Evaluate the maturity of your MDM system by answering the following questions:

1. Does my MDM support extensible data types such as XML?
2. Can internal, partner, or client services search, identify, and consume the CDM?
3. What is the effort for an application to participate in MDM as a consumer of the CDM?
4. What is the effort for an application to participate in MDM as a contributor to the CDM?
5. What is the effort to replace an external service provider?
6. Are data quality and conformance services exposed for use by applications or external parties?
7. What is the turnaround for changing the functionality of the current MDM system?
8. Can the MDM handle near real-time requests for conformance from participating systems?
9. Can the MDM exchange metadata with other MDM systems?
10. Can the MDM infer context and take action based on the semantics of the information being exchanged?

We've seen how SOA paradigms can contribute to the maturity of MDM. Now we'll focus on the other side of this marriage and see why SOA needs MDM. While SOA enables integration and data exchange through services, such integration is useless without a common vocabulary of data content and data structure. MDM defines how the enterprise establishes and maintains such vocabulary. To fully adopt enterprise SOA, an organization must first address MDM.

MDM is one of the most important components of the data services layer providing the required semantic integration of services for master data. Without an MDM system (or an ad hoc capability providing the MDM functionality), services don't know where to access the single version of the truth for "customer A" when there are multiple systems that capture "customer A" information. Moreover, this "customer A" information has to be the same in terms of structure, as well as content, when it's consumed by services.

The technical intersection of MDM and SOA occurs at the data services layer. For the data services layer to provide consistent information to consumers across the multiple data providers, data and metadata inconsistencies, discrepancies, omissions, and overlaps have to be addressed. This means that MDM functionality must be present. MDM crosses and has elements in all three areas of the data services layer mentioned above.

• Service enablement of traditional MDM functionality such as data quality and data harmonization (synchronization of data across participant systems and MDM) is exposed under the Enterprise Data Services area.
• The SOA designers and developers who are creating business services, as well as others consuming services, have to reference the organization's master data schemas. These master schemas are exposed in the Enterprise Metadata Services area, allowing consumers to draw inferences directly from their semantics.
• Finally, in the Enterprise Data Platform we find services around the management, measuring, monitoring, and reporting of the MDM system.

When the topic of semantically conformed data is raised, many firms point to their data warehousing initiatives. Consider the following typical question in that regard:

"I don't have MDM but I do have an existing Enterprise Data Warehouse (EDW) that covers the integration and data quality that seem to overlap what MDM is supposed to do. Can't I service-wrap my existing system to achieve SOA MDM benefits?"

Service-enabling the EDW is indeed a worthwhile and beneficial initiative but doesn't offer MDM capabilities. The reason is that most prior EDW efforts have concentrated on the integration and "business view" of data from disparate sources rather than the harmonization of the data at the source systems - meaning that when the source systems are service-enabled, they'll be in semantic conflict with the service-enabled EDW. In addition, EDW typically doesn't concentrate on master data, master metadata, and related data governance issues. Therefore, most EDW systems don't have the faculties for managing master data (such as an interface with workflow for business users to manage master data through its lifecycle). Finally, to successfully proceed towards an MDM solution, the initiative must be executively sponsored and business-owned versus an IT service enablement of an existing application project. To be clear, an existing EDW should, can, and will be leveraged in the new data services layer but doesn't replace the need for an MDM system.

Evaluate the maturity of your data services layer by answering the following questions:

1. Are my services impacted by changes to the repositories or databases being accessed?
2. Do my services have to call two or more repositories to read or update information?
3. Do two or more of the repositories being accessed contain overlapping information?
4. Do my analysts and developers need to understand the system internals and entity models for each interfacing system?
5. Do we have duplicate, incomplete, missing, or conflicting data across systems?
6. Can the data service layer provide a "single version of the truth" for the master data (customer, product, vendor, employee, and assets)?
7. Are my services semantically integrated to define the customer, product, or vendor?
8. Would data quality and conformance intermediary services be helpful?
9. Does my service taxonomy include services to access data and metadata across providers?
10. Is my enterprise data model exposed for consumption by the services?

The quest for the modern MDM system, and the single (data) version of the truth for SOA enablement, entails many challenges and risks. The highest business risks across both types of initiatives include: lack of executive support, insufficient business (versus IT) data ownership, and inappropriate skills or expertise. From a technical perspective, the highest risks include: inadequate performance, incorrect security of the exposed enterprise model, and ill-advised vendor selection.

Marrying MDM and SOA makes sense and offers vital benefits from both the MDM and the SOA perspective. Expect to see more such paired initiatives and vendors repositioning their products to cover the joint space better.