Informatica Perspectives

Years, ago, I came across this question in an article in Boardroom Reports:  "What do you call a hamburger that’s 99% meat and 1% garbage?"

The answer was a "garbageburger." In other words, even if a small fraction of the burger is tainted, the whole meal is tainted. The original analogy was being used to illustrate the challenges of time management, but it's an apt analogy for data environments as well. That is, if a portion of the information is bad or unreliable, trust in all the data eventually breaks down. In essence, many implementations of service-oriented architecture (SOA) taking place across companies may be garbageburgers because they are serving up unreliable information – an element that has been out of the control of SOA designers.

Sorry if I ruined anyone’s lunch, but the point had to be made.

In recent posts, I talked about the relationships between SOA and data management – how data services are needed to deliver reliable information through the SOA (as so astutely explained by Ash Parikh), and how SOA governance and data governance need to be closely linked, because one depends on the other. And they share common objectives: both encourage sharing and reuse; both break through the walls of enterprise silos; and both offer potential for greater agility for the business.

A new book, Enterprise Master Data Management: An SOA Approach to Managing Core Information (IBM Press, 2008), puts it this way: “a process is only as good as the information it processes, and similarly, information needs to be tied to the context of some process to be of any value.”

The ultimate success of SOA, then, depends upon the ability of MDM to deliver reliable, trusted data that accurately reflects the state of the enterprise. MDM and SOA also are closely intertwined, as both reach across the enterprise to make information accessible on demand. "Without MDM, applying SOA can make matters regarding information worse – not better – because low-quality data would be exposed through untrusted services, and thus consumed by even more consumers," notes Allen Dreibelbis and five co-authors.

MDM, they add, "provides the technical foundation for master data use in a SOA-style architecture. From a SOA standpoint, the service concept applied to information leads to the discovery of the concept of information as a service and providing a flexible data abstraction layer. Accessing master data is not directly done through database interfaces. Instead, access to master data uses services encapsulating a lot of functionality, such as duplicate checking or address standardization.  Master data services are one type of information service."

An MDM infrastructure should be component-based (as is SOA), and be designed along the same principles as SOA, the authors say. “Design of the MDM solution architecture should be based upon SOA design principles and concepts such as the use of Web services and the SOAP protocol.”

It’s clear that data integration and management is the next challenge for SOA. MDM establishes a single, trusted source that will ensure the credibility and success of SOAs. It’s time for SOA and MDM proponents to work together to deliver this new capability.