GCP Professional Data Engineer Practice Question

A retail company needs to redesign its IoT ingestion layer. Millions of sensors worldwide will stream small JSON readings every few seconds. The solution must

accept HTTPS pushes with end-to-end latency under 100 ms,
automatically scale without capacity planning,
survive regional outages while retaining undelivered messages for up to 7 days, and
fan-out each event to several independent Dataflow streaming pipelines for further processing. Which service should you recommend as the primary ingestion point?

Upload each JSON reading to Cloud Storage via signed URLs and trigger Cloud Functions that forward data to downstream pipelines.
Post sensor readings to a regional Cloud Run service that writes directly into Cloud Spanner for later processing.
Publish sensor events to a Cloud Pub/Sub standard topic and let each Dataflow pipeline consume from its own subscription.
Deploy an Apache Kafka cluster on Cloud Dataproc to ingest events and have Dataflow read from Kafka topics.

Report Issue

Answer Description

Cloud Pub/Sub is a fully managed, global messaging service that accepts HTTPS or gRPC pushes/pulls, automatically scales to millions of messages per second, replicates data across zones for regional resilience, and guarantees at-least-once delivery with configurable retention up to 7 days. Each downstream application can consume the same events through its own subscription, enabling efficient fan-out to multiple Dataflow pipelines. Self-managed Kafka on Dataproc would require ongoing capacity and availability management. Cloud Storage uploads plus Cloud Functions add high latency and are not optimized for streaming fan-out. Writing directly to Cloud Run and Spanner provides neither built-in buffering nor easy multi-subscriber semantics. Therefore, Pub/Sub best meets all stated requirements.

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.