GCP PCA · Question 20 · Domain 1: Designing and Planning a Cloud Solution Architecture

CASE STUDY: AutoMakers Inc

Company Overview:
AutoMakers Inc is a global vehicle manufacturer. They have recently launched a line of connected cars.

Current Technical Environment:

• 1 million connected cars currently on the road
• Cars send telemetry data (speed, engine temp, location) every 5 seconds
• Current on-premises MQTT brokers are crashing under the load

Business Requirements:

• Enable predictive maintenance to alert drivers before parts fail
• Provide real-time fleet tracking for commercial customers
• Support over-the-air (OTA) software updates

Executive Statements:

• CEO: "Data is our new revenue stream. We need to monetize this telemetry data."
• CTO: "We expect to have 10 million connected cars in 3 years. The architecture must scale infinitely without manual intervention."
• CFO: "The cost of ingesting and storing this data must be strictly controlled. We cannot pay for idle capacity."

Technical Requirements:

• Ingest up to 100,000 messages per second
• Low-latency processing for real-time alerts
• Time-series data storage for historical analysis
• Handle variable network connectivity (cars driving through tunnels)

Constraints:

• Strict budget for data ingestion
• Small data engineering team

QUESTION:
To fulfill the business requirement of 'predictive maintenance', the data science team needs to train machine learning models on the historical telemetry data. Which GCP service should you recommend for building, training, and deploying these models?