LAKE BUENA VISTA, Fla. -- CIO George Swanson has a $50,000 data warehousing problem. It's a problem he inherited, as so many IT professionals do, from a former regime.
Long before Swanson arrived to work at Buffalo Grove, Ill.-based Akorn Inc., the company had acquired tens of thousands of dollars worth of software from business intelligence vendor Cognos Inc. Akorn never implemented it, Swanson said, because, frankly, "it's such a beast."
In the last year, Akorn managed to build a data warehouse on a $100,000 dollar budget. "It's working,'' said Swanson, "except for the huge volume of records." Akorn is currently using Microsoft SQL Server and Excel pivot tables. Now the manufacturer and distributor of specialty pharmaceuticals needs a front-end solution.
Swanson must decide whether to drag the Cognos software off the shelf or spend valuable dollars on other offerings, such as tools from ProClarity Corp.
Swanson and thousands of IT professionals brought their data warehousing problems, large and small, inherited and self-made, to the annual Gartner Symposium/ITxpo 2002 earlier this month.
So enthusiastic were IT executives to learn how to accurately judge the total cost of ownership and the return on investment associated with data warehousing projects that even the late-afternoon panels were well attended, joked one analyst. There is nothing funny, though, about the news that Stamford, Conn.-based Gartner Inc. expects 65% or more of
Gartner analysts told attendees that many IT executives mistake data centers, or databases, for data warehouses. Worse, they attempt to use existing database software to drive a data warehouse, or they create expensive data marts instead of useful data warehouses. Many vendors, meanwhile, underestimate how different, and difficult, maintaining a data warehouse is when compared with online transactional processing.
During one of several sessions on these topics, Gartner analyst Kevin Strange urged IT pros to think of a data warehouse as flexible, a system that is much more than tables of data that can be queried again and again. Instead, a good data warehouse allows for building on concurrent queries. Most relational databases are just not suitable for queries such as complex analysis of sales periods -- for example, how one product sold when compared with another, compared with its own sales history, and whether it outpaced inventory.
Strange had the following suggestions for potential buyers: insist on line-item pricing instead of bundled packages, put the vendors in competitive situations by researching other offers, and don't let your vendor tell you what you need.
"Vendor-suggested configurations are often wrong," Strange said. "Sometimes they like to under-configure, just to get the business."
AT&T data process owner Sandra Layne, who is based in New Jersey, left a session led by Gartner's Ted Friedman looking as if she had just lost her best data. She said Friedman might have been talking just to her when he warned about data warehousing projects that fail because IT and business divisions do not collaborate.
"We need to keep fighting, so people working on the data talk to the people who actually know the data," said Layne, whose department is smack dab in the middle of building a data warehouse.
"We're currently composing business rules, business definitions, a whole library of terms," Layne said. "In my business, you say the word 'pick' and that means two different things to two different people."
Asked what she learned from Friedman, Layne said, without smiling: "We still have the capability to fail miserably."
FOR MORE INFORMATION
Check out a Featured Topic on data warehousing.
To provide your feedback on this article, contact Ellen O'Brien.