Learn how to quickly export data to Parquet files from SSIS using the ODBC Parquet (DuckDB) Connector. This solution makes it easy to generate high-performance, analytics-ready Parquet files directly from SQL queries or CSV files, with minimal setup and maximum efficiency.
This article explains how to write data to Parquet files using DuckDB’s COPY command, covering both direct data export and conversion from CSV files.
Prerequisites
- Download and install the ZappySys ODBC PowerPack.
- Download the DuckDB JAR file from the Parquet (DuckDB) website or the Maven MVN Repository.
- Access to either:
- SQL Server (via linked server or query execution), or
- SSIS (using Execute SQL Task or equivalent)
Writing Parquet Files Using DuckDB
DuckDB supports the COPY command, which allows exporting query results directly to Parquet format.
Scenario 1: Writing Direct Data to a Parquet File
Use this approach when you want to export query results or generated data.
Example:
COPY
(
SELECT
1 AS ID,
'John doe' AS Name
)
Scenario 2: Load a CSV File data to Parquet File
DuckDB can read CSV files directly and write the output to Parquet in a single step.
Example:
COPY
(
SELECT *
FROM read_csv('D:\CsvFiles\Countries.csv')
)
TO 'D:\ParquetFiles\Countries.parquet' (FORMAT parquet);
Execution Options
You can execute the above queries using either of the following approaches:
From SSIS
Use an Execute SQL Task in SSIS with the ZappySys ODBC–JDBC Bridge connection to run the DuckDB query as part of your ETL workflow. For detailed, step-by-step configuration instructions, refer to the Parquet (DuckDB) Connector for SSIS documentation:
From SQL Server
Execute the query directly through SQL Server using the configured DuckDB connector, enabling Parquet file generation from within SQL-based processes. For step-by-step instructions, refer to the Parquet (DuckDB) Connector for SQL Server documentation below:
Conclusion
Using the ODBC Parquet (DuckDB) Connector with ZappySys, you can easily export data to Parquet files from SSIS or SQL Server using simple SQL commands. This approach provides a fast, reliable, and scalable way to generate analytics-ready Parquet files, streamline ETL workflows, and integrate modern columnar storage into your data pipelines with minimal configuration.