To enable parallel processing in a custom transformer, what must be published?

• A. Group By Parameter
• B. Output Port
• C. Reader
• D. WorkspaceRunner

Answer: (A)

Parallel processing inside a custom transformer relies on splitting the incoming features into independent groups that can be processed at the same time. This splitting is controlled by publishing a Group By Parameter. By exposing this parameter, you let users choose a grouping key (for example, by region or another attribute), which the transformer uses to create work groups. The engine then runs different groups on separate threads, enabling true parallel execution. The other items listed don’t control how work is partitioned inside the transformer: an output port just forwards data, a reader brings data into the workspace, and a WorkspaceRunner runs another workspace. So, publishing the Group By Parameter is what enables parallel processing.