--- title: "Parquet Integration Unit" slug: "parquet-integration-unit" status: "new" updated: 2026-10-27T09:03:09Z published: 2026-10-27T09:03:09Z canonical: "docs.lobstersoftware.com/parquet-integration-unit" --- > ## Documentation Index > Fetch the complete documentation index at: https://docs.lobstersoftware.com/llms.txt > Use this file to discover all available pages before exploring further. # Parquet Integration Unit ## Description The ParquetIntegrationUnit generates a Parquet file from the Target tree. Nested structures such as objects, lists, and key-value maps are preserved and are not collapsed. For large data sets, the output is written to a temporary file rather than being held entirely in main memory. ## Prerequisites In Phase 5, the option **Use IntegrationUnit & postexecution** is enabled. > [!NOTE] > **NOTE:** > > The DuckDB library is required to process Parquet. It must be included in the installation; however, you do not need to create a separate database connection for this. ## Parameter The parameter values are entered as free text. A value must be specified for all parameters. | Parameter | Permitted values | Default value | Description | | --- | --- | --- | --- | | `Start at node` | — | — | The node in the target structure from which the Parquet file is generated. | | `compression` | snappy, gzip, lzo, brotli, zstd, lz4_raw | snappy | Compression method for the generated Parquet file. | | `parquet version` | v1, v2 | v1 | Parquet version to be used. | ## Error handling and logging If the export cannot be completed, an error is displayed with a descriptive message and logged with sufficient details. Silent errors do not occur. ## Procedure 1. Create the target structure. 2. Select **ParquetIntegrationUnit** as the Integration Unit and set the **Start at node** parameter. Adjust the compression and Parquet version as needed. 3. In a Response type "File", set the "Content" to **IU output**. Structured representation of the input data in Phase 3. The target tree receives the values transferred via mapping and passes them on to Phase 4, Phase 5, or Phase 6. Integration Units: Optional downstream transformations, such as conversion to XML format. The response determines how a profile sends processed data. In Phase 6, a response determines how and where the output data is sent or stored, e.g., via FTP, HTTP or e-mail. A profile can use several responses in parallel. ## Related - [Phase 5](/profiles--entry-tab-5.md) - [Content settings](/phase-6-response-content-settings.md) - [Parquet](/profiles--tab-3-format-parquet.md) - [Analyze file for structure](/dialogue-analyse-file.md)