Databricks Machine Learning Professional Question 212
Single answerA company is deploying a machine learning model for real-time fraud detection. The model needs to handle high volumes of concurrent requests with low latency and ensure scalability. Why would using a cloud-provided RESTful service in a containerized environment be the best solution for this scenario?
- A
It simplifies deployment and scaling by leveraging container orchestration tools like Kubernetes.
- B
It eliminates the need for monitoring tools since cloud services automatically handle all operational metrics.
- C
It provides better integration with cloud-native tools like load balancers and auto-scaling features.
- D
It ensures that the model runs faster than any other deployment method available.
Show answer and explanation
Correct answer: A
Explanation
Using cloud-provided RESTful services in containers is ideal for production-grade real-time machine learning deployments because it simplifies deployment, scaling, and management by leveraging container orchestration tools. These tools ensure that the system can handle high volumes of concurrent requests efficiently while maintaining low latency. This approach is particularly suited for scenarios like real-time fraud detection where reliability and scalability are critical.
- A. Correct.
This is correct because cloud-provided RESTful services in containers simplify deployment and scaling by leveraging features like container orchestration tools, which are designed for managing production-grade environments efficiently.
- B. Incorrect.
This is incorrect because while cloud services do provide monitoring tools, it is still the responsibility of the user to set up and configure appropriate monitoring and alerting for the application.
- C. Incorrect.
This is incorrect because, while cloud-native tools like load balancers and auto-scaling are beneficial, they are not exclusive to RESTful services in containers. Other deployment methods could also integrate with these tools.
- D. Incorrect.
This is incorrect because the speed of the model depends on multiple factors, including hardware, optimization, and the model itself. The deployment method alone does not guarantee faster execution.