Data warehouse (DWH)

A data warehouse (DWH) is a relational database that stores data in a way that it can be easily and performantly queried and analyzed.

Use cases for a DWH include

Arranging historical data
Provide Regular reports
Assist in decision process and decision making
Analyze data
Mine potentially valuable information

A typical DWH environment includes

ETL (or other processes to collect data into the DWH)
An OLAP engine
Client analysis tools
Applications that distribute the data to users

Methodologies

Bill Inmon is regarded by many as the father of data warehoses. He defined a datawarehouse as

a subject-oriented, integrated, time-variant and non-volatile collection of data to enable the decision-making process
Inmon, Building the Data Warehouse, Wiley, 3rd ed. 2002

Ralph Kimball is another pioneer in the field of data warehouses. According to him, a

data warehouse is the conglomerate of all data marts within the company. Information is always stored in the dimensional model.

Inmon's approach is considered top-down, Kimballs's bottom-up.

Relationship to data-mining

Data-warehousing: For Data-mining to take place, data from many sources (billing records, phone call records, orders etc) must be gathered together and organized in a consistent and useful way. Data warehousing allows the enterprise to remember what it has noticed about its customers.

A data warehouse can track customer behaviour over time and becomes the memory of a company. So, it makes it possible for data miners and scientist to find trends. (Data Mining Techniques, p. 5)

Relation to data marts

A data mart is a subset of a data warehouse which is optimized towards a specific subject (usually with organizational character such as finance, marketing or sales).

The conjunction of all data marts therefore equals more or less the data warehouse.

Data warehouse challenges

With all the benefits of data ware houses, I believe two challanges need to be kept in mind when implementing data warehouse:

A data warehouse stores redundant data
Data quality

Redundant data

Datawarehouses store a second copy of corporate data.

So, there needs to be a mechanism to keep the copy in sync with the operational data store.

There are some prodcuts, such as IDAA, that try to solve this problem.

More recently, a buzzword to overcome this problem is data virtualization.

Data quality vs Business agility

Running a data warehouse forces its operators to trade data quality off against business agility.