Business Intelligence, Data Warehousing, and Data Marts

The terms Business Intelligence, Data Warehousing, and Data Marts are used interchangeably by many people so I thought it would be good to explain the differences.

The data warehouse is the repository for all the data to be analysed and reported on without the means to do so. A good analogy would be SQL server 2000 before reporting services was added to it, it could store all the information you needed but there was no mechanism for getting information out.    

Data Warehouses can be really big and can provide analysis and reporting for the whole enterprise.  These big projects got a bad press for taking too long to implement and for being too difficult to change.  The response to this was the data mart, which is not simply renaming the same thing to deflect criticism. 

A data mart is a subset of information in an organisation that meets a specific business need or is just for a part of the business.  Individual data marts are implemented on the basis of need and return on investment. The idea is that the individual data marts are joined together to form the whole data warehouse.

This sounds great but has been beset with problems for 2 key reasons:

1. Each department does its own thing to implement tactical solutions, which leads to 'mart madness' where there are several implementations using different technologies.

2. Each data mart is designed differently and so they cannot be joined together for consistent enterprise reporting.

This really is just two sides of the same problem, but it is so important I thought it would be good to mention twice!

So just like every other area of IT, architecture and design must be properly addressed or all the later stages will fail.  

Technorati Tags: BI, Business Intelligence, Data Warehouse, Data Mart