State of the art and data warehouse

Principle

The aim is to follow the path from the « from raw data to information ».

Give business the information it needs to make the best strategic and management decisions.

The business intelligence system is a standard enhancement of a classic operational information system.

The notion of a "business intelligence system" is now associated with concepts and a specific methodology that must be mastered to guarantee a project's success.

Target

Advanced reporting systems will let the user freely set criteria on the analysis axis. However, only star schema architectures allow to take full advantage of browsing functions such as :

Drill down : Analysis of a fact from global to details,

Drill up : Analysis of a fact from details to global,

Drill accross : Analysis of a fact accross various axis (dimensions).

The BI System

It it established that Business Information Systems (BIS) have the following characteristics :

The BIS is seperated from the production environment but depends on it for its loading,

Business intelligence information is independant of the technical characteristics of the operational environment (using data extraction, integration, information generation, dimensional formatting functions).

Business intelligence information is « business oriented » in its format and content,

The BIS must have a dimensional data model. This model is based on axes (or dimensions) that will ease user handling,

Business intelligence information must have a specific dimension, which is time (the opposite of an operational system that only provides an instant view),

The implementation of a BIS must allow for the project specifications to change regularly so that new or refined user requirements can be taken into consideration.

However, the reference architecture for BIS must be composed of the following subsets :

the Collection and Integration system,

the Distribution and Presentation system.

The data warehouse is at the heart of the business intelligence system.

The data warehouse is associated with a loading device.

The data warehouse must host the data corresponding to the different models that make up the business intelligence system.

Specific knowledge of the characteristics and objectives of each data model that makes up the business intelligence system is a major challenge when implementing such a project.

The business intelligence system is made up of the following consecutive model:

The integration model,

The distribution model,

The presentation model.

The role of the integration model is to receive the extracted, transformed and unified data from the production environment and to feed it into the distribution model.

The distribution model must give access to data according to information goals, process any query from users and enhance the presentation model performance.

The presentation model is the model used by the BI operator. It is based on the distribution model to display business intelligence information.

The choice of the technical solution will determine the ease of implementation and evolution of the project. The ability to handle large volumes of data is critical.

The BIS is determined by the integration model and the dimensional model. The technical solution for displaying the information (reporting tool) can be easily changed. One of the key challenges lies in mastering dimensional modeling and design concepts.

Modeling results in the following concepts :

View, Facts, Dimensions,

Grouped views defining a context (business grouped information),

The contexts must cover a domain and ease browsing and information generation.