimization and resource efficiency
Show answer and explanation
Correct answers: A, B, C
Explanation
In a real-time inference system, metrics such as throughput, latency, scalability, and availability are crucial to ensure the system can handle high data volumes efficiently, respond quickly, and remain resilient to potential failures. Model accuracy and cost optimization, while relevant in broader ML workflows, are not the primary focus when evaluating ML infrastructure performance for this specific use case.
- A. Correct.
Throughput and latency are critical for real-time systems to ensure the system can handle high data volumes while maintaining low response times.
- B. Correct.
Scalability and utilization are key to ensuring the system can handle increasing workloads efficiently while making optimal use of resources.
- C. Correct.
Availability and fault tolerance are essential for a real-time system to ensure continuous operation and resilience against failures.
- D. Incorrect.
Model accuracy and precision, while important for ML model quality, are not directly related to the infrastructure's performance metrics in this scenario.
- E. Incorrect.
Cost optimization and resource efficiency are important considerations, but they are secondary to the primary performance metrics required for a real-time system.