Question

A data engineer is building a Python application that processes data into a pandas DataFrame and needs to load approximately 5 million rows into an existing Snowflake table as efficiently as possible. They are already using the Snowflake Connector for Python. Which approach should they use to load the DataFrame with the least custom code while maximizing throughput?

Accepted Answer

Use the write_pandas() function from the Snowflake Connector for Python, which stages the data and runs COPY INTO under the hood

Answer

Iterate over the DataFrame rows and execute individual INSERT statements using cursor.execute()

Answer

Convert the DataFrame to a CSV string and pass it directly to a single INSERT ... VALUES statement

Answer

Configure a Snowpipe auto-ingest pipe and point it at the in-memory DataFrame