The basic fundamentals of Data Warehousing are the subject of a huge array of books, and so I am not going to go through them again. Instead, I want to just summarise with a set of points pertinent to Data Warehousing
- A company should have just one Data Warehouse (I define ‘company’ here as an entity with a defined and unified vision and set of objectives, strategies etc.).
- Data Marts, if present, should get all of their data from the Warehouse.
- If the Warehouse is deemed ‘enterprise wide’ then the data model in place should be ‘near TNF’.
- The Data Warehouse should be maintained automatically with ETL or ELT tools.
- Reporting tools should be used to optimise reports, queries and result distribution.
- There should be a development road map and steering committee with all Data Warehouse projects, aligned to agreed business initiatives.
- The support organisation should be ‘appropriate’
- ROI should be understood and monitored.
- The Warehouse should be used for reporting but also, and very importantly, for innovative and high-payback queries as well.
- Security/privacy should be considered and balanced with ease of access.
- Independent Data Marts should be subsumed into the Data Warehouse and decommissioned.
- The Data Warehouse should be treated as a core corporate resource.
- The Data Warehouse platform should be scalable in all aspects of the word.