Once considered staid, insurers now are using cutting-edge technologies to measure risk and reward, gathering more data and enabling more users to access that data. For insurance giant The Travelers Companies, Inc. (Travelers), a new attitude toward personal insurance dictated a change of technology that allows the company to integrate data and schedule workloads.

Starting in the 19th century and continuing into the 21st century, Travelers has developed into a powerhouse insurance company, providing a variety of products across three major business units: personal insurance, business insurance and financial, professional and international insurance. As its vision changed to meet demands, the company turned to technology to stay ahead of its competitors.

For many years, the personal insurance unit targeted low-risk consumers, explains Sullivan B. McConnell, vice president of business intelligence (BI) at Travelers. Now, thanks to the ability to gather and analyze data, the company can expand its customer base with greater confidence. The strategy for the business unit "centers around better leveraging information inside and outside our enterprise," says McConnell. "It has to do with our ability to understand risk and to develop the most appropriate pricing in our industry. Given large amounts of data, it's up to us to figure out what that appropriate price is."

Sullivan B. McConnell, vice president of business intelligence (BI) at The Travelers Companies, Inc., says Teradata helped the company understand risk, optimize pricing and expand its customer base. McConnell joined the original Travelers right out of Vanderbilt University and participated in the company's Information Technology Leadership Development Program. In 1995, McConnell left Travelers to pursue his career interests in data warehousing and BI. He rejoined Travelers in 2003. Today, McConnell focuses on enterprise BI for the company.

With the change of view came the realization that Travelers needed to assess how its information technology served the new vision. Prior to 2004, the IT platforms were very diverse. "Information was siloed in a number of different systems in a number of different technology platforms," McConnell says. "We had data warehouses and data marts in DB2 on the mainframe, in DB2 in the distributed environment and SQL Server in a variety of environments." He further explains, "Different systems which store subsets of important decision-making information weren't integrated."

Because the technology wasn't letting the division achieve its business strategy of understanding risk and optimized pricing, the unit shopped for a new system. "We thought that Teradata was the best technology platform to support that strategy," McConnell says. "We had key criteria that we looked at when we evaluated the platforms. We wanted a platform with the ability to scale because we knew that our internal growth, coupled with the possibilities of mergers and acquisitions, was a key consideration." The system also had to accommodate additional users and increased data volume. "We wanted a system that was going to handle mixed workload well," McConnell explains. "Being able to handle and prioritize mixed workloads is important to us and means the system has to be highly available. What happens if components fail or pieces of the system aren't available? Does the system stay up and support the business processes?"

After doing the research, the unit moved to Teradata in December 2004. Interestingly, the business insurance and claims units did their own independent analysis and arrived at the same technology decision for each of those units: Teradata.

Lessening the burden
Teradata gives Travelers flexibility in handling mixed workloads and scheduling system resources. Unlike some consumer businesses, Travelers doesn't need queries to be answered in mere seconds. Rather, the company must differentiate "workload into different classes that have to run with different performance characteristics," notes McConnell. "Teradata Active System Management is a holistic approach to managing resources. When queries enter the system, you've got a chance to evaluate cost, what workgroup they're in and who's submitting the query." Because this occurs before the queries get accepted, McConnell can make decisions about how many types of queries can be run concurrently, thereby keeping the system from experiencing a painful slowdown.

The system's flexibility doesn't end there. With Teradata Dynamic Workload Manager, once the queries are in the system, the company can decide how to best allocate resources. "I've got the opportunity to dedicate 30 percent of machine resources to this application and 40 percent of machine resources to that application," he says. Decisions can even be made at the workgroup level with different priorities assigned to different tasks.

Meanwhile, Teradata Priority Scheduler helps the unit create essential data reports and meet deadlines. Each month, a set of derived data is processed for executive level reports. "People expect it on a particular day of the month, every month," McConnell says. "Because the system satisfies a broad range of other applications and users, there's intense pressure on the system. Teradata Priority Scheduler enables us to dedicate machine resources to a particular class of users or applications or queries. For a particular set of days at month's end, we dedicate a set of machine resources to creating this derived data set. By day of month and time of day, we're able to have the appropriate machinery resources to make a set of performance service level agreements [SLAs] for a class of workload. It's pretty granular and we've used it to great effect." How did Travelers do this before Teradata? "It was actually pretty difficult," McConnell says.

Meeting changing needs
Since Teradata was installed, "we've become more sophisticated, implementing components of the data warehouse as the user community has grown. Teradata is able to handle mixed workloads and assist in meeting the performance SLAs of our business community," notes McConnell.

But this didn't happen the first day of installation. "Honestly, for the first year and a half of our implementation, we didn't pay attention to any of the workload management capabilities because the performance of the system is very, very good," explains McConnell. "In business intelligence, when you start to get good information and good performance, more people want to use it. When more people want to use the system, you start running into situations where you are trying to meet a performance SLA. It's only in the last six months that we've really had to implement some of the advanced features of Teradata's Priority Scheduler." For instance, McConnell says, "when we have a particular workload that revolves around a certain SLA, we need to get our information on the desks of the decision makers by Monday after month close. Or, we might have a class of workload that needs a particular response time. Teradata's workload management capabilities are really strong. In my opinion, they are best of class."

Teradata Priority Scheduler has made the difference between missing deadlines and meeting them. "Before implementing Teradata Priority Scheduler, we did miss timeliness SLAs on a particular workday on occasion," McConnell says. "Since implementing it, we haven't missed an SLA with respect to data timeliness. The overall performance of queries better meets customer expectations."

Making internal customers happy is just one aspect of Teradata at Travelers. The company's strategy and Teradata's technology "puts us in a tier of companies that really gets the point of understanding risk and optimizing price," McConnell says. "It puts us in a small class of competitors that does this very well."

Confidence for the future
As McConnell looks to the future, where does he see the company's BI headed? "Because there is so much information that our business partners want to leverage, we'll be adding subject areas to this system for a long time. It underscores our decision to find a system that can scale as our needs expand. We'll add data from other operational systems that hasn't been brought into the enterprise data warehouse, and that will provide us with more insight into marketplace opportunities and understanding risk better."

Before the Teradata system was implemented, McConnell was kept up at night wondering how the company could design the system for scalability, work through all the data integration issues, and manage mixed workloads as needs evolved. Another concern was whether the company was ready to make a change and accept Teradata as a platform. "Making a change was a hard thing for us to do," McConnell says. Despite this challenge, implementation of the Teradata system not only alleviated McConnell's worries, but also instilled confidence that the technology would support Travelers' business strategy. T

Behind the solution: Travelers Database: Teradata Database V2R6.1 Server: Three production systems: 6-node 5400 Server, 6-node 5400 Server, 4-node 5400 Server, 39 nodes across dev, test, model office and production environments across all lines of business Users: 2,500 DBAs: 2 full-time equivalents (5 people) Data Model: Third Normal Form with some modest denormalizations; a few virtual data marts, and a few physical data marts in star schemas Operating System: UNIX MP-RAS Storage: 6TB in production Teradata Utilities: Teradata Load Utilities 8.1, Teradata Active System Management, Teradata Priority Scheduler, Teradata Metadata Services, Teradata Warehouse Builder, Teradata Analyst Pak and Teradata Access Modules Tools/Applications: Teradata Warehouse Miner, Cognos Impromptu, Cognos 8, Microsoft Analysis Services 2005 and products from Ab Initio, Business Objects, Hyperion, IBM and Information Builders

Shirley S. Savage has published articles on technology, energy and science.

Photography by Kenneth Scott

Teradata Magazine-June 2007