GCP Professional Data Engineer Practice Question

Cloud Bigtable is designed for petabyte-scale, time-series and IoT workloads that need sustained hundreds of thousands of writes per second and single-digit millisecond reads based on a row key such as session ID. It automatically splits tablets as the table grows, so no manual sharding is required. BigQuery can ingest streaming data but optimizes for analytics over large datasets; individual record access typically takes seconds, not milliseconds. Cloud SQL offers familiar relational semantics but scales vertically and cannot handle the required throughput or data volume without complex sharding. Cloud Storage is cost-effective for lake storage but is an object store; retrieving individual sessions would involve listing and reading files, resulting in latencies far beyond the sub-10 ms requirement.