What is the spill sequence when a query cannot fit in memory?

• A. Spill to remote storage only
• B. Spill to local disk first, then to remote storage
• C. Spill to cache
• D. Spill to temporary tables

Answer: (B)

When a query won’t fit in memory, the system handles the overflow by spilling intermediate data to storage so processing can continue. The most efficient path is to spill to the local disk on the compute node first. This keeps the data close and minimizes latency. If even more data must be held and memory is still under pressure, the engine spills to remote storage as an overflow. This sequence—local disk first, then remote storage—lets large operations complete without exhausting memory.

Spilling to cache isn’t the standard path for handling overflow during query execution, as caches are meant for faster access of recently used data rather than providing overflow capacity. Temporary tables are user-visible objects for storing results, not the internal spill mechanism used by the engine.