The Matrix: Revisited

Try this versatile matrix to ease data planning and integration.

Matrix Extensions

One of the beauties of the bus matrix is its simplicity. You can reuse the familiar tabular format to convey other DW/BI "relationships." These extensions aren't a substitute for the DW bus matrix, but are intended as complementary opportunities to reuse the framework.

Opportunity matrix. Once the bus matrix rows have stabilized, replace the dimension columns with business functions, such as marketing, sales and finance. Based on the business requirements gathering activities, shade the cells to indicate which business functions (columns) are interested in which business process rows. This is a useful tool to assist with the prioritization of matrix rows.

Analytics matrix. While numerous analyses focus on the results of a single business process, more sophisticated analytics and data presentation tools, such as dashboards, require metrics from multiple business processes. In this case, reference the stable bus matrix rows but list the complex analytic applications as columns, shading the boxes to indicate which business processes are needed by each application to convey the prerequisite building blocks.

Strategic business initiatives matrix. As a variation of the analytics matrix just described, you can list the organization's key initiatives or executive hot buttons as columns mapped to the underlying process metric rows. This clarifies the need to tackle the underlying components in order to support the broader business initiatives.

Detailed implementation bus matrix. A single business process matrix row sometimes spawns multiple fact tables or OLAP cubes. For example, this occurs when there's a need to view the metrics at both atomic and summarized levels of detail or with both transactional and snapshot perspectives. In this scenario, the matrix rows are expanded to list individual fact tables or OLAP cubes, along with their specific granularity and captured or derived metrics. In this situation the standard dimension columns are reused.

We've described the benefits of creating a data warehouse bus matrix, but what happens if you're not starting with a blank data warehousing slate? Have several data stores already been constructed without regard to common master reference data? Can you rescue these stovepipes and convert them to the bus architecture with conformed dimensions? We'll discuss these issues in a upcoming column on conforming your nonconformed data.

Margy Ross is a President of the Kimball Group. She's focused on DW/BI consulting and education since 1984. write to her at [email protected].

Required Reading

Quick Study

The data warehouse bus matrix serves multiple purposes, including architecture planning, data integration coordination and organizational communication. Here's a brief primer and glossary of related terms.

  • Conformed dimensions. Dimensions are conformed if both have attributes sharing the same name, definitions and values. In other words, the dimension attribute values are drawn from the same data domain.
  • Data warehouse bus architecture. The bus architecture is based on standardized dimensions and facts that let separate data marts, fact tables or OLAP cubes coexist and integrate. In this context, the term bus is not a large motor vehicle. Instead, it refers to the term's early meaning in the electrical power industry (a conductor for collecting electrical currents and distributing them) and commonly used in the computer industry to describe the standard interface specification that lets peripheral devices usefully coexist. Ralph Kimball is credited with associating these fundamental bus concepts to the data warehouse delivery and presentation environment.