OLAP Methodology

The Data Warehouse Lifecycle resembles most systems development lifecycles in that the warehouse requirements must be gathered. Then they must be analyzed. From the analysis flows the design, which is the input to the build and load processes.

Note: This process assumes that the data warehouse is being produced from the enterprise/application database pair and the examples reflect that. However, the methodology will work perfectly well if you have to design a data warehouse from legacy application databases. The steps are the same. The deliverables are the same. It is just more difficult to work with standard application data models, that are not component-based.

The Business Intelligence (BI) Workshop is the process that produces the data warehouse requirements. The inputs to this process are the existing business and system documentation and the subject-area knowledge of the business unit managers and staff.

Prior to the workshop, the business users are directed to bring their business system operating manuals and manual procedures documentation to the session. System documentation should be provided from the systems support group.

Data Mart Analysis is the process that produces the model of the data mart, the star schema. The inputs to this process are the outputs from the BI Workshop.

Data Mart Design is the process that produces the definition of the data mart fact table and dimensions. The inputs to this process is the star schema from Data Mart Analysis, plus all supporting documentation.

Query Design is the process that produces the initial set of BI queries. This activity should be executed with the close participation of the business managers and users. It is during this process that the user receives their initial training in the use of the data mart and building queries.

The Business Intelligence Workshop is a process that is conducted as a user-driven, joint application development (JAD) workshop and facilitated by the OLAP Architect. The workshop produces the requirements for the data marts and the business models in the form of the Business Intelligence (BI) Survey.

The workshop should contain up to 20 managers and staff members from the business. There are a couple or three people from the OLAP Architects group. There may be some individuals who are curious about the process and wish to audit the workshop, but they should be considered observers and remain quiet in the back of the room.

There are several formal deliverables produced within the BI Survey:

        • The Business Context Model identifies the organizations, people and systems
          participating in the business problem.
        • The Entity-Relationship Model documents the business objects that support these           relationships.
        • The Object Examples identify document the elements of the business objects.
        • The Business Objects Model identifies candidates for the data warehouse.
        • The BI Queries identifies the requirements of the business models.
        • The Candidates Data Marts Table identifies potential data marts.

One of the underlying objectives of the BI Workshop is to produce a common understanding of the business problem, business problem elements and candidate data marts. The OLAP Architects and business unit managers and staff have to produce the formal requirements for the data marts and business models collectively for the process to be successful.

The business users are teaching the technical people the business. The technical people, in turn, teach the business user what's in the marts and how to use them to achieve their goals.

There should be three workshops, with several weeks between them. The first workshop produces the draft BI Survey. The second workshop produces the revised BI Survey. The third workshop produces the final BI Survey. The OLAP Architects use the intervening weeks to work the problem, ask the business user any follow-up questions and produce the deliverables.