Google Professional Machine Learning Engineer Question 387
Single answerGoogle Cloud PlatformYour team has developed a REST API-based machine learning model inference service. The service needs to handle unpredictable traffic patterns with potential spikes during business hours. You want to deploy this service on Google Cloud to ensure scalability, low operational overhead, and cost efficiency. Which Google Cloud service is the most appropriate for this use case?
- A
Google Cloud Run
- B
Google Kubernetes Engine (GKE)
- C
Google Compute Engine
- D
AI Platform Prediction (now Vertex AI Prediction)
Show answer and explanation
Correct answer: A
Explanation
Cloud Run is the best choice for deploying a REST API-based service that needs to handle unpredictable traffic patterns. It offers serverless execution, automatic scaling based on demand, and low operational overhead, making it ideal for this machine learning inference use case.
- A. Correct.
Correct. Cloud Run is a fully managed service that enables you to run containers in a serverless environment. It automatically scales up or down based on traffic, making it ideal for handling unpredictable traffic patterns with low operational overhead and cost efficiency.
- B. Incorrect.
Incorrect. While GKE provides container orchestration, it requires manual setup and management of clusters, which increases operational overhead. It is better suited for scenarios requiring more customizability and control.
- C. Incorrect.
Incorrect. Compute Engine involves managing virtual machine instances, which results in higher operational complexity and is not ideal for scaling unpredictable workloads efficiently.
- D. Incorrect.
Incorrect. Vertex AI Prediction is designed specifically for serving machine learning models but doesn't provide general-purpose REST API hosting. It is better suited for serving models trained on Vertex AI or TensorFlow.