Data warehouses are specific databases used by decision-support applications.
Setting up and running a warehouse within a company follows specific processes, distinct from those used to create DBs.
As far as data extraction is concerned, warehouses are fed from a variety of data sources, such as databases, files and web documents. We need to ensure that all this data is consistent, and that it is regularly updated (refreshed) in line with the needs of decision-makers.
Warehouse structuring must be adapted to its intended use. The data models used to structure and manipulate conventional databases are generally unsuited to warehouses; new multidimensional models have been proposed to offer decision-makers a simple representation of data.
Warehouse data is often manipulated using data analysis software. For this reason, data must be selected according to certain criteria or dimensions, using ad hoc operators that aggregate or, on the contrary, divide them according to the axes of study.
Finally, the evolution of the warehouse is not only linked to the data extractions it regularly receives from sources. Its schema can also be modified over time to adapt to changes in analysis processes.