When you configure the stage process, you need to map the landing table columns to the staging table columns. To configure load by row ID, map the landing table column that contains ROWID_OBJECT values to the ROWID_OBJECT column in a staging table. For example, you can provide ROWID_OBJECT values to the MDM Hub by mapping the INPUT_ROWID_OBJECT column in a landing table to the ROWID_OBJECT column in a staging table.

The load process compares the values from the ROWID_OBJECT column in a staging table with the ORIG_ROWID_OBJECT column values of the cross-reference table that is associated with a base object. If the values match, the load process updates the records in the cross-reference table.

The load process updates the existing cross-reference record with the incoming record if the values of the PKEY_SRC_OBJECT and ROWID_SYSTEM columns match. Otherwise, the load process inserts a new cross-reference record that has the values of the incoming record, and merges with the existing cross-reference record. The load process updates the base object record with the winning cell values from the cross-reference records.

When you load a HUB_STATE_IND value of -1 for a specified source and specified row ID, the HUB_STATE_IND value is set to -1 for all cross-reference records of the specified row ID, regardless of the source. The MDM Hub considers the base object record to be deleted since all cross-reference records have a HUB_STATE_IND value of -1.

The initial data load for a base object inserts all records into the target base object. Therefore, enable loading by row ID for incremental loads that occur after the initial data load.