Columns

Paradyne Corp.'s Daniel Bond

• By Jack Vaughan
• July 12, 2001

"We're not 'building the city of Boston,' we're building a few streets." This is Daniel Bond talking. He is speaking about data warehousing and he speaks from many years of information technology (IT) experience. Like others the data warehouse manager at Paradyne Corp. exists in a deluge of buzzwords -- everything from data marts to Web-enabled Olap. At the end of the day, he must deliver useful systems to users that are largely immune to buzzword mania.

When he talks about building 'paths' and not whole cities, he addresses a key issue in data warehousing today. The question is: Does the data warehouse with a modest goal -- often called the 'data mart' -- better support an organization's efforts to exploit corporate data? Conversely, is a larger corporate information infrastructure required to first ensure that dispersed data marts do not become troublesome islands of automation?

In fact, Bond somewhat walks a line on this issue. He is wary of overblown projects, but sees the value of infrastructure. "We're building just enough [infrastructure] to get from the wharf to the warehouse, and," he said, carrying the city-building analogy a little further, "from the warehouse to the bus. Like everyone else, I know about massive projects that go nowhere."

Bond spoke to us over 'eggs over medium' at Boston's Sheraton hotel during last summer's Data Warehouse Institute conference. In a recent project at Paradyne, the Largo, Fla.-based manufacturer of data communications equipment, Bond implemented a data warehousing infrastructure using a Web server and browsers as the ultimate delivery means. The goal was simply to connect Web-based customers with mainframe-based data -- especially to retrieve current order information "hot from the manufacturing scheduler" on the company's IBM mainframe. Today, the system allows customers (who are primarily value-added resellers) to check order status of equipment on line. Links to in-house reporting systems are expected.

Back-end data sources included CA-IDMS VSAM flat files, as well as Sybase, AS/400 and DB2/400 databases. Tools and servers were largely obtained from Information Builders Inc., New York City. These included MVS EDA Hub, Unix, MVS servers, Focus Six and Web Focus products.

Paradyne offers a range of products, including frame relay and new RADSL (rate adaptive digital subscriber line) equipment, that require increasingly complex and expensive technical support and inventory control. The company, founded in 1969 and claiming the title of first producer of both the 14.4-Kbps and the 19.2-Kbps modem, is in a "start-up" mode having been spun-off by former owner, AT&T. Evolution was not without pain. The company's changing hands resulted in an information systems (I/S) numbers' reduction from 25 to three. Consultants provide legacy support.

Among rules of thumb employed by Bond during the design phase: The level of automation should fit the level of support that not only is required, but is also budgeted, deliverable and maintainable. Paradyne deliberately developed a partnership with a selected vendor ( here, Information Builders) in order to avoid 'analysis paralysis,' obtain valuable consulting and to (possibly) influence future products.

In fact, the Web reporting application was required by contract with a major customer. This likely accelerated the usual imperative to move quickly. The pilot was accomplished in three-and-a-half months. There was a quantifiable internal pay-off, Bond said. Sales administrators have reported a 40% reduction in order status calls, thus, in Bond's words, freeing up staff for other tasks.

The final output of Bond's system is Html. Did he consider the hot language of the day -- Java? "No," he said. "I think the first responsibility of I/S is to provide a minimum of architecture to solve a business problem. I wanted to avoid what is essentially hand-coding versus using a 'Case-like' tool. We are dealing with 'green-screen' people who are very intolerant of delay."

He said: "The purpose [of the system] is not just to reduce costs. It is also to draw our customers closer to us."

In discussing the issue of a data warehouse's scope, Daniel Bond noted that chief data warehousing strategists such as Bill Inmon, founder of Pine Cone Systems Inc., Englewood, Colo., have stepped back from original proposals to design large, fully encompassing infrastructures before going ahead with a data warehouse. "Just build enough infrastructure," advised Bond, "but take the time to do it."

An architect thinks ahead, he noted. But you do not have to make the analysis and design stage a long drawn-out process. "A day of thought could be sufficient," said Bond.

All in all, Bond told participants at a Data Warehouse Institute conference session, Web technology can fit in with strategic I/S goals. It is strategic to put more control of subject matter in the hands of the people who know their data, while allowing I/S to retain control of infrastructure elements, including the data warehouse, he said.

-- Jack Vaughan