Order

Preserve ordering

When the dataset is read by DSS, preserve its "native" ordering (order of the rows in the files, files in enumeration order). This disables read parallelism.
When the dataset is written by DSS, preserve the "native" ordering of the stream of data. This disables write parallelism and only a single file will be written.
This only applies when DSS itself writes the dataset. For example, it does not apply for Spark.

Parallelism

Force single output file

When the dataset is written by DSS, force a single output file (regardless of the kind of input). Automatically implied if when preserving ordering.

Write bucketing

When the dataset is written by DSS from a "single-stream" source (like a SQL database), DSS can automatically "redispatch" the rows accross several files to give an opportunity to downstream recipes to better parallelize. Entering more than 1 here will enable this feature and create multiple output files. Multiple output files may still be created when running a recipe with input datasets that contain multiple files (see above)
Incompatible with preserving ordering or forcing single output file.

Output file name

Output file base name

Base name (without extension) of the output file, for single-output-file cases