Building a data warehouse

Architecting a massive data warehouse can be daunting. This month's special report focuses on the best practices for building one.
Written by ZDNET Editors, Contributor
A management guide of how enterprises can build an effective data warehouse.

The term data warehouse is an ambiguous one. This is because applied in different contexts, data warehouses can take on different look and feel, and produce different results. A spreadsheet application running on a laptop can be one company’s makeshift data warehouse, as opposed to another that features the latest Oracle-SAP application tandem riding on a million-dollar storage array tethered to a redundant server.

But if the shape of data warehousing projects can come in extremes, their basic management blueprint will not be too dissimilar. And they invariably stem from a few common--and common sense--doctrines.

What are they?

The first step is to embarking on a data warehouse building project is to ask the pertinent questions. Asking the right questions, however, is not as straightforward as it sounds. A good approach is to start by interviewing both your IT department and users. You’ll find that obvious questions like “What do you want in a data warehouse?” will not yield answers as effective as “What kind of information is important to you?” can.

And if all this tedium is too much for you, an outside consultant or service provider can be your ticket. Or you can also partner with another company to build your next data warehouse. Whichever approach you take, however, understand that embarking on a data warehouse building project is a project unto itself.

Next is getting the key ingredients.

A high-level view of what constitutes your data warehouse building blocks starts with issues like whether to use off-the-shelf parts or develop from scratch, strategies for getting data into the data warehouse; applications and structure for data mapping; desired data granularity; and analytics. So what spokes will you use to support your data warehouse?

If you are using Oracle’s 9i, read our guide on data warehousing with 9i.

Ready to roll-up your sleeves and get to the task? Checklists can help you stay on the straight and narrow. Well, actually not so narrow: Here, we offer three checklists, on how to improve a data warehouse, smoothen implementation and blind-spots to consider.

Finally, niceties like business intelligence add the finishing touches. Here’s a guide to creating business intelligence in a data warehouse for SAP users.

Have fun building!

Editorial standards