Articles: information management
Large data warehousing
Author: Wally Bock
According to David Wells, a principle analyst at the firm Ovum, large data warehousing projects carry a high risk, but a federated data mart approach can reduce failure rates and improve efficiencies.
Wells says that failure rates run up to 70% on projects that often cost millions and take years to implement. He believes that most projects that fail do so for organizational rather than technological reasons.
With the advent of data mart tool kits in the last few years, the large projects can be broken down into smaller federated operations. That increases the likelihood for success at the same time as it may decrease the overall costs.
WALLY'S COMMENT ... This is a variation on a theme that I heard Tom Peters state several years back "...it's very hard for anything big to get anything done."
If you're in a large organization, be very, very leery of super-big, super-large projects extending across an entire organization and taking a long time. They almost never work. Come to think of it, they hardly ever work in smaller organizations, either. If you must do a large project, break it down into lots of smaller projects.
A combination of Net/Web technology and changed leadership thinking are leading us to ways to put multiple smaller projects together. The smaller projects can be tailored, shaped, and driven by needs close to the front line, and pick up energy from local players because of that.
An excellent example of how this can work is the Eli Lilly Intranet. The Net for this huge, world-wide company is managed by a very small staff. Why? Because it's driven and run primarily by local users. The core technology staff is responsible for setting some basic standards and for helping local units make technological choices.
Another company who seems to have done that well is IBM. They've developed what they call a "no excuses toolkit" which sets out standards for Web publishing for IBM units. This gives IBM something of the best of both worlds. It encourages local freedom to publish and to make decisions driven by local and specific market conditions. It also makes possible a common look and feel for IBM projects regardless of country or market.
Now, what if you're a small organization? Then I think there are two learning points here. First, big projects are best done in small pieces. Get one done, move ahead.
The second learning point relates to data warehousing. The idea of bringing data all together so you can see relationships and use it in various applications is not something that's limited to large companies with huge budgets. That basic principle can drive a variety of projects in the small to mid-sized organization. Buy the concept, not necessarily the software and make the idea of data warehousing work for you.
This material originally appeared in Wally Bock's Briefing Memo Newsletter of 15 June 1998.
Copyright © 1998 Wally Bock
A Common Sense Approach to the Digital Age
Speaker - Author - Columnist - Consultant
Article used with permission