Two different methods are there for loading data in dimension tables. They are as follows:
>> Direct or fast: In this method, all the keys and constraints are disabled prior to loading the data. Once the complete data is loaded, it is legalized against all the keys and constraints. In case the data is found to be invalid then it will not be included in the index. Plus, all the future processed on this data is skipped.
>> Conventional or slow: In this method, all the keys and constraints are legalized prior to the data is loaded. In this way, it helps in maintaining data integrity.