Incremental UOW

Incremental UOW processing examines records' timestamps in the source UOW tables to determine which records have changed since the last processing job in any mode (UOW or Reload, Full or Incremental), and creates the execution set for those records' UOW key (subject or subject visit) and no others. If the program has never been run in any mode, all units of work are included in the execution set. It performs insertions, updates, and refreshes but no deletions.

As with Reload processing, UOW's Incremental mode is faster than Full mode, so you may want to use incremental UOW processing frequently and full UOW processing less frequently but regularly, to ensure that data is appropriately deleted.

Note:

The end result of Incremental UOW processing is the same as Incremental Reload processing. Both process all new and changed records and delete no records. The end result of the two Full modes is also the same.

Which process will be faster depends on the volume of changed data being processed and whether changes are concentrated in specific units of work or spread fairly evenly across all units. Compared to Reload, UOW processing has overhead costs in detecting affected subjects or subject/visits, but it is more efficient in that it processes records only in units with changes, not all records.

In general, UOW will probably be faster than Reload when the number of incremental changes is small or concentrated in relatively few units of work.

In addition, if you use custom programs in transformations, using UOW processing takes care of finding incremental data changes; your code does not need to handle that.