By evaluating data dependence relationships as part of the process of integrating a data warehouse with a master data repository, one begins to see patterns of at least three key types of accesses. A search or locate access queries the master repository to determine if an entity already exists within the master data set, a read accesses the data, and an update will modify the data in the master repository.
In most cases, a data warehouse would be used for reporting and analysis purposes, and typically would not impose any updates to master data entities. However, it is not unheard of to have business processes in which individuals interacting with the data warehouse might be compelled to update a master copy, so it is important to review the end-user usage scenarios to properly assess the dependence relationships. Based on an understanding of the set of data attributes that are ultimately shared by more than one business process, those dependence details can be used to determine if and how the master data model would need to be adjusted to support integration with the data warehouse, including
- Modifications to the set of master data attributes managed as shared items
- Modifications to the set of identifying attributes to ensure unique identifiability as well as ability to ensure precise and accurate entity resolution
- Modification of existing master data model elements (e.g., types and sizes)
- Creation of new master data entities
- Addition of new master data model elements
Aside from directing modifications to the models, these dependence requirements will also suggest the need for methods for assuring consistency between source and target among the MDM repository model, suppliers to the MDM repository and the consumers of data from the MDM repository.