The correct answer is B because bytes spilling to external storage is a clear indicator that a warehouse is undersized. When a Snowflake warehouse doesn't have sufficient memory to process queries in memory, it spills data to external storage, which significantly impacts performance. This spilling occurs because the warehouse lacks the memory capacity to handle the query workload efficiently. The community discussion shows 100% consensus on answer B, with references to Snowflake documentation about performance impacts from disk spilling. Other options are less suitable: A (filter nodes) relates to query complexity, not warehouse sizing; C (high processed rows) could indicate efficient processing; D (partitions scanned vs total) relates to data pruning efficiency, not warehouse capacity.