MDM Storage
MDM Storage is created automatically when the server starts for the first time, and the generation of the structures for the storage is based on the defined model. The MDM Storage structure is transparent to users while the implementation details of the structure are completely hidden.
MDM storage includes a repository of cleansed, matched, and mastered data consisting of the following:
-
The current data repository stores the instance records, or source records, from the different source systems.
Storing data from the source systems has a critical purpose and provides many benefits. All records from the source systems are stored in their cleansed form, which allows systems to retrieve cleansed versions of their own records, and comparison against existing data is done when receiving updates.
This makes it possible to efficiently handle updates in MDM and cross-system distribution of source records.
-
The Matching Key tables for all matched entities. These tables enable incremental matching, as well as identification of records when used as a service.
-
Other technical tables.
-
-
The historical data repository stores previous versions of both instance and master data.
History contains data in a defined scope (specified entities, specified columns) and in defined frequency (all changes vs. change triggered by some specified columns). Data is stored as BLOBs.
The MDM storage relies on a database and is platform-independent. All databases commonly found in the enterprise are supported as long a JDBC driver is available for them. However, some databases allow for performance optimization techniques, so a selection of those technologies for the MDM storage is recommended.
The MDM storage tables should not be accessed directly. Instead, the user should use only the online or batch interfaces to interact with the MDM repositories. The use of logical transactions is another reason to rely only on the standard interfaces to access the MDM repositories.
Was this page useful?