Data warehouse rescue
When the health insurance division of a well-known South African insurance group encountered difficulties with their newly implemented data warehouse, they contacted PBT Group to help resolve these complexities, commencing a rescue mission with winning results.
The company procured QlikView based on its ability to develop high quality reports with basic analytical capabilities in relatively short periods of time.
The problems that they encountered were:
- QlikView struggled with the number and complexity of the business rules
- The quality of the data
- The volume of raw data
- Performing longitudinal analysis on the data
PBT Group’s intervention
Some of the biggest challenges included the number of undocumented business rules. Knowing that it is unlikely to document all business rules upfront, we adopted a more agile approach to development.
We began by creating a solid data model, framework and architecture. Then we started populating the data warehouse, testing as we progressed. No project is without challenges, and here the production migration was delayed by an unavailable production server. In addition, the development server was underpowered and as a result some fact tables could be loaded and tested on a subset of data. However, everything passed UAT, with the caveats imposed by environmental constraints.
The project resulted in a well-designed data warehouse that will eliminate multiple versions of the truth, hide complex business rules from the end-users and therefore make analysis a lot easier, even for novice users.
Despite constraints and challenges, we developed a data warehouse with about 20 fact tables and even more dimensions in a complex environment in about 650 labour hours, which included a POC. We have now been asked to bring in additional data into the data warehouse and to continue with the QlikView front-end development.