RESEARCH & RESOURCES

Semarchy Accelerates MDM

MDM often seems like an endless process of planning. MDM specialist Semarchy touts a new take on MDM that it says emphasizes doing.

Justly or unjustly, there's a perception in the market that master data management (MDM) is a boondoggle for everybody but the business: software vendors, integrators, outside consultants, everybody but the business benefits. If you want MDM, so the conventional wisdom goes, you're going to pay through the nose for it. Consequently, MDM has long been the domain of the Big Guys: Global 2000 firms and the like. Given what's involved, it's hard to see how that could change.

But could it? Recently, we've seen the emergence of a handful of vendors -- upstarts and veterans alike -- who say it can. Players such as Liaison Technologies, Reltio Inc., and RedPoint Global Inc. (among others) are targeting what they see as an opportunity in the MDM market: a pent-up demand for MDM-for-the-rest-of-us. You can add upstart player Semarchy Inc. to this list, too.

Each of these vendors touts what it claims is a new, cost-reducing take on MDM. Collectively, they claim that the MDM we have -- for which businesses are accustomed to paying beaucoup bucks -- is a vestige of the past: of the strictly-structured, data warehouse-centric environments of the early 2000s. Liaision, RedPoint Global, Reltio, and Semarchy all claim to drive down MDM costs by leveraging new topologies (the cloud or REST), new technologies (Hadoop and its combination of massively parallel processing and distributed storage), and agile, iterative development and delivery techniques.

"People tend to look at MDM as just being really insanely complex. Once they get these [MDM offerings] installed, it's almost impossible to keep them moving at the pace at which the business is moving," says Matt Dahlman, Semarchy's vice president of North American operations.

"The sales cycle in MDM is noticeably longer, and the perception is that MDM projects just traditionally take an extremely long time: it's going to take at least 18 months, and it's almost always going to take longer than that. That's the type of attitude you see when [customers] go into it."

Semarchy is perhaps the upstartiest of the New MDM Upstarts. It was launched four years ago by a team of ex-Sunopsis technologists -- one of whom was Dahlman himself. (Sunopsis was an extract, transform, and load data integration specialist that Oracle Corp. acquired in 2006.)

Semarchy was founded in France and has a large presence in the EU; at this point, however, it's relatively new to the U.S. market. That said, it touts several U.S. customers, including a prominent purveyor of eyeglass apparel and a hugely successful Tex-Mex fast-food chain.

Its flagship MDM product is Convergence, which enables what Semarchy calls "evolutionary" master data management. Convergence doesn't depend in any way, shape, or form on Hadoop. Reltio, Liaison, and RedPoint Global make the Hadoop platform the linchpin of their MDM messaging: they each have different pitches, but all three claim to leverage Hadoop's schema-optional distributed storage and built-in parallel processing capacity to simplify data storage, management, and transformation.

Convergence doesn't run on Hadoop. Instead, it consists of homegrown technology that's optimized for the Oracle database. (Semarchy Convergence will run on Linux or Windows systems; it's programmed in Java.) Dahlman describes Semarchy's approach to MDM as "evolutionary" because it emphasizes an agile, iterative approach to master data management. The important thing is to start simple -- e.g., with the business domain that's the best candidate for MDM -- and to iterate from there.

"The term we use is 'evolutionary MDM,' which means that we're delivering value in days, not weeks or months. Our approach is to take the business domain that has the most manageable data set, and start with that one," says Dahlman. "We know we're not going to get everything right the first time; even in the best case, [the first iteration will] be incomplete. Realistically ... you're going to be over-matching this, or you're going to be under-matching that. Whatever the details are, we make it really easy to deploy updates. Based on the metadata of our model, we automatically update the MDM structure based on the changes of your attributes. We're able to automatically ... change the workflow and change the matching rules. We take care of the little nitty-gritty that just sits behind the scenes. We see that as really important this idea of an iterative approach."

The point is to do stuff, says Dahlman: i.e., to produce tangible results, to work iteratively -- using a software design model that makes it possible to accelerate iterations -- to get things right. Traditional approaches spend an enormous amount of time planning to do stuff; Semarchy's philosophy borrows from the disciplines of agile software development and continuous delivery. The latter approach is used in the DevOps (development and operations) development model.

"These customers started with a particular iteration, I listed customers and products as the most common MDM domains, but a lot of folks jumped in to location, or managing store hours, details like this. The French equivalent of the FCC is using Convergence to manage broadcast frequencies, so we're managing radio frequencies for them. This is one of the benefits of the MDM multidomain model. The point is, some of the folks started with one domain and after they had success with that, they moved to some other thing," Dahlman explains.

"The most recent simple example [of this] is working with [a purveyor of designer sunglasses], in Southern California. The original motivation [was that] ... people like to design their own sunglasses -- pick their own colors, pick the lens type, etc. In SAP, where they're managing their products, it's a long process to be able to handle this, and [in some cases] they weren't able to do it. What they're doing is they're defining base-level products in SAP as always but then they send it out to Semarchy and we go through our validation rules. In some cases, they end up importing the data back into SAP after they've manipulated it the way they need to. It gives them the flexibility to handle a really complex product."

This particular company started with a single business use case -- i.e., MDM for a single (albeit, particularly complex) custom sales application -- and has been expanding its use (and its MDM strategy) from there. "With evolutionary MDM, you don't have to get everything correct before you start; you don't have to worry about completely fitting it into existing processes, tying everything together," he concludes. "Think of it this way: we want to approach this problem like Salesforce, not like SAP. The problem is that most solutions approach [MDM] like it's [an] SAP [deployment]."

TDWI Membership

Get immediate access to training discounts, video library, BI Teams, Skills, Budget Report, and more

Individual, Student, & Team memberships available.