GCP Professional Cloud Architect Practice Question

MedDeviceCo streams 5 TB of patient telemetry data daily from thousands of devices. The solution must (a) ingest the high-throughput stream reliably, (b) surface aggregated metrics to clinicians within five seconds, (c) archive all raw events unchanged for seven years at the lowest feasible cost while keeping data encrypted in transit and at rest, and (d) support a weekly machine-learning training job on the full historical data set with minimal operational effort. Which Google Cloud design best satisfies these functional and non-functional requirements?

Send device data through an external HTTP(S) Load Balancer to GKE pods that persist events in Firestore; export collections nightly to Cloud Storage Standard and run weekly Dataproc Spark jobs for model training.
Publish messages to Cloud Pub/Sub, use Dataflow streaming to write aggregates to BigQuery while archiving raw data to a Cloud Storage Archive bucket, and run weekly BigQuery ML training jobs on the stored data.
Invoke Cloud Functions from devices to insert records into Cloud SQL; export daily SQL dumps to Filestore for seven-year retention and train models in Vertex AI using imported CSV files.
Ingest data through Cloud IoT Core directly into Cloud Bigtable replicated across regions; query it from BigQuery using federated connectors and train models in AI Platform on exported snapshots.

Report Issue

Answer Description

Cloud Pub/Sub provides globally distributed, highly durable ingestion that scales to millions of messages per second without manual sharding. A streaming Dataflow pipeline can branch the stream, writing near-real-time aggregates to BigQuery and archiving the unchanged raw events to a Cloud Storage Archive bucket, the lowest-cost tier for long-term retention. Dashboards can query the streamed aggregates in BigQuery within seconds, and BigQuery ML can train models directly on the data without export, satisfying the weekly retraining requirement. All chosen services encrypt data in transit and at rest and are HIPAA-eligible. The alternative designs either cannot meet the ingest scale and latency requirements (Cloud SQL, Firestore), rely on a deprecated service (IoT Core), or use a more expensive storage class, so they fail one or more stated constraints.

Ask Bash

Bash is our AI bot, trained to help you pass your exam. AI Generated Content may display inaccurate information, always double-check anything important.