Data warehousing is an integral part of generating Business Intelligence. Often the simplest mistakes can cause the failure of a data warehousing project. The article The Anti-Architect -- How not to design and roll out a data warehouse
Some companies may want to just make part of IT in charge of data warehouse support, but, depending on how important the warehouse will be to your company, permanent full-time employees might be necessary. Kimball says it may even be necessary to go as far as having "data warehouse support people ... physically located in the end-user departments."
Another unusual mistake that Kimball mentions is attempting to fully train users too quickly. The idea of continually training the users on the query tools doesn't seem too appealing, but the extra time and patience should ensure that your users fully utilize the data warehouse. He also recommends that you wait to train until the warehouse is up and running, so that the users can be trained with real data.
Kimball also debunks the idea that physically centralized data warehouse is best. He notes that a centralized system may be more expensive and difficult to implement. And he says it is especially less secure. Having a distributed data warehouse may require more ingenuity and improvisation, but overall, it might be a better decision for the company.
Read more of this article at http://www.intelligententerprise.com/020114/502warehouse1_1.shtml.
Ralph Kimball is the author of the best-selling The Data Webhouse Toolkit (Wiley, 2000).
For more information, check out searchCRM's Best Web Links on Data Warehousing.
This was first published in January 2002