Mastering Data Warehouse Aggregates: Solutions for Star Schema Performance

7.2. The Aggregate Project

Like any software development effort, there are many ways to organize a data warehouse project. The dimensional model lends itself particularly well to an iterative design/build cycle, which fits in with many popular methodologies. Through two or more carefully scoped iterations, the schema design is loaded with data, reviewed, and refined. This allows users to review actual work products against live data, a process that generates valuable feedback on the database design.

Iterative design is not unique to data warehousing. The approach has been employed in software engineering projects for decades. Iterative elements can be found in projects that include prototyping, joint application development (JAD), rapid application development (RAD), or any project approach in which user feedback to working software products is solicited. Iterative design can be incorporated into various project methodologies, including top down, waterfall, and spiral approaches.

Regardless of approach, the inclusion of aggregates in a data warehouse project will always require certain activities. This discussion divides the tasks among a common set of project stages: strategy, design, build, and deploy. The tasks can easily be remapped into any chosen project methodology. They can be used to incorporate aggregates in a larger data mart development project, or to organize a project that adds aggregates to an existing data mart.

In keeping with the approach to incremental deployment ...