A data warehouse is a database that has been specially designed for reporting. Data is collected from various different sources – mostly the operational systems - then reorganised in a way that enables quick reporting.

“What’s wrong with the database we already have?”. This is usually the first question organisations ask, and it’s the right question to ask. After all, why go to the time and expense of creating a database just for reporting, when the data you need is already stored electronically? There has to be a good reason to do this. And there is, several in fact.

Firstly, there’s speed. In a system with many transactions the database is optimised to write the data to the database as quickly as possible. This usually uses an organisation known as ‘normalised’. However, to get the data out quickly you need exactly the opposite, known as ‘denormalised’ data. What this means is that, to query the data you often need to look at several tables at once, and that’s inefficient.

Also related to speed, there’s the problem of several users trying to query the data at the same time as others are trying to put data in, which can slow the systems down considerably.

Variety of data sources. Another reason companies build data warehouses might be termed ‘accuracy’ or ‘completeness’. Having defined your KPIs or your reporting needs, you might find the data for this is in several places, not just the organisation’s main operational system. You may need some data from legacy systems, for example, or certain information vital to KPIs is only held on a spreadsheet. All of these sources can be combined in a data warehouse build whereas it would usually be against company rules - or illegal – to modify the operational systems to deal with this.

Additional calculations. It may be necessary to include calculations in your data, and these can easily be added to a data warehouse.

Keeping history. Finally, data warehouses allow you to store a history that often is not otherwise available.

All of the above are good reasons why organisations find it useful to build and maintain a data warehouse. A typical data warehouse is updated daily – usually overnight, but this need not always be the case. It all depends on the business requirements that you built it for.