If you are going to build a quick and dirty custom business intelligence environment, some things have got to give.
One thing is likely to be in the modeling area. One approach to a quick solution is to accept the source data model as the business intelligence data model. This will result in less functionality available in the user interface tool, but for those environments where the main obstacles to getting needed data are the limitations of the operational environment (saturated access, lack of or inconvenient batch windows), this can be effective.
Since the operational models will be used, practically verbatim, to get this quick model up, other sources that follow will feed other areas of the business intelligence model. This approach lacks integration but the ETL will be straightforward.
It also lacks scalability since, to get to the mass deployed BI environment, you need modeling for data access. I have yet to see such a schema in an operational environment, so this will have to be modeled eventually.
In the next tip, I?ll outline another approach. However, regardless of approach, the quick and dirty custom business intelligence environment is not a bad idea -- as the first step in a more robust environment to follow. It proves the concept, proves data can be moved from point A to point B, and gets the users started with a point solution meeting an acute need in a timely fashion.
For more information, check
This was first published in January 2002