Mastering Data Warehouse Aggregates: Solutions for Star Schema Performance

7.4. Summary

The dimensional data warehouse is usually implemented as a series of data marts. When preceded by a planning project, incremental implementations fit together like the pieces of a puzzle, forming an enterprise data warehouse. When their coordination is not planned in advance, the result is incompatible stovepipes.

Aggregates are commonly excluded from the scope of the first data mart. This allows the team to focus on development of the new set of competencies required by the data warehouse. They can be added to the data mart in the future, once the team has developed a familiarity with the data warehouse lifecycle. Alternatively, every project may include aggregate implementation in its scope.

The process of designing and implementing aggregate tables mirrors that of the base schema. It can be understood as a set of tasks and deliverables in each of four major project stages: strategy, design, build, and deployment.

This chapter has enumerated the specific activities and deliverables necessary at each project stage. These tasks can be reorganized to suit any development methodology. They can be executed for aggregates alone, or incorporated into a larger development effort that includes the base schema.

Attention to aggregates is not limited to development projects. Once placed in production, some maintenance activities are required. As you have seen, specific responsibilities can be assigned to existing data warehouse roles, ensuring that nothing is overlooked.