GCP Professional Data Engineer Practice Question

A health-monitoring startup receives a global stream of 300 000 heart-rate readings every second, amounting to about 90 MB/s of incoming data. The platform must continuously ingest this stream, allow patient dashboards to retrieve the most recent reading for any individual with a 95th-percentile latency below 10 ms, and retain 30 days of history so that a nightly batch job can scan tens of billions of rows for population-level analytics. Which Google Cloud storage approach best satisfies these requirements while minimizing operational overhead?

Insert all readings into Cloud SQL (PostgreSQL) with high-availability replicas for dashboards and export CSV files to Cloud Storage for batch analytics.
Persist readings in Cloud Bigtable for ingestion and low-latency lookups, then export daily snapshots to BigQuery for analytical scans.
Write JSON files to Cloud Storage with object versioning; dashboards fetch objects via signed URLs, and Dataflow reads them nightly for analytics in BigQuery.
Stream records directly into BigQuery and serve both dashboards and analytics from the same table.

Report Issue

Answer Description

Cloud Bigtable is designed for very high write throughput-scaling to millions of rows per second-and supports single-digit-millisecond point reads, making it well suited to time-series heart-rate telemetry. Using a row-key schema that combines patient ID with event timestamp allows dashboards to fetch the latest reading within the required latency. Because Bigtable automatically handles sharding and replication, operational overhead stays low. A daily export (via Dataflow or an external table) to BigQuery lets analysts run cost-effective SQL scans over 30 days of history.

The alternative approaches do not meet all requirements:

BigQuery alone can ingest streaming data but interactive point lookups typically take seconds rather than milliseconds.
Cloud Storage offers economical retention but cannot deliver sub-10 ms reads for individual records.
Cloud SQL would need extensive sharding and still struggle to sustain hundreds of thousands of writes per second while keeping latency in the single-millisecond range.

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.