Data warehouses can be either normalized or dimensional. Normalized data is simple and lacks context, while dimensional data provides a comprehensive picture but can be difficult to navigate. Most data warehouses find a middle ground between the two.
There are two major data warehouse systems; normalized and dimensional. In a normalized structure, data is limited to a simple presentation of factual information. There is no context or background to the data beyond what the user is willing to relate. In a dimensional system, information comes in a context of other facts that show what the data as a whole is. In this case, there is a wealth of information available whether you want it to or not.
The two major data warehouse systems are the two extremes. In most data warehouses, a middle ground is achieved between these two. The actual descriptions are for the purest form of the style, though rarely encountered.
Normalized data is the easier to implement and manipulate of the two data warehouse systems. In this style, information is reduced to individual facts with no connection to other data. For example, a product serial number and product name are merged without additional information. The information is available to any user who might want it, but they have to do the work to make it mean anything.
To make sense of information in normalized data warehouse systems, the user collects related information to piece together a whole picture. To find a customer’s phone number, the above information can be linked to a serial number and account number of the person who purchased the item. Then the account number and name could be located. Finally, the name and phone number are found. Each of these steps is a separate database query made by the user to gather information.
Dimensional data is the exact opposite. In general, these data warehouse systems are the easiest for humans to use, but the hardest to change or manipulate. When information is gathered, everything is combined into one big data sphere. Instead of a serial and product number, an entire purchase invoice would be entered at once.
If a user were looking up a phone number in a dimensional database, the process would be different. The serial number would yield an entire history for that customer, names and dates of everything purchased and any service calls or returns. Additionally, every address and phone number the customer has ever used would also be directly available. The picture is very comprehensive, but perhaps so comprehensive that the required information is hard to find.
Protect your devices with Threat Protection by NordVPN
