Data Warehouses are structured to hold large amounts of information, generally in what is known at as a Star Schema, or Dimensional Modeling form. Conventional database systems use highly normalized data formats so that they will execute transactions and queries as fast as possible, in minimal time and space. Data Warehouses use a more relaxed format. De-normalization is encouraged so that the raw data will make sense to users as they are exploring it. For example, rather than having a single record in a table contain customer information, that information might be replicated across a whole series of tables to simplify querying for users.
OLAP tools are generally designed to work with de-normalized databases.
Data being pushed into a warehouse is usually "staged". Data staging occurs when a periodic process reads data from sources (often the business' primary databases), scrubs this information for quality, de-normalizes it, and writes it into the Warehouse.
Data Warehouses are usually accessed (queries) via "data marts", which are purposes-specific access points to the warehouse. Data Marts are designed to answer the probable queries of a given kind of user.
Normally a data warehouse will not store reliable and current information on an individual business activity. It is normally used for collective processing only. Computing in data warehouses are often refered to as On Line Analytical Processing OLAP?, contrasting it to On Line Transaction Processing OLTP? used for normal business activities. Data from Enterprise Resource Planning (ERP) systems and other related business software systems is imported into the data warehouses periodically for further processing.
![[Home]](wikipedia.png)