When stored data in Cloud Spanner performs poorly under load, which data configuration issue can cause this problem?

The correct answer highlights a crucial aspect of database performance in Cloud Spanner, specifically related to primary key selection. When using Version 1 UUIDs, which increase monotonically, all writes are directed to the same partition in the database. This can lead to hotspotting, where a significant amount of load is concentrated on a single node or partition rather than being distributed evenly across the system. This can severely degrade performance under load, especially when there are high write operations, as the database struggles to handle requests effectively.

In contrast, optimal primary key designs typically aim for randomness or distribution to provide balanced access across partitions. This randomness helps prevent bottlenecks and ensures that the load is spread across multiple nodes, preserving performance during peak usage times.

The other options present various concerns but do not directly lead to load performance issues caused by partitioning and distribution. Therefore, the use of monotonically increasing UUIDs as primary keys is particularly problematic for performance in a horizontally scalable database like Cloud Spanner.