BI On A Budget

Cobbling together technology from several sources and vendors has been the traditional approach to BI. The reason is simple: Database vendors have historically provided very little BI functionality. Consequently, many BI applications use SQL syntax to extract data from the database in bulk, stage it at the application layer, and then transform the data into required granularity. Then, the BI application provides the added value.

This traditional approach continues not out of necessity, but because it's the model most familiar to project planners. Today's leading RDBMS vendors, including IBM, Oracle, Microsoft, and NCR/Teradata are building BI functionality into the database. They've been moving steadily down a path of supporting BI functions inside the database, including data mining, OLAP, ETL, spatial analysis, and advanced statistical and analytic functions for regression, covariance, sampling, ranking, and more.

The benefits of exploiting the inherent BI function of your RDBMS are significant and varied:

In-database BI functions not only lead to a "single version of the truth" across the enterprise — regardless of end-user tool or application — but also reduce costs. You no longer have to establish infrastructure (CPU resources, data storage, network bandwidth, and software) to support the traditional BI model. Instead, the database itself does data preparation, in an environment established and optimized for precisely this role.

Focus on What You Need

As a solution strategist and data architect, I'm often brought into accounts to establish equilibrium between the organization's current technology, the budget available, and the new requirements. Over half of my clients have opted to invest in full-featured, big name products. Unfortunately, many use only a fraction of the product's capability, which translates into a very expensive implementation as opposed to a meaningful, effective investment. For others, price is more influential; they've tried to implement less costly technology but found that the approach created more work and more costs in the end.

The balance comes when you can reshuffle tools to exploit their features and supplement them with technology that can fill any gaps discovered on the way to implementing your desired BI applications. The supplemental tools need to be relatively inexpensive when the BI budget is tight. (As a side note, I try to avoid completely dropping existing technology.)

The experience of reshuffling and rebalancing BI environments has led me to always examine the three key architectural approaches described in this article: nontraditional combinations of technologies, exploiting Web services, and expanding the use of in-database BI functions. There are other ways and means to achieve BI on a budget, but investigating these strategies first will definitely reduce investment costs while providing robust BI solutions.

Michael L. Gonzales [[email protected]] is the president of The Focus Group Ltd., a consulting firm specializing in data warehousing. He has written several books, including IBM Data Warehousing (Wiley, 2003). He speaks frequently at industry user conferences and conducts data warehouse courses internationally.