Introduction
Organisations can utilise data warehouses to integrate data, automate data operations, and employ the central repository to support all reporting, Business Intelligence(BI), analytics, and decision-making across the organisation.
However, creating a data warehouse architecture might be difficult. The design patterns involved in developing and managing the data warehousing environment can be complicated, especially concerning the scaling, the total cost of ownership, flexibility, and accessibility.
The Inmon and Kimball approaches are the primary strategies for data warehouse architectural design.
In this article, we'll dig deeper into what each author offered, underline the relative cons of each strategy, and compare and contrast the two techniques.
Inmon Approach
The Inmon approach is known as the top-down or data-driven strategy, in which we start with the data warehouse and break it down into data marts. These data marts are then specialised to satisfy the demands of other departments inside the firm, such as finance, accounting, and human resources.
Bill Inmon offered a top-down approach to data warehouse architecture which is known as the Inmon approach:
- Begin with the corporation's data model. Identify all of the data sources that the company has access to.
- Identify the essential entities (customer, product, order, etc.) and their respective linkages based on the data and understanding of business needs.
- Create a thorough, logical model using the entity structure. The logical model includes all of the properties of each entity, as well as their respective interactions and codependencies, in great detail. According to data modelling terminology, the logical model creates logical schemas for entity relationships.
- Build the physical model from the logical one. Extract data from various sources, alter it and integrate it into a normalised data model using ETL operations. Each normalised data model stores data in the third normal form to avoid redundancy. The data warehouse's core is the normalised data model.
- Create data marts for different departments. For all reporting needs, data marts are used to access data, and the data warehouse serves as a single source of truth.
Advantages
Inmon's architecture has several advantages.
- Flexibility: Inmon's approach adapts faster to changing business needs and data source alterations. Inmon's architecture is more versatile due to the ETL process design that results in normalised data. The architects alter only a few normalised tables, communicating the modification downstream.
- Single source of truth: Because of the normalised data model, the data warehouse serves as a single source of truth for the entire organisation.
- Less prone to errors: Because normalisation minimises data redundancy, both engineering and analytical procedures are less susceptible to errors.
- Completeness: Inmon's approach incorporates all Enterprise data, ensuring that all reporting requirements are met.
Disadvantages
Inmon's architecture has some drawbacks.
- Cost of initial setup and regular maintenance: The time and cost required to set up and maintain Inmon's architecture are far greater than the time and investment needed for Kimball's architecture. Normalised schemas are more challenging to build and maintain than their denormalised counterparts.
- Skill requirement: Highly skilled engineers are required for Inmon's method, which is harder to come by and more expensive to maintain on the payroll.
- Extra ETL is required: Separating data marts from the data warehouse necessitates the employment of more ETL processes to generate the data marts, resulting in increased engineering overhead.
518+ registered 
