Types of Data Warehousing Architecture
A Data Warehouse is referred to as a subject-oriented, integrated, time-variant and non-volatile collection of data which makes a management’s decision-making process much better as mentioned by the famous Bill Inmon. Once a decision has been made to build a Data Warehouse, there are three different approaches that can be taken into consideration and they are as follows:
1. The Dimensional Approach
This approach was introduced by Ralph Kimball, who is also known as the ‘Father of Business Intelligence’, who uses a Dimensional Model to model the data warehouse. In this approach data is segregated into two segments called ‘Facts’ and ‘Dimensions’. Facts are usually all the measurable data, where as the dimensions are the attributes which relates to these facts. An overview of the architecture behind this approach is shown below.
Advantages of Dimensional Approach
- Make information easy to understand
- Retrieval of information is highly efficient and fast
- More scalable
Disadvantages of Dimensional Approach
- Loading the data warehouse with data from different operational systems are complicated.
2. The Corporate Information Factory (CIF) Approach
This approach was introduced by Bill Inmon, who was also known as the ‘Father of Data Warehousing’, who uses an Entity Relationship Model to model the data warehouse. Usually the data is stored based on the normalization criteria, closely resembling the structures in the source system. All these created entities get converted to physical tables when the database is implemented. An overview of the architecture behind this approach is shown below.
Advantages of CIF
- Straightforward to add information to the database.
- More structured and easier to maintain
Disadvantages of CIF
- Joining data from different sources will be difficult due to the number of tables involved. and
- Requires substantial investment and long-term commitment
- Time consuming to be built
3. Data Vault Approach
This approach was introduced by Dan Linstedt in 1990s and is based on the concept of Hubs, Links and Satellites. The keyword hubs refer to the unique business keys in master tables where as Links are the relationships and associations between the Hubs. Satellites contain descriptive data which points to links. In this approach there is are quality check-points, validations or cleansing when it comes to the process of loading data from the source systems. An overview of the architecture behind this approach is shown below.
Advantages of Data Vault
- Provides extended functionality and flexibility
- Allows parallel data loads and process large amount of data
Disadvantages of Data Vault
- This model is only power user accessible
- Introduces many joins and integration issues
Now, let’s consider a high-level comparison across these methodologies.
