The process of dividing the underlying data into subsets that can run in multiple processing threads. When administrators enable the Data Integration Service to maximize parallelism, the service increases the number of processing threads, which can optimize mapping and profiling performance.