GCP Professional Data Engineer Practice Question

Your e-commerce platform lets sellers upload high-resolution product images averaging 2 MB each. Upload traffic can spike to 40 000 images per minute during holiday sales, while buyers worldwide expect image loads in under 100 ms. Images are rarely modified after upload, but marketing wants to recover accidentally overwritten versions and automatically move images older than a year to cheaper storage. Without provisioning or managing servers, which Google Cloud design best satisfies these requirements?

Store images in a multi-region Cloud Storage bucket with Object Versioning enabled and lifecycle rules that move noncurrent versions to Nearline; serve the bucket through Cloud CDN.
Insert each image as a BYTEA column in Cloud SQL for PostgreSQL, create cross-region read replicas, and cache image URLs using Cloud CDN.
Save each image as a binary field in a Firestore document and trigger Cloud Functions to move documents older than one year to Coldline Storage.
Write Base64-encoded image chunks to Cloud Bigtable rows partitioned by image ID and expose them to users behind an external HTTP(S) load balancer.

Report Issue

Answer Description

Cloud Storage is optimized for unstructured objects and automatically scales to thousands of writes per second-enough to handle 40 000 images per minute-without server management. Placing Cloud CDN in front of a Cloud Storage bucket provides edge caching worldwide, helping meet the sub-100 ms latency goal. Using a multi-region bucket replicates data across multiple Google Cloud regions within a continent, improving availability and regional read latency, while Cloud CDN handles global delivery. Enabling Object Versioning preserves previous copies of overwritten files, and lifecycle rules can transition noncurrent or aged objects to Nearline or other lower-cost classes automatically. Relational databases like Cloud SQL are not designed for multi-megabyte objects at high write rates, Bigtable is tuned for wide-column key/value workloads rather than large binaries, and Firestore has a 1 MiB document size limit, so these alternatives cannot meet the throughput, size, or cost requirements as effectively.

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.