Ensure that the data warehouse is business-driven, not technology-driven

Define the long-term vision for the data warehouse in the form of an Enterprise data warehousing architecture

Provide flexibility to allow for significant changes in data warehousing functionality and integration of new business requirements

Identify best practice solutions and introduce new functionality in a piloted or prototyped manner

Reduce dependence on individuals who have key information

Do not build "virtual" data warehouses that access data directly from source environments and have no target database

Avoid "stovepipe" data marts that do not integrate at the metadata level with a central metadata repository, generated and maintained by an ETL tool. Replace existing stovepipe or tactical data marts by developing fully integrated, dependent data marts, using best practices

Buy, don't build data warehousing components

Use a data-modeling tool to perform logical and physical modeling. Ensure that a single logical data model drives the generation of all physical, target data models

Create a hub-and-spoke architecture, with the ETL tool as the Hub, to extract data from all data sources, resolve inconsistencies in data sources, and generate clean, consistent target databases for decision support

Use a 2nd-generation ETL tool to automate the extraction, transformation, and load functions