The vendor hype machine is up and running, and master
data management (MDM) will be in your favorite software salesperson’s messaging and in their quotas by the time you read this. That hype drives business for a while, but soon
MDM will have to prove its mettle.
So should you listen or care? Is MDM a fad, as some believe, or is it here to stay? I believe information management architectures are ripe for innovation as the transition continues to more
operational forms of business intelligence (BI), moving it back to its most effective place alongside business operations. MDM fits right onto that path, offering key supporting data for operational BI.
Most data warehouses only address mastering data on the periphery. Think back to the origin of your data warehouse. Like many, its creation was probably due to operational system limitations to querying the data (at the same time as it was running the business). The immediate knee-jerk reaction to this was to copy the data (from the one system giving you the problems) to another one, which refreshingly did not have operational obligations. Let’s call that the data
warehouse release. It can expand to other systems (and hopefully integrated and cleansed data).
The major challenges to the data warehouse strategy lie in the sourcing of the vast amounts of transactional data. Master data (i.e., dimensional data) gets the second billing in this scenario because there are only so many problems that can be dealt with at once.
Those who addressed their data warehouse with some top-down planning and vision, and averted the continual re-evaluation of their core approach, have already accomplished some of the goals of MDM - namely data integration, data modeling and data quality. Experienced information management professionals will already understand the innate
value of these items, which constitute most of the rationale for MDM today. However, whether they get addressed in the data warehouse or in a new, separate data
hub seems to be an act of fate or bias. That’s the reality. The data flow should be MDM hub first for master data integration, modeling and data quality, which can then feed the data warehouse.
However, even if you have a best-practice data warehouse, you may still be in need of some of the value proposition for MDM. Specifically, MDM hubs are placed in operations (logically and often physically) for brokering master data operationally.
Few have solved the problem of making the “clean” data warehouse available to the operational environment - in other words, making the data warehouse part of a closed-loop system with operations. Several challenges, mainly operational and span of control in nature, have seen to it that data warehouses provide post-operational analytic functions only. The MDM hub is built with this in mind. It’s very operational in nature, providing (and sourcing) clean master data to operational and data warehouse systems.
So far, I have spoken architecturally of MDM. I will now say the same thing about the value of MDM as I have said about data warehousing and data quality - there is none on the surface. You cannot build an MDM hub and sell it on the market to gain revenue (at least I don’t think you can). It needs to assist other revenue- and expense-targeting applications.
More reviews about the http://www.dmreview.com/channels/master_data_management.html