Q: How long would a typical BI Portfolio project take?
Q: You've made it pretty clear that you consider the enterprise data warehouse to be the foundation of a company's BI portfolio. However, many BI applications claim to work without the need for any data warehouse at all; they simply pull and integrate data directly from various source systems. If this is true, why would we ever go through the hassle and expense of building an EDW?
A: It's true that most BI applications provide the functionality to integrate data from disparate systems, wrapping complex calculations in a semantic layer that insulates business users from database code. However, it's important to understand two important distinctions:
- Despite the sophistication of these semantic layers, they can't persist data that simply doesn't exist in the source systems (i.e., snaphshot measure groups, slowly-changing dimensions, etc.) This is especially true in the multifamily industry, where our systems typically store only the current state of an element.
- The formidable task of integrating data - matching up an entity in the financial system with an entity in the property management systems with an entity in the HR system for example - still has to be accomplished. No matter how fancy the BI application, they can't magically integrate and normalize data without well-defined business rules, error handling, data cleaning, etc. In other words, the same tasks still need to be accomplished with or without the use of a data warehouse. The EDW is a long way from becoming an endangered species.