|
Data Marts and Data Warehouses
The debate continues as to which is better for an organization: data marts or data warehouses. Data warehouses have a very large
scope, one that can take millions of dollars to build and years to implement before value is realized by the enterprise. Our approach
is to build a data warehouse, one data mart at a time. A data mart is faster to implement and can eventually bring an enterprise to a
complete warehousing solution.Tackling the information needs of an enterprise by building one subject area at a time, we can begin delivering value almost immediately. By contrast, warehouse solutions may cause organizations to wait for years before any kind of ROI is realized. We begin building a data mart, using the subject area of sales as an example. The result, commonly referred to as a "Star Schema", would interact something like this:  In building the sales data mart, we not only record measurement values such as sales dollars and quantities, we also examine the dimensions of who, what, where, when, and why about those sales. While we define these dimensions, it is important to consider how they will be shared, coordinated, and conformed with other subject areas that will be developed at a later date. The information needed to satisfy the subject area is found in existing company production systems. The information is extracted and loaded into the data mart. After data extraction and loading take place, the data mart contains all of the pertinent sales figures and dimensions. The mart is then indexed and designed for data retrieval. Congratulations, the company now has a functional data mart that can begin delivering value immediately! Once the initial mart is up and running, other subject areas are targeted for a data mart. If the next subject area tackled is inventory, several shared dimensions can be drawn from the sales mart. For example, products are part of the inventory subject area, as is time. We are also adding a new dimension location. Now the process can begin again by finding the information necessary for inventory (in addition to the shared information from the Sales mart), extracting it, and finally building an Inventory mart. The result of this new addition to the "Star Schema" is shown below:  As depicted in the diagram above, there are now two operational subject areas delivering value to the enterprise. Best of all, the shared and conformed dimensions between them allow the users to begin asking questions across subject areas. Questions like, "How much inventory do I have of my best selling product? Or my worst?" With each mart added, the enterprise is becoming more coordinated, and even companies who are reluctant to take on a warehouse project will eventually have one. We just build it one data mart at a time. |
Prove It! See it for yourself. VantagePointSM rapid implementation process. Contact Janet Amos to find out how. |
| Contact Us | Careers | Site Map |
