Question 27
Domain 4You are building an ML model to detect anomalies in real-time sensor data. You will use Pub/Sub to handle incoming requests. You want to store the results for analytics and visualization. How should you configure the pipeline?
Correct answer: C
Explanation
Pub/Sub is used to ingest the streaming sensor events, and Dataflow is the service for real-time stream processing and anomaly detection. Vertex AI hosts the ML model, and BigQuery stores the results because it is the analytics warehouse for querying and visualization.
Why each option is right or wrong
A. 1 = BigQuery, 2 = AutoML, 3 = Cloud Functions
B. 1 = BigQuery, 2 = Vertex AI, 3 = Cloud Storage
C. 1 = Dataflow, 2 = Vertex AI, 3 = BigQuery.
Pub/Sub is the ingestion layer for the sensor event stream, but the processing step must be a managed streaming engine; Dataflow is the GCP service built on Apache Beam for low-latency, event-by-event pipelines and is the correct place to run the anomaly-detection transform. Vertex AI is the managed platform for training/serving the model, and BigQuery is the correct sink for downstream analytics because it is the serverless warehouse designed for SQL querying and dashboarding over large result sets.
D. 1 = Dataproc, 2 = AutoML, 3 = Cloud Bigtable