GCP Associate Cloud Engineer Practice Question 54: Bigtable IOT Timeseries

Google Cloud Associate Cloud Engineer · Test 6 · question 4 of 10

Databasesmedium

A logistics company operates several hundred thousand vehicle sensors that each publish a telemetry reading every second. The company needs a managed database that can absorb this constant stream of single-key writes with consistently low millisecond latency, scale linearly to billions of rows as the fleet grows, and support fast range scans of a sensor's history by timestamp for downstream analytics. Which database should the company use?

Select one answer
Show answer and explanation

Answer:

  • D. Cloud Bigtable, because its wide-column NoSQL design delivers high-throughput, low-latency writes at massive scale with rows naturally sorted for time-series range scans

Cloud Bigtable is purpose-built for exactly this profile: high-throughput, single-digit millisecond writes and reads on key-value data that scales linearly by adding nodes to a cluster, up to billions of rows and petabytes of data. Google's own documentation lists IoT telemetry, such as sensor and meter readings, and time-series data, such as per-device metrics over time, as canonical Bigtable use cases, because a row key that starts with the device ID followed by a reversed or bucketed timestamp keeps each sensor's history physically sorted and contiguous, making range scans fast.

Cloud SQL for PostgreSQL is a relational engine that scales vertically on a single instance; it cannot sustain hundreds of thousands of concurrent single-row writes per second at low latency without extensive sharding, and its strong transactional guarantees add overhead this workload does not need. Firestore is a document database tuned for mobile and web app data with automatic per-field indexing, but that same indexing adds write overhead and its throughput model is not designed for continuous massive-scale sensor ingestion. BigQuery is a data warehouse optimized for large batch and ad hoc analytical queries over already-loaded data; streaming inserts into BigQuery have materially higher latency than Bigtable writes and it is not meant to serve as the low-latency system of record for a live ingest pipeline.

Reference: Cloud Bigtable overview

Practice Test 6 ›

Study guides for Google Cloud Associate Cloud Engineer