The loading of the dimension table can be represented in two different ways:

Conventional (Slow): In this way, the constraints and keys goes for validations that are done against the data. These validations are performed before the loading of the data takes place in the data warehousing. This ensures the data integrity that is maintained throughout the loading phase. As the name suggests it is slower method that has to be implemented. This way deals in too much complexity in creating the database.

Direct (Fast): In this, all the constraints and the keys that are used with the data are disabled. The validations are performed after the data is loaded in data warehousing. These validations include everything that are applied on constraints and keys. Data are checked for invalid bit or dirty bit, if anything like this is found then the data is not included in the index and all the processes that require the data are skipped.