A Dimension Table is an object related to an On-Line Analytical Processing (OLAP) Database. Its function is to define how something relates. An example would be an employee dimension table that has a field that relates to the employee identifier of the employee’s manager. This is called a parent-child relationship. It would allow a system to quickly build a structure of an enterprise’s chain of command. Adding another field that shows the employee identifier of the vice president that the employee is under would be an alternative rollup of the employee dimension.
Parent-child relationships are not the only way of relating data. The standard relationship, also known as a star relationship is when a series of fields in a table define how data is related. An example would be a User-Defined Accounting Key (UDAK). An account table might have the fields entity, department, and account. The fundamental problem with standard relationships is that there is the danger of ragged dimensions. A dimension is ragged when a major component is missing, like if we were missing department in an UDAK combination. Most modern OLAP packages, like SQL Server Analysis Services have settings and tools to eliminate this problem. Another relationship type is a snow-flake relationship. It is pretty much similar to a standard relationship, except that it uses multiple tables.