Master Data Management – An Introduction
Writing by Manishanker on Wednesday, 21 of July, 2010 at 5:49 am
In any organization, it is often found that the key data whether related to product or customer is maintained in multiple systems, which may some times lead to inconsistent view and duplication of the data. Master Data Management (MDM) is a set of processes that seek to ensure that an organization does not use multiple (potentially inconsistent) versions of the same master data in different parts of its IT systems.
Before appreciating MDM, let us understand Master Data and its classification. Master data can be defined as data that is non-transactional in nature, shared or used by most systems. It generally falls into categories such as People, Things, Places and Concepts, which are further classified based on the organization’s business model. Customer and Product master are the most critical master data to be managed for any organization.
Often organizations have a hard time creating, finding, and managing data that is complete, as different IT systems would have been established at different times using different technologies by different business units.
MDM and its processes focus on the need to clean up the inconsistencies from the legacy systems data as well as to create an accurate, timely, and complete set of master data for the growth of an organization. It is comprised of a mixture of business processes or applications, methods, and tools. The important activities/processes that will be performed during the appropriate phases of MDM implementation include, identifying the source databases containing master data; collecting and analyzing the metadata information for master data; finalizing the common master data definitions; creating master data model(s) for MDM DB and XMLs; defining data transformation rules for transforming legacy source data to MDM database; identifying or building toolsets required to create master data by cleansing, transforming, and merging the data; collecting and harmonizing the unique instances of data to populate the shared MDM repository; integrating with existing and new consuming applications to synchronize the data periodically via a SOA approach; and establishing data governance policies and procedures to maintain accuracy, cleansing process, and timely synchronization of data.
There are various architectural approaches for developing a Master Data Repository, with Hub – Repository being one of the most common approaches, where core master data is managed with a single repository and the data is not replicated to multiple systems. In this case, all the source systems provide updates to the central hub and the consuming systems receive data via central services (SOA), whereas the common interfaces/services perform the data cleansing, transformation, and mapping to the MDM repository.
The popular MDM products in the market include IBM InfoSphere Master Data Management Server for Product Information Management, IBM InfoSphere Master Data Management Server for Customer Data, SAP NetWeaver Master Data Management Server, Oracle Master Data Management Suite, Talend MDM (Open Source), Kalido Master Data Management, Siperion Multi-Domain MDM Hub and Tibco Collaborative Information Manager.
MDM implementation may vary from organization to organization depending on the need and time of implementation. Testing the MDM, which is key in MDM implementations, covers various scenarios such as, large volume of source data from multiple sources with different data standards; initial data migration testing from different sources to MDM database; periodical data synchronization between source systems and MDM DB; testing of multiple technology components; multiple failure points and consuming systems interfaces; and also validating the consuming systems especially if it is a product with no proper documentation.
Category: Software, Software Testing
- Add this post to
- Del.icio.us -
- Meneame -
- Digg
No comments yet.