DENVERA well-managed data warehouse can dramatically improve an
organizations decision-making capabilities, several experts said during a recent
A data warehouse is one of the most effective ways for information technology to add
value, said Kimberly Baker, vice president and general manager for data warehouse systems
at NCR Corp. of Dayton, Ohio. But failed projects have scared off some organizations.
Specialists at the Government Information Technology Executive Councils
Information Processing Interagency Conference last month agreed that data warehousing
projects must be carefully managed to provide valuable data that will become an integral
part of the organizations decision-making process.
Data warehouse projects face a high hurdle to achieve success, said David Isaac, a
project manager at MRJ Technology Solutions Inc. of Fairfax, Va.
While other projects can succeed just by getting the system to work, data
warehouse projects succeed only if they get the system to work and they demonstrate
ongoing value, he said.
Measures of success are less concrete with data warehouse projects, Isaac
said. Most IT systems have clear-cut requirements, he noted. But the goal of a data
warehouseto help people make better decisions fasteris more subjective, he
A data warehouse lets the organization manage data from multiple sources to answer
questions that had been unanswerable, Baker said. Data warehousing is a process, not
a product, she said.
John Price, chief of the Information Systems Branch at the National Institutes of
Healths Center for Information Technology, said NIH developed its data warehouse
after executives realized how difficult it was to get accurate information consistently.
Price said the NIH data warehouse provides information on budget and finance, travel,
procurements, property management and human resources.
Before the data warehouse was implemented, it would often take weeks to get answers,
and staff would take weeks to create a report that would often be outdated by the time it
was presented. Executives were relying on old information to make decisions,
Typically, the initial application of the data warehouse is for historical data, Baker
said. But as the data warehouse becomes more integrated into the organization, the demands
often increase, she said.
An advanced data warehouse can do predictive modeling, Baker said.
Price said the NIH warehouses success has spurred users to push for other
applications. The more you put on, the more people want, he said.
Data must be managed properly before a data warehouse can provide a real return
on investment, said Stan Becton, president of Acton Burnell Inc. of Alexandria, Va.
One critical step is to establish a framework for the data warehouse as part of the
organizations IT architecture, Baker said. That framework will include specifics on
the data sources and users.
That all has to be laid out at the beginning, she said.
The IRS last year created a data warehouse to help identify tax cheats, said James
Alzheimer, assistant to the director in the IRS Office of Compliance Research.
There were problems putting the IRS data into a usable format, he said. IRS needed
a data warehouse designed for users with little computer proficiency.
The agency decided to use data sampling that let the agency design a fast system that
met all the essential requirements, Alzheimer said.