Answer:
A Data Warehouse is a corporate database that integrates purified information from the various sources in the organization. This information must be homogeneous and reliable, it is stored in a way that allows its analysis from many different perspectives, and that it is in turn of the optimal response times. For this, the information is highly denormalized and modeled in a way quite different from transactional systems, mainly the star models (star scheme) and snowflake (snowflake scheme) that we will study later.
A Data Warehouse is much more than what we have commented so far. According to Bill Inmon it is characterized by being:
Theme-oriented: data is organized by topics to facilitate user understanding, so that all data related to the same element of real life is linked together. For example, all customer data can be consolidated in the same table, all product data in another, and so on.
Integrated: the data must be integrated into a consistent structure, eliminating the inconsistencies involved between the various operational systems. The information is structured in various levels of detail to suit the needs of users. Some of the most common inconsistencies that we usually find are: in nomenclature, in units of measure, in date formats, multiple tables with similar information (for example, having several applications in the company, each with customer tables) .
Historical (time variant): the data, which may vary over time, must remain reflected in such a way that these changes are consulted and the reality that was at the time they were stored is not altered, thus avoiding the problem that occurs in the operational systems, that specifically only the state of the present business activity. A Data Warehouse must save the different values that take a variable over time. For example, if a client has lived in three different cities, he must specify the period he lived in each of them and associate the facts (sales, returns, incidents, etc.) that occurred at any time to the city in which He lived when they occurred, and did not associate all historical facts with the city in which he currently lives. If a customer, during all the time we have been selling has gone through three civil states (single, married and divorced) we must know what marital status he had at the time we made each of the sales.
