Master Data Management or MDM for short has become an essential and integral portion of BI landscape today. If we take step backwards, the need for MDM was realized only during the previous decade. Before a formal MDM tool was implemented in an organization, each application was running in a separate silo with different standards for Master Data.
Although there are several characteristics for Master Data, a simple definition is “any data that does not change often and against which transaction data is referred to“. Some examples are Customer, Vendor, Product, Employee etc. In a typical BI Data Model, the dimensional data can be classified as master data.
One of the primary goal of any BI project is to provide “single source of truth” to the end user. Problems started arising when integrating different systems into a consolidated warehouse, wherein the consistency of source data became a bottleneck. The example demonstrates a typical scenario where master data starts to deteriorate in quality. This problem was typically handled by ETL tools by checking the data and making sure they conform to standards. This approach of having an uniform master data only for reporting purposes only solved the problem temporarily because it did not address the real reason behind fragmented data.
An MDM tool was in need of the hour for organizations because data quality was becoming a source of concern. Although the need for MDM was first felt in BI space, now the adoption is much wider and integrated across all applications in the entire landscape.