GCP Professional Data Engineer Practice Question

A streaming pipeline ingests 25 000 sensor events per second and must keep the last 24 hours of data available to a mobile app with single-digit millisecond read latency. The app always fetches records by (device_id, event_timestamp) and runs no ad-hoc analytic SQL. Data volume will grow to tens of terabytes within a year, and the team wants a fully managed store that can add or remove nodes to scale writes without downtime. Which sink should you implement as the serving layer?

Cloud Spanner
BigQuery
Cloud Bigtable
Cloud SQL for PostgreSQL

Report Issue

Answer Description

The workload is write-heavy, time-series oriented, and requires low-millisecond point reads using a composite key. Cloud Bigtable is a managed wide-column NoSQL database that scales horizontally to millions of writes per second, stores petabytes, and delivers sub-10 ms latency when rows are accessed by row key. It can add or remove nodes without downtime. BigQuery is optimized for analytical SQL and cannot guarantee single-digit millisecond reads. Cloud Spanner offers strong consistency and SQL features, but its per-row latency is generally higher and the schema flexibility of Bigtable better fits time-series data. Cloud SQL cannot scale to tens of terabytes or 25 000 writes per second without sharding and operational overhead. Therefore, Cloud Bigtable is the appropriate sink.

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.