Although master data management systems have many different configurations, they are essentially designed the same. This design provides a structured environment in which the master data management (MDM) system can provide precise and reliable management of data and data processes. At the heart of all of these systems is the master data management hub, a database in which master data is collected, cleaned and stored. The master data management system may use various hubs to administer different sets of data, such as customer or product data. Each hub generally uses one of three common types: transaction/repository, registry or hybrid.
In a transaction/repository type hub, all applicable data is stored and accessed for a single database which must contain all the information needed by the different applications which access it. All data is merged and integrated, and published to individual data sources after it has been linked and matched. This type of hub permits one source of data to be created, similar to data governance, thus reducing duplication by making duplicates easier to identify as data is collected and cleaned. There are disadvantages to this model; current applications may require modification to be able to use the master data. In some cases this modification is not possible. Applications and services that function as a temporary interface between the master data management software and data-dependent applications may be required, which can be costly.
Registry hubs are just the opposite of transaction/repository hubs; master data is kept within native application databases. This type hub allows for applications to remain fairly intact because the data is managed within native databases. As the database grows this type of hub can become inefficient because each time a request to access master data is made, the data must be found, a request distributed between the numerous databases and a list of requested data formed; all this must be done in real time. Additionally, duplicate data entries may be found in different databases or even within the same database. Cleaning and consolidation of the individual databases would be the best solution, but is not always practical. A further disadvantage is that when new databases are added to the hub registry, new keys must be added to current tables.
Hybrid type hubs are a combination of transaction/repository hubs and registry hubs and try to address some of the problems present in both. The hybrid system merges the advantages of both systems; master data is left on the native databases, generating keys and IDs to access this data, but duplicating some of its important features to the hub. The hub can service the more common requests and queries need only be sent for less-used attributes, resulting in increased efficiency. The disadvantage of the hybrid hub is that since it stores replicated data from outlying databases, it may encounter updating issues and deciding which attributes to store, naming to be used and format to store them in can create problems.
