GCP Professional Data Engineer Practice Question

Your company streams 5 PB of user-generated HD videos and ingests 50 TB of new content each day. Viewers around the world expect video playback to begin in under two seconds. Product managers require a real-time dashboard that shows per-video engagement within 50 ms, and analysts need to run ad-hoc SQL queries on several months of usage logs. Compliance rules mandate that every uploaded video be retained for seven years at the lowest possible cost. Which Google Cloud storage architecture best meets all of these requirements?

Host videos on a Filestore Enterprise NFS share replicated across regions; store engagement metrics in Cloud SQL; process logs with Dataproc running Hadoop on regional Persistent Disk.
Persist both videos and logs as BLOB columns in Cloud Spanner with multi-region configuration; store engagement metrics in Firestore and run analytics queries directly against Spanner.
Write video binaries directly into Bigtable; cache engagement metrics in Memorystore; archive logs to Cloud Spanner tables for analytical queries.
Store videos in a multi-region Cloud Storage bucket with lifecycle rules that move objects to Archive after 90 days; keep engagement metrics in Bigtable; ingest logs into BigQuery for ad-hoc SQL analysis.

Report Issue

Answer Description

Cloud Storage is purpose-built object storage that can serve large, unstructured assets such as video files globally with high throughput. Using a multi-region bucket delivers low-latency access worldwide, while Object Lifecycle Management can automatically transition objects to cheaper Archive storage after the initial period of heavy access, satisfying long-term retention at minimal cost. Cloud Bigtable is optimized for high-throughput, sub-10 ms reads and writes, making it well-suited for serving the per-video engagement counters that power a near real-time dashboard. BigQuery is a serverless, columnar data warehouse designed for interactive SQL analytics over terabytes to petabytes of data, making it the most efficient option for analysts querying months of usage logs. The alternative designs either store large video binaries in services not optimized for object delivery, lack automated tiering to low-cost archival classes, or rely on systems that cannot guarantee the required latency or elasticity for real-time metrics and ad-hoc analytics.

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.