The Data Integration Service might split a Filter transformation if the filter condition is a conjunction. For example, the Data Integration Service might split the filter condition "A>100 AND B<50" into two simpler conditions, "A>100" and "B<50." When the Data Integration Service splits a filter, it moves the simplified filters up the mapping pipeline, closer to the source. The Data Integration Service moves the filters up the pipeline separately when it splits the filter.

When you choose the normal or full optimizer level in the Developer tool, the early selection optimization method is enabled by default. When you use the auto optimizer level, the early selection optimization method is enabled in the following cases:

The mapping runs in the native environment or on the Blaze engine.
The mapping contains a data source that supports pushing filters to the source.

You can disable early selection if the optimization does not increase performance.

The Data Integration Service ignores early selection optimization if a transformation that appears before the Filter transformation has side effects. The Data Integration Service cannot determine if the SQL transformation, Web Service Consumer transformation, and Java transformation have side effects. You can configure early selection optimization for these transformations if they do not have side effects.