Value-Driven Data Warehousing

A'dam - May, 24th 2007 - "I am tired of IT people. They'll talk to you for half an hour about what they do, but by the time they're finished, you still don't have any idea what they've told you or why it's important," so complained a friend of mine in a recent conversation. What was the irritant giving rise to my friend's frustration? Ahem … my own ineffective attempt to explain to him what it is that I, a data warehousing/business intelligence (DW/BI) practitioner, do.

Sometimes the right metaphor is helpful. It can clarify abstract concepts for the uninitiated and, even for the expert, be a means for synchronizing designs and vocabularies and analyzing problems. To that end, let me propose a metaphor for describing what DW/BI is all about and, perhaps more importantly, suggest where the field is broken: the metaphor is that of an information supply chain.

This metaphor - the information supply chain - is in some ways a simple extension upon thoughts already well developed by others, most notably Bill Inmon's Corporate Information Factory (CIF), and yet the supply chain metaphor doesn't seem to enjoy much use in the marketplace, at least not as a framework for holistically describing both the DW and BI, together.

Inmon describes the enterprise data warehouse and the various satellite systems that surround it - operational data stores (ODSs); extract, transform and load (ETL); data marts, etc. - as similar to a brick-and-mortar factory. The factory inputs many disparate raw ingredients (data from all the company's operational systems), subjects those raw data inputs to a series of processes wherein the data is refined and integrated (ETL) and finally spits out a finished good on the other end (corporate information). Whether a practitioner leans more toward Kimball's dimension bus architecture or Inmon's CIF as the guiding vision for how to build a data warehouse (or like this author you feel the two approaches are more complementary than contradictory), Inmon's metaphor is simple to understand and helpful for explaining what data warehousing is all about.1

Extending the metaphor to encompass the field of business intelligence is beneficial, since definitions of BI seem pretty slippery, and it can be unclear how BI overlaps or is different from the enterprise data warehouse. I believe the reason for this confusion, at least in part, comes down to how the term "BI" is often used - that is to say - it is used to convey purpose, rather than function. BI is typically defined as: the collection of tools and techniques used to help companies make better decisions. This definition is very broad. It expresses a worthy purpose, but it casts a huge net within which almost anything can fall and thereby blurs important distinctions.

Read the full article: DMReview.com