One of the accusations occasionally leveled against service-oriented architecture is it doesn't live up to the organization's expectations. A recent blog post by Gartner analyst Andrew White offers at least one good reason why this might be happening, and along the way, he explains the connection between SOA and master data management.
White looked at a recent Gartner survey of SOA deployments in the retail industry. SOA appealed to them, in part, because it can be applied to multi-channel integration and that is, as White calls it, "one of those last bastions of integration complexity" for retailers, who have either acquired or developed a range of systems to address one channel. The result, writes White, is a "ton of spaghetti integration (point-to-point) across channels to integrate product, data, order, customer and price data."
SOA to the rescue, right? Well the problem with this approach is SOA doesn't solve the chaos in the underlying data. And that is where MDM should come into play, except retailers aren't pursuing MDM with the same hoozpah they're using with SOA, White points out. His fear is that they wrongly believe SOA will solve this problem and they won't need MDM.
It seems SOA and MDM go together like peas and carrots. And White's not the only one who thinks so. Kyle Gabhart does, too.
Gabhart is a a SOA evangelist and the SOA Solutions Director for technology education and mentoring company Web Age Solutions. In a November post on SOA and MDM alignment, Gabhart explained how the two support each other, surmising:
MDM drives how the enterprise manages data. SOA drives how that data is accessed and utilized in a business context. SOA and MDM were made for each other.
Of course, there's a catch. There's always a catch. While the two may work well together, the relationship is not without conflict, as Gabhart explains in a series of articles published this month at Sys-Con Media.
The problem is in the data model. He explains how the data model required for SOA artifacts can come into conflict with the master data data model in the first article. In the second article, he explains the questions that come up when SOA is already in place and you add in MDM.
Both pieces are a bit more technical and will probably appeal to technologists more than business readers, so I'll sum up: Your enterprise architect-or whoever is in charge-needs to do some serious planning and know a lot about data modeling or you'll have problems. Given the potential expense of both SOA and MDM, I think you'd be well-advised to take the data modeling question seriously.