Many commercial applications involve sorting and searching. SEARCH is part of the underlying database functionality although in many cases, particularly in the MVS community, COBOL applications still use raw or indexed-sequential files rather than a proper database. SORT is a valuable problem to parallelise because sizes are such that it invariably involves backing store, usually disk.
The correspondence between SORT and SEARCH on COBOL tables and on SQL relations is obvious, and there is much previous work in this area from the RDBMS community, hence where possible, calls will be planted to the underlying RDBMS (Kumar et al., 1995).