Sorting input data according to Group By ports significantly improves Aggregator performance. When data is pre-sorted, the aggregator doesn't need to build and manage a large aggregate cache in memory; it can process groups sequentially. Passing unsorted data requires caching all rows until all groups are complete, increasing memory usage and processing time.