Question 770
Engineers believe they have resolved this failure: A Spanner table uses increasing integer identifiers and write throughput collapses as a hot key range forms. Which follow-up evidence is most convincing?
Explanation
Correct Answer
Writes distribute across splits and application latency returns within bounds. This distinguishes actual recovery from a merely healthy-looking console. The targeted correction was: Redesign supported primary-key ordering or use an appropriate distributed key component.
Incorrect Answers
The result would be meaningful for a problem where a Bigtable schema stores all meter readings under a single row key and update contention rises as ingestion expands. For this question, closure depends on confirming that writes distribute across splits and application latency returns within bounds.
Even if this observation were recorded, it would support a case where a multinational application needs continued transactional availability after one region fails but its Spanner database was deployed regionally. It would not verify that writes distribute across splits and application latency returns within bounds.
That result would matter when Cloud SQL for MySQL is configured with public addressing while the application design specifies database access only through a private VPC. It does not establish that writes distribute across splits and application latency returns within bounds, which is the evidence needed here.