Metadata management is often overlooked, misunderstood or assumed to be taken care of automatically as part of an MDM initiative. While it’s true that databases can generate metadata reports based on a logical data model, there’s a lot more than that to metadata management in an MDM initiative.
As companies start to include multiple domains such as customers, products, suppliers, etc. in their MDM initiatives, the collaborative lifecycle management of the master data across business and IT functions will become a challenging change management undertaking.
One of the key benefits to addressing metadata management is to lower cost of ownership by documenting the entire end-to-end process for master data at the metadata level that greatly enhances change control across business and IT functions.
“Metadata is structured information that describes, explains, locates, or otherwise makes it easier to retrieve, use, or manage an information resource. Metadata is often called ‘data about data’ or ‘information about information’” (National Information Standards Organization, 2004).
Metadata within an MDM initiative includes all of the following: the source systems, MDM hub, data quality tools, business process management (BPM) and workflow tools, and enterprise application integration (EAI) tools.
The need for metadata for different semantics between the source systems and the MDM hub is the most apparent, as a source and hub may call the same thing by different names, e.g. part number versus SKU.
However the business rules, transformations and data rules in the data quality, BPM and EAI tools are often overlooked. Yet the complete data lineage from source to hub and back to source needs to be fully documented at the metadata layer.
So if managing metadata is so important, how does one go about managing it?
There are three approaches to managing metadata: a centralized approach, a distributed approach or a federated (hybrid) approach.
The centralized approach stores all metadata centrally, providing easy access to information, scalability and performance. This approach is more complex to integrate initially than the distributed approach, requires continuous synchronization and is also a single point of failure.
The distributed approach takes advantage of each participating system’s own metadata management capabilities and simply reads the information as required. In essence, this is simply a web portal that queries the relevant information from all participating systems on demand.
The federated approach is similar to the distributed approach in maintaining references to the metadata information for all participating systems. But also like the centralized approach, it can store metadata information locally, thus taking advantage of the best of the centralized and distributed approaches.
Choosing the best approach for an organization depends on a number of factors. The metadata capabilities of some of the existing systems in use by an organization may prohibit the federated or distributed approach, so a thorough assessment of your current tools is a necessary consideration.
Also if you are planning on upgrading existing tools, make sure that the selections fit into your metadata management approach.
MDM initiatives will require significant attention to be paid to semantics across the enterprise and metadata management in order to be successful.