Central Repository - Data Warehousing
Central vs. distributed
Almost all implementations for business functions, or so-called Applications, are stand-alone
solutions focusing on their specific problem. Many man-years of development are providing
excellent and sophisticated software to ease the end-users day-to-day work.
One problem they are all facing, namely the lack of input data comes with this island
approach. On the whole, it is finding the right data and creating an
interface feeding data into the application that proves the most challenging.
Central solutions, by definition, provide all data created and required by their inherited business applications. Within additional Data Marts, certain pieces of the overall data are duplicated for drill-down research. Changes to the source data have to travel through the complete set of storage locations until the new information becomes available to everyone.
These solutions are normally integrated and work harmoniously together, as long as all components suit your business needs. As soon as better tools exist that specifically solve a certain business problem, again the same issues arise.
"How can this external application be integrated into the integrated solution?"
The purpose of a central repository is to store all required input and output data of existing business applications. It would be better still if it were to only store information on finding data within each individual application.
Main advantages of a central meta-data model are:
- Central system to find all required business information
- Transparent understanding of your enterprise's data-flows
- Keep the master of specific information where it belongs, within the specific applications. Do this as
long as performance is acceptable
- Everyone uses the same data
- Store long-term history of aggregated data (e.g. daily and monthly results) for ten or more years in order
to have an extended information platform for research
- Create additional ideas out of your integrated information system
- Create only bi-directional interfaces between the central repository and each application, no network