Impact of Matching Configuration Changes
To allow MDM to work in incremental mode, the Matching step creates a repository table for each matching. This table stores data (specifically, attributes defined as a partition) and matching keys.
For optimal performance, matching keys are stored in a denormalized format in the matching key table. Therefore, any modifications that could potentially impact matching key values require those keys to be recalculated.
This includes changes to:
-
Partitioning rules and their order.
-
Key rule order.
-
Key rule components (attributes) and their order.
The last two items affect tiebreaking when records belong to multiple key rule groups. -
Transformations applied to columns used in these configurations.
To ensure accurate matching after such changes, reprocess the affected entities using the approach best suited to your requirements:
-
Reprocessing without rematching (Rematch Type column is set to
NOMATCH): Records are reprocessed but not rematched. In this case, only new or rematched records are affected by the changed rules. -
Reprocessing with rematching (Rematch Type column is set to
REMATCH): Two strategies control how existing manual matching decisions are handled:-
Existing manual matches are preserved (
rematchStrategyset toPRESERVE_MANUAL). -
Any existing manual decisions are discarded (
rematchStrategyset toREMOVE_MANUAL). Applies new rules to all records, including existing matching groups.Keep in mind this will remove any previous manual matches, as described in Working with Matching.
-
Was this page useful?