Technical overview of virtual MDM

Whether your business is banking, security, healthcare, or perhaps consumer goods, understanding your customer is the key to your success.

With volumes of data that are stored across multiple source systems and the often dynamic state of that data, organizations are faced with challenging integrity, profiling, and relationship issues. Using IBM® InfoSphere® Master Data Management adds value to your organization by increasing data integrity and providing a 360-degree view of your data and the relationships between data objects.

In the realm of Master Data Management, implementation of a virtual MDM operational server supports environments where master data is created in a distributed arrangement across multiple source systems and remains fragmented across those sources. The virtual MDM operational server provides a central indexing service for the distributed data. By implementing a virtual MDM operational server, you can combine, compare, review, and resolve potential data issues.

Virtual MDM configurations can be implemented when you use either InfoSphere MDM Standard Edition or InfoSphere MDM Advanced Edition.

Using a sophisticated and flexible data model and attribute weighting, the MDM probabilistic algorithms identify and link records across the distributed source systems. The MDM operational server receives updated transactions from the source systems. The deployed algorithms identify data issues and linkages across the systems that potentially represent the same member or object. The algorithms detect issues that are based on various data elements (member or object attributes specific to your business environment). The algorithms standardize each data element and compare both similarities and differences to determine the likelihood of a match. As issues are detected, the MDM operational server assigns comparison scores and manages the records by using threshold settings. This process gives you a complete member picture. This process does not affect the data that is stored within your local source systems; record data in the MDM database is a subset of information available from the sources.

The MDM operational server uses statistical analysis and other processes to determine the likelihood that records might pertain to the same party. No process for linking records is perfect and, given the imperfect nature of statistical analysis, inaccurate input data, and other uncertainties, the result of the analysis that is processed by the operational server might contain errors. These errors can include false positives and false negative. False positives are records identified as possibly pertaining to the same party, which actually relate to different parties. False negatives are separate records that are not identified as pertaining to the same party, but actually do relate to the same party. A specific linkage weight does not necessarily have equivalent relative importance across the comparison of different files. Independent verification of the accuracy and completeness of linkage results, and the decisions made that are based on the linkage results, is required.

The following diagram shows the elements of the virtual MDM architecture.
Figure 1. Concepts in virtual MDM
The diagram shows the elements of the virtual MDM.