The Data Warehousing

A data warehouse is a subject-oriented, integrated, nonvolatile, and time-variant collection of data in support of management's decisions. The main purpose of Data warehouse is reporting and analysis. It contains various components like retrieving data, analyzing data, extract and loads the data, manage data dictionary etc.

In data warehousing, there are two approaches that are most commonly used.

  1. B. Inmon
  2. Ralph Kimball

Both these approaches are different from each other. We will now discuss the Inmon approach of data warehousing.

Inmon approach of data warehousing is also known as relation approach. It is a top down approach. This approach is based on a flexible relational design. Generally it is believed that relation model is not optimal in defining requirements. In this the data are spread in parent-child relationship. So the data needs multiple look ups to fetch each and every record. This approach is based on 3rd normalization form. In this approach the data is designed relationally. This can be sued to support OLTP databases and data mining warehouses. In the top-down approach the data is transformed, integrated and then transferred to the operation data store. Then it is loaded into the data marts so that the end users can access it. There are some techniques that Inmon recommends. They are as follows.

  1. Each table may be physically merged by creating views.
  2. By using embedded arrays, data may be refactored.
  3. Data that is not been used regularly may be kept in separate tables.
  4. Derived data may be placed within database.

The inmon approach has the following levels.

High level

Here in the high level Entity-relationship diagram is used. These are same as the ER model of Databases. There are some special symbols that are used in Entity-relationship diagrams. Entities are shown in oval shape. Relationship are shown in edges and cardinalities are shown using arrows. ERDs are shown at the highest level of abstraction.

Mid level

Here for each major entity, area or a subject a mid level model is created. It contains 4 basic constructs. They are

  1. Primary grouping of data
  2. Secondary grouping of data
  3. Connector
  4. Type of data

Physical data model

It turns the DIS into relational tables. It adds time element in the keys and the physical data model. [2] [3]

Kimball Approach

The Kimball design is developed along the requirements of the end user. It is a bottom up approach. It bridges the gap between relational database and multi dimensional databases. In this approach the data marts are connected with the bus structure. These data marts can be viewed as data warehousing. The arrangement makes the data marts makes the data warehouse as virtual reality. Here the architecture is multidimensional. In this approach, the redundancy can be decreased by normalization of the data. This normalization is known as snowflaking. Kimball approach has data marts on tables. These tables contain data that can be easily aggregated. In this bottom-up approach, the data is extracted from the existing systems. All the data is stored into data store and later on more data is added or updated. The data store contains a fresh copy of data as it is updated. Then the data is summarized and is available to end users. [2] [3]

Common factors of both approaches

No matter which approach is used to develop data warehouse, the basic definition of data warehousing is always the same. The concept and way of thinking are different.

"It is an integrated, nonvolatile, and time variant collection of data to support management decisions." [1]. OLAP systems are responsible to integrate large amount of data from various business units. This data must be consistent. In both approach they agree that end user validation of data mart is significant. Both the approaches have the concept of staging but the names are different for each approach. In Kimball it is known as backroom and Inmon calls it warehouse. They both agree that independent data marts cannot fulfill the needs of end users on enterprise level for precise, timed and relevant data. [2] [3]

Differences in both approaches

In Kimball approach Iphysically data warehouse does not exist. The data warehouse consists of data marts that are conformed to dimensions. He believes that the raw data can transformed into the staging area. Kimball approach is a bottom-up approach while the Inmon uses top-down approach. The Inmon approach is more robust and flexible then the Kimball approach. The Inmon approach supports gradual development while the Kimball approach does not support that. Inmon approach focuses on end user requirements and so it can reproduce the data in a form that can be used to organization.


In my opinion Kimball model is more efficient and a better approach to create better design. Kimball approach starts with small modules and so it is easier to implement. Using this approach creates a quicker and simpler way to create a data warehouse for organization.


  5. Kimball, Ralph & Ross, Margy (2002). The Data Warehouse Toolkit: The Complete Guide to Dimensional Modeling
  6. Stephen J. Smith, Alex Berson ( December 1, 1997). Components of Data Warehouse,

Please be aware that the free essay that you were just reading was not written by us. This essay, and all of the others available to view on the website, were provided to us by students in exchange for services that we offer. This relationship helps our students to get an even better deal while also contributing to the biggest free essay resource in the UK!