Data Warehouse World: Checking out success, failure, and myth
Building data marts before developing a data warehouse can be one of your biggest mistakes
New York (July 30, 1998) -- To avoid a million-dollar software implementation mistake, Data Warehouse World was the place to be this week.
That's because measuring the success or failure of projects and debunking myths surrounding software trends were the main topics of discussion at Data Warehouse World and related trade shows.
Show organizer Digital Consulting Inc. (DCI) combined Data Warehouse World with its Sales Force Automation, Marketing Automation and Enterprise Resource Planning Summit shows, expecting about 11,000 attendees in all.
Data warehouse and Enterprise Resource Planning (ERP) projects, especially, can cost millions of dollars for large companies. In an effort to guide users and help them avoid common software implementation pitfalls, several keynote speakers focused on debunking common beliefs that may lead to costly mistakes.
The tone of several panels was set by Bill Inmon, chairman of the board of Pine Cone Systems Inc. and a data warehouse consultant. Inmon was co-founder of Prism Solutions Inc. and is often referred to as the "father of the data warehouse."
"I've been reading that 70 percent of data warehouse projects fail," Inmon said. "This is poppycock."
In both the general press and trade press recently there have been reports that a majority of data warehouse projects fail, which may turn off many companies to technology that can give them a strategic and beneficial business edge, Inmon said.
Close examination of the research projects that are the basis for these reports of failure leads to a very different conclusion, he said.
In a heated preamble to his keynote speech earlier this week, Inmon said he contacted several companies that have researched data warehouse projects. Some researchers say a data warehouse project is a failure if the goals of the implementation change within a year, he said.
"This is like saying...Bill Gates is a failure because he didn't graduate from Harvard, because his original goal was to go to college and get a degree," Inmon said.
However, companies can make costly mistakes, he said. Building several data marts before developing a data warehouse can be one of the biggest mistakes a company can make, he said, refreshing a theme that has been discussed at several Data Warehouse World shows.
Data marts are a subset of data warehouses, containing less data and less history, and are customized for a specific department. Data warehouses contain more detailed data, which the data marts summarize for the purposes of different departments in a company.
Building the data marts before the data warehouse can make it difficult to integrate different departmental-level data. Data marts require a separate interface to be built for each data source -- for example, transaction processing applications.
Data marts roll up detailed data in different ways. "The result is that the sales department will give you a different number for your revenue than the marketing department...you end up with a big mess," said Inmon.
On the other hand, there are some trends that can save companies big money, and allow them to facilitate analysis by "out-of-the-box" thinkers, said Inmon.
To save money, companies with vast amounts of historical data should consider "near-line" data storage -- storing rarely used data on tape, for example, rather than on disk, Inmon said.
Companies might also consider creating what Inmon called "exploration warehouses" where analysts can do large queries against huge amounts of data. "If someone comes to you and says `I've been thinking about this for years and if I can do a 72-hour query I might be able to prove this idea that could save the company millions of dollars,' you probably are going to put that person off," because a data warehouse normally can't be tied up for that amount of time, Inmon said.
But a company could lose a valuable new way to do business by not allowing such out-of-box thinking to flourish, he said. To accommodate such requests, companies should consider exploration warehouses that can do large queries very quickly. Such data warehouses perform analysis quickly because they store the data in memory, rather than on disk. They use what is called token-based technology, which uses a technique akin to "bit-mapping" data to shrink it down to fit in memory, Inmon said.
For ERP software, there are also a number of trends that could save users money -- and some myths that could cost companies a strategic advantage, according to John Bermudez, a keynote speaker at DCI's ERP Summit here and group director for enterprise research at AMR Research, a Boston-based consultancy.
One of the main myths about ERP implementations is that most of them are failures, Bermudez said.
"The vast majority of companies successfully replace their existing systems and improve processes," said Bermudez. However, he also said that "many companies fail to achieve all their goals because they are unable to sustain (an ERP) project."
The reason for this is often that key people in an ERP implementation, if successful, are promoted or wooed away by a company's competition -- leaving people unfamiliar with the implementation to manage it, he said.
Many companies are learning to ensure that remaining employees on an ERP project receive proper training, and are attempting to capture knowledge of an ERP implementation through computer-based training, Bermudez said.
One currently popular myth that may lead companies into acquiring ERP software they can't really afford is that as large ERP vendors such as SAP AG create slimed-down versions of their software for the under US$200 million market, many smaller players will go out of business, according to Bermudez.
"Large vendors can't be all things to all people and many smaller vendors will survive in niche markets," Bermudez said. These vendors include Symix Systems Inc., Intentia, and Lilly Software, he said.
On the other hand, it is not true that best-of-breed applications for sales force automation and specialized business analysis will prevent players like SAP and other vendors of integrated packages from getting into these markets, Bermudez said.
"Companies still like to get integrated applications from one vendor -- it saves a lot of development costs trying to tie applications together."
--Marc Ferranti is a correspondent with the IDG News Service
If you have technical problems with this magazine, contact email@example.com