For a business person, the technologies of business intelligence are the reporting, analytical & forecasting tools. But in the background, these tools commonly depend on a very complex & powerful entity known as a data warehouse.
When committing to the establishment of a business intelligence environment, decision makers should understand the pivotal role played by a data warehouse. In addition, every information user in an organisation with a data warehouse needs to be aware of its potential to greatly enhance their decision making.
Why is a data warehouse the heart of BI?
A single business system (e.g. General Ledger) will usually provide a built in reporting facility. 3rd party reporting and analysis tools are often used for further insight into the data in that particular system.
Single business system reporting
However, very few business systems can build reports combining data from another business system. In order to build a report based on data from 2 or more business systems, a common approach is to extract the data from those systems into a separate repository where it is combined into a single set of data. This is the data warehouse. Reporting & analytical tools connect to the data warehouse rather than the original business systems.
Multiple business system reporting via a data warehouse
As data from more business systems is added to the data warehouse, its potential to empower a wide community of information users grows exponentially. Business intelligence relies on combining data from multiple or even all the systems in a business and the data warehouse plays a pivotal role in enabling this.
A data warehouse is not the only method of reporting on multiple business systems but is usually the best approach (see The advantages of a data warehouse).
An overview of how a data warehouse works
A data architect creates an 'enterprise data model' - a standard format for data across the organisation. This means a standard way of describing entities such as 'customer', 'product' etc.
Then, data is extracted from source business systems and transformed into this enterprise data model format. The transformed data is loaded into the data warehouse which is usually a specially configured mainstream database such as Oracle or Microsoft SQL Server. The overall process is known as ETL (Extract, Transform, Load) and can be very complex when widely differing source data are transformed into the standard model.
Loading typically occurs overnight but can be more frequent to support near real-time reporting.
The data warehouse can be directly accessed by business intelligence tools to produce reports, analysis & forecasts.
In some cases, a further stage of data processing is employed to produce OLAP Cubes. Well architected Cubes present the data in business terms and make reporting on very large volumes of data almost instantaneous. See Data Warehousing 101 for a more detailed technical insight to data warehouses.
