While other analysts tend to talk about the people and process problems of MDM, Aaron Zornes (@azornes) sees it a bit differently. Zornes is the chief research officer for the MDM Institute, the leading research and advisory consultancy exclusively focused on master data management, and he contends that the tools are also a major part of the problem. Vendors have a history of not delivering the technology solutions organizations need to support active, integrated data governance, he explains to IT Business Edge’s Loraine Lawson.
Lawson: What do you mean when you say “passive” and “passive aggressive” data governance?
Zornes: We were saying we need to get some definition about what governance, data governance or master data governance is. It’s people, processing and technology, and what’s been lacking is the software technology.
All the consultancies in the world have been saying, “Well, it’s all about the people and the process.” But in our client experience, if the people and process generate a Visio chart and a PowerPoint, a Word doc, Excel spreadsheet and then that’s it, that doesn’t really manage the quality of your data or the mastering of your data. It’s a one-way, dead-end street. It doesn’t pass off to the rest of the master data environment.
We were raising Cain basically with the major software vendors, IBM, Oracle, the upstart MDM vendors, the best-of-breed guys, everybody, saying, “Look, we need something to help us define and manage what master data is. And yes, you can give us a data model. It might be part of a customer data model or product data model and we can evolve it, but there’s a lot more to the data governance than that.
We need to know where the data comes from that fills in the master data view … Who is the trusted source for address changes? Who is the trusted source for this credit risk rating? Who is the trusted source for marital status change? Who is the trusted source for an address change or who is the legal entity identifier trusted source that we tag each of these master customers or organizations with?”
Now you’ve heard the words data stewards, data trustees, data governance councils, that’s all great stuff and companies like Cognizant and EMC have been making a lot of money building software systems to do that work flow to help the trustees, the stewards, the directors of data governance, to help those people identify what data’s being mastered, where does it come from, and who can access it and what is the audit trail and yada yada yada.
So enterprises have been paying a million or two, to companies like Cognizant and IBM Professional Services, Oracle Pro Services, and Informatica, to build this for them and it was always custom work. And even that didn’t generate the data quality rules that then execute in the MDM hub. So if we go back a step, did you ever - were you in the industry when there was something called CASE … Computer Aided Software Engineering?
Lawson: No. What’s that?
Zornes: CASE stood for “computer aided software engineering: In the ‘80s, ‘90s, there was this big trend called higher order software. We want provable software, higher order software. When we write software, let’s prove that it is going to execute the way it’s supposed to execute. So they created two categories of tools: a design tool and an execution tool, Lower CASE and Upper CASE - just to be cute, that’s what they called it.
Upper CASE were design tools, which let you design what your systems were going to do. You visually diagram it and show it to the users and they say, “Yep, that’s it.” Then you push a button and the upper CASE tool would generate code for the Lower CASE tool to execute against the database.
We’re in a similar situation where primarily we have lower CASE MDM. We have MDM hubs that can execute rules about MDM all day long. The MDM hubs have the ability to execute all these rules, but how do you create these rules? How do you define them? What is the customer master? What is the product master? What is the vendor master? There’s also a bunch of rules that have to be adhered to, based on your industry.
The way we used to build MDM systems was sort of cowboy style; we’d wing it. We’d get a data model and we’d say, “Okay, we’re going to get some data from here, some from there.” We get it up and running but then there’s no process for continuity.