There are 3 known methods to CDC in Data Services, two of which are source based:
Source Based CDC
-Using a timestamp, date, datetime, or int column to uniquely identify the next "band" of rows to be loaded into the delta process in combination with a status table, global variable, and where clauses of the query transform. This is demonstrated in the Tutorial file provided with Data Services in the CDC chapter.
-Using the Database native CDC functionality, available in Oracle, SQL Server and I believe Attunity
Changes are replicated to change tables in the source, marked as insert/update/delete. Data Services can natively read these tables through the use of CDC Datastore objects and Map_CDC_Operation transform.
Target Based CDC
-Must be used if the methods above are not available. The Table_Comparison transform is capable of comparing all source records to the target table to determine if an INSERT or UPDATE is required for that row.
If the "Detect Deleted Rows" option is ticked as well, the transform will also generate DELETE records. Be warned - this tick box only functions as desired when the full dataset is coming from the source system! For example, if you send 10 rows to a target with 100 rows, 90 will be marked as deletion due to absence of these rows in the source dataset.
Bob