Time Dimension in Data Warehouse

In any data warehouse date and time plays a vital role in keeping track of when a particular event has taken place. However, we do not store the time information in the same way as we do in the OLTP database. In a dimensional model a separate time dimension is normally used. This time dimension contains separate attributes for year, month, and day and so on. This time dimension is initially populated with the data of a particular period like ten years. Whenever an event/transaction needs to be recorded in the fact table the time key from this dimension for the corresponding time is obtained from the time dimension.

Surrogate key for time dimension is auto incremented just like any other dimension but it is always helpful to have it the right format like for date: 05-Mar-2010 the time key should be 05032010. This proves to be extremely useful as queries can be done directly based on the time key and thus improves performance.

The following figures show how to create and populate time dimension in Sql Server Analysis Services (SSAS):

OLAP Video Tutorial

This is an exciting video I came across where OLAP concepts are explained in a easy to understand way using cartoons.

Dimensional Modeling

Dimensional Modeling refers to the technique used to design logical models for database which can support OLAP and Data Warehousing operations. It differs from the relational modeling or ER modeling in the sense that it can be used for data warehouse as well as relational databases.

Dimensional modeling is based on two concepts called facts and dimensions. Facts are usually numeric values or something that can be measured or aggregated. For example: monthly sales amount. Dimensions are groups of hierarchies and attributes that are used to define facts. For example product and its color. In a dimensional model a group of facts forms a fact table while a group of dimensions form a dimension table.