Sorts the unloaded rows into clustering index order using DFSORT. This phase is CPU-intensive and requires the SORTWK datasets. The amount of sort work space needed is approximately 1.5 to 2 times the size of the unloaded data.