Google Professional Data Engineer Question 41
Single answerGoogle Cloud PlatformYour company is planning to build a machine learning-based recommendation system that is expected to serve millions of users daily. The company’s business requirement is to ensure low latency for recommendations, scalability to handle future growth, and cost efficiency. As a Data Engineer, how should you map these business requirements to an appropriate Google Cloud architecture?
- A
Use BigQuery for real-time data processing and predictions, as it is highly scalable and cost-effective.
- B
Deploy a Vertex AI model and use AI Platform Prediction for serving recommendations with autoscaling enabled.
- C
Store all data in Google Cloud Storage and use Cloud Functions to process and serve recommendations on demand.
- D
Use Firebase Realtime Database for storing all data and processing recommendations directly within the database.
Show answer and explanation
Correct answer: B
Explanation
The business requirements call for a scalable, low-latency, and cost-effective solution for serving machine learning-based recommendations. Vertex AI and AI Platform Prediction are tailored for deploying and serving machine learning models efficiently while offering autoscaling capabilities to manage future growth. This makes it the most appropriate choice for this scenario.
- A. Incorrect.
BigQuery is a great tool for analytics and batch processing, but it is not suitable for real-time predictions due to its higher query latency.
- B. Correct.
Vertex AI and AI Platform Prediction are specifically designed for machine learning model deployment and serving real-time predictions. Autoscaling ensures that the system can handle spikes in traffic, making it a suitable choice for scalability, low latency, and future growth.
- C. Incorrect.
Cloud Functions are good for event-driven processing but are not ideal for serving real-time recommendations at scale. They may not meet the low-latency requirements for machine learning-based applications.
- D. Incorrect.
Firebase Realtime Database is optimized for real-time data syncing but is not a suitable choice for machine learning predictions or processing at scale.