RESEARCH & RESOURCES

Model-Driven Architectures + BI: The Perfect Pairing

Successful enterprises recognize that data is their most valuable resource. Those that figure out how to move, refine, and process this resource for enhanced BI will be at a huge advantage.

By Bob Potter, Senior Vice President and General Manager, Rocket Software

Major defense contractors, automotive manufacturers, and government agencies have been using model-driven architectures (MDA) in recent years to build data applications for their core operational systems. Yet there is one frontier that MDA has yet to influence, and where the potential payoff could be dramatic: Business intelligence.

MDA dictates that users need to separate their business logic from their underlying platforms, and software should be designed for flexibility and adaptability. In most cases, BI data architectures remain rigid, so building a solid and scalable BI architecture requires the integration of a seemingly endless array of rules and attributes related to a huge variety of function-specific data that is often isolated or too locked down in a proprietary business system to be useful. More often than not, the enterprise tries too hard to model its entire business to the point where the BI system no longer reflects the current business model when it is finally deployed. As a result, organizations that set out to build cross-functional BI architectures are often setting themselves up for failure.

It doesn't need to be this way. Forward-looking enterprises that are willing to adopt MDA will develop BI applications that deliver useful, made-to-order data analytics rapidly. These applications can pave the way for better revenue forecasting, cash-flow stewardship, payroll optimization, benefits administration, risk management, and a wealth of other "back-office" decisions that ultimately improve fiscal health.

One byproduct of MDA is a dashboard that is conceptualized by business-users from various disciplines rather than one that exclusively reflects IT capability. This is because MDA, at its very essence, is a development approach that starts by asking the question "Why?" What is the purpose of the application we're building? What business problem are we trying to solve? Once the answers to these questions become apparent, MDA developers work backwards in collaboration with the end users they are assisting. This requires visualization and cross-functional consensus about what data is relevant and what insights are desired. All data sources must be considered and the architecture must be flexible and not limited to pre-defined drill paths and models of execution.

The application development platform, the coding, and the strategies for data sourcing flow from a comprehensive vision and clear understanding of BI needs. With a clearly articulated rendering of the end product, interdisciplinary project teams can work more effectively towards breaking the silos that prevent data integration and getting to a single version of data integrity. Once users are able to show how closely guarded data can be useful in ways that were previously unimagined, it becomes easier to engage important stakeholders in the process of redefining business intelligence and realizing data governance. A complete picture that reveals the wealth of available data can also reveal pathways for solving seemingly intractable business problems.

Unfortunately, because most BI data resides in silos, the architectures are too often built for narrowly prescribed purposes and modeled accordingly. What happens when the human resource department has a need for ERP data but can't access or process it in a meaningful way? With more agile warehousing, it becomes possible to iterate, scale, and maintain logical, business-driven models for retrieving, presenting and utilizing business information in real time across all departments. In other words, form follows function. MDA architectures allow data to flow from the right places with greater ease and precision.

Unfortunately, in today's world, it's often the other way around. The data models established by IT teams (without input from the actual stakeholders) don't reflect a holistic and nuanced understanding of enterprisewide BI needs. As a result, when a problem arises and an analyst queries a set of data that is unavailable, he or she goes to IT and hopes that the in-house developers can somehow create a bespoke solution, which ultimately will be as narrowly functional as every other data application the company uses. If IT can't help, the analyst has to write his or her own application, usually using Excel with incomplete access to the relevant data. All the while, quarter after quarter, the business problem continues and grows. Without timely insights (measured in hours, not months) from fresh, intuitive aggregations of data, the windows close on opportunities to quickly identify and fix problems.

With great frequency, we hear that we are living in a "data economy," but what does this mean? Economics is the study of finite resource allocation and how these resources get to where they are most needed as seamlessly as possible. For a successful enterprise today, data is the most valuable resource. Enterprises that figure out how to move, refine, and process this resource for enhanced BI will be at a huge advantage. However,, like any other economy, success or failure ultimately boils down to whether the infrastructure is built to harness economic potential. MDAs exist for this purpose and only this purpose.

Bob Potter is senior vice president and general manager of Rocket Software's business intelligence/analytics business unit. He has spent 33 years in the software industry with start-ups, mid-size and large public companies with a focus on BI and data analytics. You can contact the author at bpotter@rocketsoftware.com.

TDWI Membership

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

Individual, Student, & Team memberships available.