Data warehouse and its characteristics

A data warehouse is a subject-oriented, integrated, time-variant and non-volatile collection of data in support of management's decision-making process (W. H. Inmon). It is a central repository of historical data collected from several operational sources and organised for analysis and reporting.

Characteristics

• Subject-oriented: organised around major subjects such as customer, product, sales, not around day-to-day operations.
• Integrated: data from many sources is made consistent in naming, units, formats and encoding.
• Time-variant: data is kept as historical snapshots (5-10 years); every record has a time element.
• Non-volatile: data is loaded and read, but not updated or deleted in place; only refreshed periodically.

Other features:

• Large and historical: stores years of detailed and summarised data.
• Read-optimised: designed for complex queries, not for transactions.
• Multidimensional: data is modelled as facts and dimensions (star/snowflake) so it can be sliced and diced.
• Metadata driven: a metadata repository describes content and lineage.

Example: a bank warehouse integrates data from savings, loan and card systems, organised by customer, with 10 years of history, which is never updated after loading.