Question

A data engineering team runs the same set of dashboard queries every morning against a large sales table. They configure their virtual warehouse to auto-suspend after 60 seconds of inactivity to control costs. However, they notice that the first morning queries consistently run slower than subsequent ones, even though the underlying data has not changed overnight. Which characteristic of the compute layer best explains this behavior?

Accepted Answer

When a warehouse suspends, its local SSD cache is dropped, so queries after resume must re-read data from the storage layer until the cache is repopulated.

Answer

The result cache in the cloud services layer expires after 60 seconds, forcing a full recomputation each morning.

Answer

Auto-suspend permanently deletes the micro-partitions associated with the warehouse, requiring them to be regenerated on resume.

Answer

A suspended warehouse loses its clustering keys, so the first queries must re-cluster the table before returning results.