Data continuity a linchpin for Teradata

Thousands of Teradata’s customers flocked to Seattle last month to attend the company’s annual Partners conference, and while the common goal of many of the attendees was to share ideas, the projects the customers were working on were very different.

Executives at the Windber Research Institute (WRI) – a biomedical research facility in Windber, Pa. – realized when starting the company almost three years ago that it would need to invest in a solution that would help them manage vast amounts of data and information.

Windber needed a data warehouse that could integrate data not only from the research facility itself but also from the Internet, said Nicholas Jacobs, CEO of WRI. One of the biggest challenges facing the institute was that “there was nothing out there on the shelf or developed that would fit our needs.”

Windber’s president of professional services, Holly Rigby, said it had looked at other vendors before choosing a data warehouse, but said that Teradata’s reputation is why the company won WRI’s business.

“Our intuition about them was that they would be someone who would work hand-in-hand with us collaboratively because we knew it would need to be something that was done right,” Rigby said.

WRI has been working with Teradata – the data warehousing division of NCR Corp. – since March and recently finished phase one of its project, which was to integrate the Teradata platform. Rigby said phase two – making the institute’s data mining tools more user-friendly to researchers, allowing them to manipulate data more easily – will begin this month.

With Teradata’s data warehouse in place, WRI will be able to bring different data solutions together – including its genomic and clinical data along with its demographic history data, said David Rosendale, database administrator for WRI. “[We will] be able to track over time aggression of disease cells versus non-disease cells or genes.”

He added that when choosing Teradata there were some “heated discussions” with members of the institute’s IT team “because everybody has their opinion over who would be best.”

While not over the choice of vendor, there were heated debates at Planning Systems Inc. (PSI) about how the Teradata product should be implemented.

Reston, Va.-based PSI – a provider of software systems engineering – partnered with Tarleton State University and Teradata almost three years ago to work on a project for the United States Department of Agriculture (USDA).

The goal of the project was to put a global data mining system in place to look at the annual cost of crop insurance fraud to the government, said Mike Cross, director of Texas operations for PSI. “Crop insurance is delivered through individual insurance companies and they were probably doing some of their own internal data mining, data warehousing, but no one was looking across the entire spectrum and focusing solely on this concept,” Cross said.

He added that last year alone US$34 billion in liabilities were reported and over US$4 billion in compensation was paid out on losses. So far the project has cost US$6 million and on that investment, Cross said they returned about US$180 million in savings through a combination of identifying broad waste and abuse, and giving reimbursement on compensation payments or in the prevention of insurance fraud.

Cross said that although PSI has seen great savings in its project with Teradata, ROI isn’t guaranteed.

“You can’t build a warehouse and all of a sudden save money. You’ve got to exploit that [data] and exploitation can’t come from Teradata, it’s got to come from the people in charge,” Cross added.

Another piece of advice Cross had for companies planning to implement a data warehouse is not to overwhelm themselves with information. Just because the data is available doesn’t mean it has to be collected, he said.

“Data for the sake of data is getting to be a rampant problem. If I want to be able to collect temperature data every tenth of a second, I can do it. [But] do I really need to know the temperature every tenth of a second, or is the temperature every 15 minutes good enough?” Cross said.

Planning is also key when implementing a data warehouse, Cross said, adding that out of the six months it took to get the project off the ground, four of those months were dedicated to planning – which made some members of Cross’ team angry.

“Most people want to get out there and grow data and look at it, and I refuse to let them do that because I know the pitfalls of that. Once you go down that path it’s very difficult to get off it,” Cross said.