A genomics company keeps raw sequencing outputs that Spark jobs, a BigQuery dashboard, and a Presto cluster must all read. Storage should avoid duplicate copies. Which approach enables shared access?
Show answer and explanation
Answer: D
Keeping data as open columnar files such as Parquet in Cloud Storage lets many engines read the same bytes. BigLake tables register those files so BigQuery can query them, and BigLake's governance layer applies fine-grained access controls. Spark and Presto can read the same files without a duplicate copy. Native BigQuery tables are stored in the warehouse's proprietary format, Cloud SQL is not built to serve as a data lake, and Bigtable is a key-value store rather than a shared analytics format.
Why the other options are wrong
- A. Native tables are stored in BigQuery's proprietary format, so Spark and Presto would have to load a separate copy to read the same data.
- B. Cloud SQL is an operational database that is not designed to hold large genomic files, and federating queries from each engine would still leave data stranded there.
- C. Bigtable is a sparse NoSQL store for key-based access. It is not an open file format that Spark and Presto can scan as a shared dataset.