photobank.kiev.ua - Fotolia
Having worked in the IT industry as long as I have, I know how much we all love TLAs; i.e., three-letter acronyms. In recent years, one such set of letters, MDM, has come to be used in two ways in data management circles, both important in their own right and, increasingly, in tandem.
I'm talking about master data management and metadata management; and yes, the latter technically doesn't shorten to MDM as an acronym, but some people do so nonetheless. While the two are distinct tools, the rapid growth of mobile devices has made an integrated metadata and master data management process crucial to enabling accurate data analytics and effective business decision-making.
Let's start by reviewing what metadata and master data are. To use an old rubric that is still a valid definition, metadata is data about data. The simplest explanation involves an entry in a relational database or a spreadsheet. On its own, $552.12 is just a set of characters. It's the metadata about those characters that tells us they're in a numeric field used to represent currency. Metadata gives the data meaning.
While the basic importance of metadata seems clear and obvious, there are wider issues. When you have two systems that need to communicate, they can't do so unless you can translate data between them. But, to continue with the example from above, currency can have different metadata descriptions in the two databases. One of the most obvious problems is in the name of the fields: Do fd_sales_territory1 and Region 1 Sales Total refer to the same piece of data or to different things?
If an IT department can't use metadata to properly identify information, the results of data analysis applications become questionable. That's where the other MDM comes in.
Lessons learned on managing master data
One of my current soapbox statements is that people in IT need to apply the lessons from the PC explosion in the late 1980s and early 1990s to the mobile device explosion happening now. When PCs emerged, what was centralized and tightly controlled information was suddenly in the hands of many people who could modify it separately and create their own data definitions. That led to a lot of inconsistent data and confusion among different users.
The need to mediate between centralized and decentralized information to ensure data consistency and continuity was one of the reasons for the growth of both metadata and master data management. That sales data example? It was a critical issue to developers building sales force automation and business intelligence systems in the wake of the PC boom.
The growth of the data warehouse was one attempt to address the mediation issue. For years, people talked about using enterprise data warehouses (EDWs) to create a single source of truth for all of an organization's information. However, the complexity and unwieldiness of EDWs became apparent as companies moved forward on deployments.
Data warehouses typically included metadata management tools as part of the setup for handling the extract, transform and load feeds used to populate them. The tools helped normalize data for users to better understand (and no, I don't mean the relational database definition of normalization). What evolved from that was a separate focus on ensuring consistent definitions of data across applications and departments.
That's how master data management came to be a discipline, designed to formalize a process to see enterprise data in a consistent and accurate way. The master data management process subsumes the metadata management one in EDWs, with the ultimate goal of implementing common definitions of master data across a company's systems.
Mobile push changes MDM equation
While IT was starting to get a handle on the two MDMs, laptops, smartphones, tablets and other mobile devices came along and rapidly expanded the ability to collect and process data in the field. Apps and applets are popping up (pun intended) all over the place. The mobility movement makes managing master data and metadata even more difficult.
Each mobile application has the potential to see metadata differently than others do. That means you need an expansion of mediation between systems to make data understandable. It's true that many mobile apps are leveraged on top of cloud frameworks, such as Salesforce, helping to normalize data in this sense. But there are competing cloud platforms, and many existing application vendors -- Oracle and SAP, for example -- offer their own apps. IT teams must carefully analyze the metadata in new mobile systems and ensure that data can be linked together in a coherent and accurate way.
Master data management is a key component of doing so. No new mobile applications should be approved until it's known how the data being captured in them meshes with a company's master data definitions.
When a busy salesperson is told by her buddy about a nice new app to download, or when a business unit wants to buy mobile software for all of its users, will the captured data match what corporate execs need to see? If it isn't comparable with the rest of the organization's data, its value is severely limited, and the larger business goals of the enterprise are put at risk.
Metadata and master data management both have their places and leverage each other. When you hear a vendor talk about its MDM product, make sure that you know which of the two it means -- and that it isn't giving the other one short shrift. Creating a combined metadata and master data management process will become even more complicated if you end up with tools that don't mesh well together.
Five basic steps involved in setting up a master data management program
Read an excerpt from a book on managing multi-domain MDM initiatives
Consultant David Loshin on the need for metadata management in data lakes