ExpertThis question is part of a case study — click to read the full scenario(Case 16)
CASE STUDY: AutoIoT
Overview: Connected car manufacturer. 1M vehicles sending telemetry every 5 seconds.
Business: Predictive maintenance alerts, real-time fleet tracking, monetize anonymized data.
Executives:
- CEO: "Leverage AI to predict failures."
- CTO: "Current MQTT brokers crashing. Need fully managed, scalable ingestion."
- DPO: "Vehicle location is sensitive. Strip PII before analytics."
Tech: Ingest millions of msgs/sec, real-time stream processing for anomalies, store raw data for ML, sub-second queries for dashboards.
Constraints: Vehicles lose connection and send late batch data. ML models updated weekly. Strict analytics budget.
Which architecture should you design for the data ingestion and processing layer to replace the crashing MQTT brokers?
GCP PCA · Question 17 · Domain 2: Managing and Provisioning a Solution Infrastructure
CASE STUDY: AutoIoT
Overview: Connected car manufacturer. 1M vehicles sending telemetry every 5 seconds.
Business: Predictive maintenance alerts, real-time fleet tracking, monetize anonymized data.
Executives:
- CEO: "Leverage AI to predict failures."
- CTO: "Current MQTT brokers crashing. Need fully managed, scalable ingestion."
- DPO: "Vehicle location is sensitive. Strip PII before analytics."
Tech: Ingest millions of msgs/sec, real-time stream processing for anomalies, store raw data for ML, sub-second queries for dashboards.
Constraints: Vehicles lose connection and send late batch data. ML models updated weekly. Strict analytics budget.
How should you handle the constraint where vehicles lose connection and send late batch data?
CASE STUDY: AutoIoT
Overview: Connected car manufacturer. 1M vehicles sending telemetry every 5 seconds.
Business: Predictive maintenance alerts, real-time fleet tracking, monetize anonymized data.
Executives:
- CEO: "Leverage AI to predict failures."
- CTO: "Current MQTT brokers crashing. Need fully managed, scalable ingestion."
- DPO: "Vehicle location is sensitive. Strip PII before analytics."
Tech: Ingest millions of msgs/sec, real-time stream processing for anomalies, store raw data for ML, sub-second queries for dashboards.
Constraints: Vehicles lose connection and send late batch data. ML models updated weekly. Strict analytics budget.
How should you handle the constraint where vehicles lose connection and send late batch data?
Answer options:
Use processing-time windowing in Dataflow to ensure all data is processed immediately upon arrival.
Use Cloud Dataflow with event-time windowing and configure allowed lateness to process delayed messages.
Drop the late data at the Pub/Sub layer to maintain real-time dashboard performance.
Store all data in Cloud Storage first, then run a daily batch job to sort the timestamps.
How to approach this question
Full Answer
Common mistakes
Practice the full GCP Professional Cloud Architect Practice Exam 4
50 questions · hints · full answers · grading