Data warehousing - a means to an end

To be successful, data warehousing should be implemented not as an end in itself, but as a means to an end. It must be designed in the broader context of a business intelligence strategy.

A data warehouse is essentially a database where data is collected to be analysed. Its purpose must be its defining characteristic - which leads to the need to define the corporate business intelligence strategy.

"The challenge of any such strategy is to create intelligence from data," says Bruce Jones, manager of sales support for the South African subsidiary of the global SAS Institute. "An organisation must first understand its business issues, define the required intelligence and then establish the value to the business. Only then can it develop its business intelligence corporate models."

In a typical large enterprise there may be over 100 different data sources running on numerous different platforms and devices.

All this disparate data needs to be extracted from these sources, then transformed so that it can be transported into a repository where it can be stored meaningfully. This process is known as ETL Q (extract, transform and load - with quality).

"Once data is intelligently stored, it can be used to generate reports and analyses that lead to actionable conclusions," says Jones. "SAS software is accepted globally in its ability to access data from any source on any platform. This is one of SAS' key strengths. It does not matter whether the data comes from the Web, legacy systems, ERP systems, campaign management systems, call centres or cellular phones - it can easily be accessed and turned into knowledge."

The data is then stored in the data warehouse according to business need. Thus, if the business need is for sorting or ranking, the storage will be relational. If, however, the business need is for trends, patterning or prediction, parallel storage will be used.

"Intelligent storage allows for proactive exploitation of the data, and the ability to automate for changing conditions," says Jones.

Business intelligence software enables business users to see and use large amounts of complex data. The tools are generally of three types:

• .         Multidimensional analysis software, which gives the user the opportunity to look at the data from a variety of different dimensions;

• .         Query tools, that allow the user to ask questions about patterns or detail in the data; and

• .         Data mining tools, that automatically search for significant patterns or correlations in the data.

"The tools chosen, and the user interfaces employed, should be based on intelligence needs," says Jones. Analytical intelligence goes one step further by looking into the future as well as the past. It does not merely look in the rear view mirror, but looks ahead to predict future happenings based on past performance.

"Data warehousing, then, is a means to an end," says Jones. "It is a vital link in the intelligence value chain which is made up of business intelligence planning, ETL Q, intelligence storage, business intelligence and analytical intelligence."

According to Jones, when data warehouses fail it is often because they are not seen in the broader context of a business intelligence strategy, and were designed with little understanding of business needs.

"Other reasons for failure include not enough senior management and business buy-in, too few permanent resource allocated, and too much emphasis on the IT components of the data warehouse," he says.