Question 260
A database team wants to prevent a recurrence of this issue: An application wants semantic search over product text but has no plan for storing or indexing embedding vectors. Which preventive practice is most relevant?
Explanation
Correct Answer
Include embedding size, indexing, recall, and update behavior in database requirements. This is preventive rather than a one-time repair; it addresses the underlying failure: The selection process considers only string search and ignores vector retrieval requirements.
Incorrect Answers
Here, the selection process considers only string search and ignores vector retrieval requirements. The option concerns the separate risk of a situation in which a Bigtable replica in another cluster is queried immediately after a write and returns an older value.
For the situation in which a TimeSeries Bigtable workload uses only a leading timestamp in its row key and struggles at high ingestion rates, it makes sense to test row-key prefix entropy with expected ingestion distribution. Here, the selection process considers only string search and ignores vector retrieval requirements, so the safeguard misses the recurring fault.
The proposed check would include patching, backups, monitoring, and staffing in database TCO analyses, helping with cases where a small team installs and patches its own database on virtual machines despite having no custom engine requirements. The recurring exposure described here is that the selection process considers only string search and ignores vector retrieval requirements.