Chapter 1. What Is a Data Warehouse?

Congratulations?you've joined a team either building or about to build a data warehouse. Do you really know what you've gotten yourself into? This may seem like a stupid question, but I've found that what people call a data warehouse varies significantly. In fact so much so, that I treat the term "data warehouse" with deep suspicion. I apologize for being so skeptical, but I've found that over 90% of what people call a data warehouse is open for debate! How do you tell someone his or her data warehouse is not really one without starting a fight?

A few years ago, there was no such thing as data warehousing. Now we hear about data warehouses everywhere and everyone seems to be building them. Success stories abound in technical and business journals. Many database conferences now have a data warehousing track or special interest group (SIG). Moreover, businesspeople have bought into them "hook, line, and sinker." They all want data warehouses and data marts. Now, they even want them via the Web! These are most often referred to as Web houses. That's the good news?there's plenty of demand.

But, demand for something by itself is not sufficient justification. For example, I would like to retire from the workforce right now. But as my wife kindly reminds me, it does not make sense given our financial reserves. Far too often, I've seen data warehouses being built for all the wrong reasons:

  • Businesspeople ask for one since it's in vogue to have one.

  • The chief information officer (CIO) decides to sponsor a data warehousing project initiative.

  • Information Systems (IS) management submits a data warehousing proposal for funding.

  • IS management combines several reporting systems into a warehouse.

  • IS management renames an existing reporting system a data warehouse.

The point is that a true data warehouse should solve a genuine business need and thus be sponsored by the businesspeople who will benefit from it. Moreover, a true data warehouse follows some very specific design guidelines we'll be discussing in this book. Something is not a data warehouse simply because someone wants it to be or says it is.

Why am I making such a fuss over this? It's actually quite simple. The techniques espoused in this book will only work for genuine data warehouses. These exact same techniques will either not work or actually make things worse for entities that are not data warehouses. As such, this chapter is actually quite critical in terms of your data warehouse's success.