Unloading Data

This section explains how to unload tables and query results, using the ybunload client tool. The ybunload client is a high-performance parallel export tool that unloads data to files on the client system. The client connects to the manager node, then distributes the work of reading and writing the data to the compute blades. Data can be returned to the client in compressed (GZIP) or uncompressed files. The unloaded data can be streamed out to multiple output files or combined into a single file, depending on your requirements.

Download and install the unload client (as part of the ybtools package) from Yellowbrick Manager or the Customer Support download site. Then run the Linux and Windows executable programs:

  • ybunload on Linux
  • ybunload.exe on Windows
Note: Yellowbrick recommends that you upgrade to the latest version of the ybtools clients when you upgrade the cluster so that your client and server versions correspond. The client tools are backward-compatible but not always forward-compatible.
The following table shows which formats and storage targets are supported for ybunload operations:
Supported Formats Supported Storage Targets Notes
csv Local file system csv is the default unload format.
S3 or S3-compatible object storage
text Local file system
S3 or S3-compatible object storage
parquet Local file system
S3 or S3-compatible object storage
Azure Blob storage, Azure Data Lake Storage Gen2 Data unloaded to Azure object storage must be unloaded in parquet format.