Intelligent Enterprise

Better Insight for Business Decisions

RSS
Webcasts
Whitepapers
Subscribe
Home

August 10, 2003

Preserve the Data Warehouse

BI capabilities embedded in transactional systems degrade confidence

by Jim Gallo

Diogenes is one of my favorite historical characters. He's the philosopher who wandered around ancient Greece searching for an honest man but got so frustrated in his quest that he settled for the company of four dogs instead. Like Diogenes, business intelligence professionals have dedicated a large part of their working lives to finding truth — a single version of it — by creating data warehouses (DWs) and dependent data marts, metadata repositories, and of course, data architectures. Yet vendors in the transactional world are selling their own versions of BI/DW. So, which solution holds the "truth"?

I can't turn around without bumping into a salesperson offering BI/DW inside of an application. These technologies are in CRM, ERP, sales automation solutions, purchasing systems, and so on. And to make matters worse, most of these solutions have a different report writer for their transactional application (typically built by the application vendor) than they do for their BI solution (typically one the application vendor gets from an original equipment manufacturer).

Buyers often believe promises of a quick return on investment and a lower total cost of ownership because of slick packaging and the promise of quick implementation. Yet the very same vendors extolling the virtues of these embedded BI/DW products bypass the IT group when selling their point solutions, avoiding the critical "enterprise" test.

Can We Skip a Step?

Let's go back to the beginning: Why did data warehousing become such a viable concept in the first place? Remember when data and systems were in complete disarray and we survived through Herculean efforts and a web of extracts? Remember when the business users sat in a room arguing over whose numbers were right? Read any book on data warehouse, and you'll learn more than you ever wanted to know about why the concept of the DW is important. Data warehousing was born to eradicate these problems. BI was born as a way to add true value to the DW.

Now roll forward to today. The point solutions that offer embedded BI/DW have taken on a life of their own. And from what I've been reading in articles by many vaunted industry analysts, these products are gaining increasing credibility. In some cases, these are the same folks who, just a few years ago, were talking about DW as the be-all and end-all.

What are people thinking? I thought we'd learned that dependent data marts built around conformed dimensions really do save us time and money and actually do deliver a single version of the truth. We should also know by now that independent data marts fall apart when someone in the business asks to integrate the internal data. So how are these point solutions any different from independent data marts? They're not!

Take Back What's Yours

I encourage you not to take the ready-fire-aim approach of point solutions, but to revisit the fundamentals of BI/DW. By subverting the very notions of BI/DW and deploying point solutions, we're allowing the vendors and users to take what is not theirs. Yes, it's the business's data and systems, but it's our responsibility as BI professionals to ensure that the architecture and processes deliver value in both the short and long term.

Consider a recent example. One division of a business has an ERP installed, and another division is seriously considering one from a different vendor. Another group within a third division is looking at a purchasing system, and a corporate department is looking at a human resources system. All four of these solutions offer BI/DW as a "value-added" component. (Yet, they're separately priced!) All four use different front-end tools, and at least two of them offer a different "completely integrated" extract, transform, load (ETL) tool. (The other two vendors don't want to talk about ETL.) Now, to make matters even worse, the corporate executives want a unified view of data across all divisions.