To optimize parallel loads, what compressed data file size is recommended by Snowflake?

• A. 1-2 MB
• B. 10-50 MB
• C. 100-250 MB (or larger) in size compressed
• D. 500-1000 MB

Answer: (C)

Maximizing parallel loads relies on having enough independent files for Snowflake to process at the same time. Each file can be loaded by a separate loader thread, so you want enough work units to keep the compute warehouse busy without creating excessive per-file overhead. The recommended compressed file size is 100–250 MB because it strikes a balance: large enough to amortize startup and I/O costs per file, yet small enough to generate many concurrent loaders. If files are too small, overhead dominates and throughput drops; if files are too large, you get fewer parallel tasks and underutilize the warehouse. So aiming for around 100–250 MB compressed per file helps maximize parallelism and overall load performance.