Dimensional modelling reflects best practice

Dr Ralph Kimball, CEO of the Kimball Group, will present the afternoon workshop, "Best Practices for Building the Data Warehouse", on 13 November at the Straight Talk on Practical BI Techniques for Proven Results seminar, to be hosted by Sybase SA's BI Practice, in association with ITWeb.

Kimball says delegates will come away with an increased knowledge of best practice and a fundamental understanding of the goals of master data management.

In addition, he says, they will learn practical dimensional modelling techniques of ranging complexity, all of which will be demonstrated by means of specific industry scenarios.

The workshop will kick off with an introduction of the classic best practices for building effective data warehouses, Kimball says.

He will demonstrate how the most important data warehouse subject areas are implemented using the techniques of dimensional modelling.

The basic concepts of dimensional modelling will be outlined, including slowly changing dimensions, which show how changes are tracked in time; and conformed dimensions, which he says solves the integration problem.

Kimball will review the three types of fact tables and how to extend them in order to support real-time data warehousing. He will describe the content of these fact tables as well as give practical examples of how these three choices are adapted for different industries and how they appear in practice.

Kimball describes dimensional modelling as the "organisation of data that starts with the natural measurements that occur in all businesses and presents these measurements in a simple and intuitive way".

He says the measurements themselves reside in fact tables, and the context of those measurements in dimension tables. The power of this approach, he explains, arises from the fact that all fact and dimension tables look similar and have predictable paths to implementation.

Kimball says: "Dimensional models have proven to be the most understandable and performance-oriented organisation of data; other approaches are harder to understand and run slower when responding to queries."

The most important best practice approach is to systematically capture the true business needs and translate these needs into specific metrics, such as key performance indicators, and required data sources, Kimball says.

"Modelling disciplines and technology cannot make up for missing business requirements or inadequate data."

Once the business needs and the required data sources have been identified, he says, the basic rhythm of the data warehouse is extract, transform and load.

Related stories:
Identify BI users' needs
Customer behaviour key to BI
Data warehousing guru heads to SA