In today's fast-paced world, everyone is under constant pressure to produce at a rapid pace. When we eat at a restaurant, we expect to be served quickly; when we are shopping, we don't like waiting in line. Similarly, our business users expect the business intelligence team to deliver requested capability quickly, and anything that is perceived as delaying the result is not well received. Data modeling is one development step that does not directly generate code, and, adding insult to injury, we try to insist that the business user participates in the modeling effort. In this setting, it's not surprising that we're often asked to skip developing a business data model. It leaves us with two challenges: explaining why this data model is important and finding a way to develop the model as quickly as possible.

Business Data Model Importance

A business data model (a.k.a., conceptual model or enterprise model) is a representation of the information used in an enterprise from a business perspective. The model encompasses the business rules about the information and is independent of organizational, procedural and technical constraints. At any point in time, there is only one business data model in an enterprise. This model is portrayed in an entity relationship diagram and is represented in (at least) third normal form. In that form, there is no data redundancy, and business rules are enforced through relationships and referential integrity. Four areas in which the business data model helps are: designing the physical databases, describing project scope, guiding data profiling efforts and supporting transformation mapping.  

Physical database design basis: Just as organizational charts help classify people within the enterprise, the model helps organize data and provides a foundation for developing physical data structures. Databases that are designed based on the business data model are inherently consistent with each other because they are based on a non-redundant set of business rules. For transactional systems, a subset of the business data model should be used for designing the databases. For BI environments, the model provides the subject orientation that can be applied regardless of the architectural approach. It can help build a normalized structure for a hub-and-spoke architecture and provides the foundation for the business rules affecting conformed dimensions in a bus architecture. 

Project scope definition: BI projects are defined based on the data they need to address. The business data model can be used to identify the affected business entities and attributes. This is particularly important when there are multiple simultaneous projects so that overlaps and gaps can be identified and addressed. Other related advantages include identifying the data stewards who need to be involved and providing information to assist in estimating the effort.  

Data profiling support: One of the critical steps in data warehouse development is data profiling. This step examines the source data to understand quality deficiencies so that appropriate remedial actions can be taken. The business data model provides authoritative definitions and business rules for data elements and entities. These provide the basis for determining what needs to be profiled and for comparing the actual data to what it should be (according to the definition and rules) so that quality deficiencies can be identified and addressed. 

Transformation and integration support: While the business data model should have been the basis for operational database designs, this is not the case for most legacy systems. (Most companies did not develop the business data model prior to developing their legacy systems.) The business data model can provide the focal point for developing integration and transformation rules. The source data stores can be mapped to the business data model, and the business data model can be mapped to the target data store. With that information, the physical transformation and integration logic can be developed. 

These are only a few of the benefits that the business data model can provide a BI project.

Model Development Approach

Developing the full business data model using the traditional approach can take several months, and the business will not see a benefit until the model is used to help development efforts. A better option is needed. Two approaches have successfully been applied to quickly provide business data model support for BI development activities.  

The first approach is to develop the model incrementally based on the scope of each data warehouse segment. This approach limits the modeling activities to those that support the specific project. The business data model expands over time as more data is added to the data warehouse. There is some risk that as the model expands, rework may be needed due to relationships that are subsequently discovered, though this risk can be mitigated by using experienced modelers and gleaning information from publicly available data models.

The second approach takes a more global view. With this approach, we envision the scope of the data warehouse for the coming two to three years and develop an entity level business data model for that scope. This can often be accomplished in a few weeks. With this model in hand, it is attributed incrementally based on each project's scope. A little upfront effort is needed with this approach, but the rework risk is reduced because the relationships for several increments are identified upfront with the expanded view. 

The business data model is part of the foundation for a BI program. If this model does not exist when the BI program is initiated, it should be developed incrementally, possibly with an entity-level model for the first few years of the program created first. Once the model is in place, the project team will benefit in several areas, including database design, scope definition, data profiling and data capture, transformation, and integration design and development.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access