You need to load a large CSV file (1 TB) into a Snowflake table. Which of the following methods would generally provide the fastest loading performance?

Question

Options

• AUsing the Snowflake web interface to upload the file.
• BUsing a Python script with the Snowflake Connector to insert rows individually.
• CUsing Snowpipe with a properly sized warehouse and a file format object optimized for CSV files.
• DUsing the 'COPY INTO' command with multiple parallel streams and a properly sized warehouse
• Eare significantly slower due to network overhead and lack of parallelism.

• A
  5% (1)
• B
  10% (2)
• D
  67% (14)
• E
  19% (4)

Explanation

COPY INTO with parallel streams (D) is the fastest method because Snowflake is architected for bulk loading - it splits large files across multiple parallel micro-partitions and leverages the warehouse's compute nodes simultaneously, achieving throughput that scales with warehouse size.

• A (Web UI) is limited to small files (typically under 50 MB) and is single-threaded by design - completely impractical for 1 TB.
• B (Python row-by-row inserts) is the worst possible approach for bulk data: each INSERT is a separate transaction with round-trip latency, making 1 TB effectively impossible.
• C (Snowpipe) is optimized for continuous, micro-batch ingestion of streaming/event data, not one-shot bulk loads - it has queuing overhead and is not designed to maximize throughput for a single large file.
• E is a sentence fragment that appears to be a distractor description leaked into the choices, not a real option.

Memory tip: Think of COPY INTO as Snowflake's "bulk loader" - whenever you see a large, one-time file load, COPY INTO + a large warehouse + staged files (in S3/Azure/GCS) is the canonical answer. Snowpipe = streaming; COPY INTO = bulk.

No community discussion yet for this question.