The sample percentage that you specify may differ from the actual sample percentage during a test run. The sample percentage that you specify represents the chance that each data is included in the sample.
You cannot use native sampling if the underlying Oracle tables or views from which you are sampling do not have unique primary keys. Instead, use PowerCenter sampling.
If you run sampling on a table pair, you select one table to sample. In a master-detail relationship, sample the detail table because it is the bigger table. When you sample the bigger table, the joiner processes a smaller set of data. Table A is the master table and table B is the detail table.
Some tests might have different results based on whether you enable data sampling.