A major retailer had developed and built a data warehouse over the years, based on individual projects leading to redundancy and slow performance. The root cause for this failure was the way the data warehouse was 100% funded on a project base. This led to erratic situations like the use of six (6!) different time dimensions and a ballooning database of 1.500 tables. That, and a few technical quirks made business intelligence and reporting a real pain.
Retail is detail but in datawarehousing, the retailer needs to see the entire picture as well.
The objective of the mission was to improve performance and reliability. So we analysed the architecture, proposed a mix of data warehouse funding from an infrastructure perspective mixed with project based funding and proposed best practices for the design and development of the warehouse infrastructure.