Union Pacific Corp. taps into its 4-terabyte data warehouse for serious strategic planning, everything from analyzing the prices it charges its customers to scrutinizing the traffic patterns of its freight cars. But for day-to-day tactical decision-making such as scheduling train crews, Union Pacific operates a number of separate databases--even though Paul Evans, the railroad's enterprise data warehousing senior manager, acknowledges that managing multiple data systems is a challenge.
Teradata, the NCR Corp. division that supplies Union Pacific's data warehouse technology, this week will debut a major upgrade of the Teradata suite that's designed to help erase the gulf between IT used for strategic and tactical business intelligence. Teradata 7.0, unveiled at the vendor's user group meeting in Las Vegas, is the first real embodiment of "active data warehousing," the vendor's term for data warehouses that can process data in real time and provide analysis capabilities to front-line employees such as ticket agents and call-center workers.
"In the end, your strategy is only as good as the execution of your strategy," says Stephen Brobst, Teradata's chief technology officer.
Building a data warehouse for strategic and tactical analysis has been practically impossible until now, industry analysts say, because data is organized differently for each. Strategic queries analyze huge amounts of historical information across dimensions such as time and geography. Tactical queries, in contrast, are narrower in scope and involve less data, but they require faster answers using data that's updated frequently or even in real time.
"Data warehousing started out as a strategic view of data. But today, more people need to know what's going on with their business right now," says Kyle Prescott, database administrator at UnumProvident Corp., an insurance company. "I see a huge need for active data warehousing for the 'now' decisions, the more tactical decisions."
Teradata 7.0 offers enhancements for building data warehouses for both kinds of analysis. A new partitioned primary index feature organizes database tables for strategic and tactical queries. A new query priority-scheduling tool balances a data warehouse's workload, giving priority to queries from, say, a ticket-counter agent serving a customer over a marketing analyst using the warehouse to research ticket sales.
Last week, Teradata struck a deal to resell Informatica Corp.'s data-integration software, including its PowerCenter RT real-time product, for quickly moving high volumes of data from operational systems into data warehouses. That, combined with other Teradata 7.0 enhancements, will help data warehouse managers update information in real time or near-real time.
"We're definitely looking at doing more consolidation and turning the data warehouse into a platform for broad decision-making," Evans says. Union Pacific has already begun increasing the capacity of its data warehouse to 7 terabytes in anticipation of merging it with the railroad's tactical databases using Teradata 7.0.
The warehouse, now updated daily, will be refreshed as frequently as once per minute. Evans expects the number of employees who use the warehouse to grow from around 4,000--primarily managers--to as many as 10,000 workers within five years.
Wells Fargo & Co., which has been running a pilot data warehouse project based on Teradata for about a year, may leverage the system's new active data warehousing features later next year for online personalization and event-triggered communications applications, says John Ahrendt, senior VP of Internet data services.
However, Giga Information Group analyst Philip Russom says the need for real-time data warehousing is limited to certain applications for the moment. E-commerce applications such as personalization that require an immediate response are among them, he says. Real-time alerts about production yield problems in manufacturing plans may be another.
Many of the 150 enhancements in Teradata 7.0 are designed to boost processing power and automate laborious data warehouse management tasks. Queries can be processed at the same time the warehouse is being updated without performance degradation, a vast improvement from running complex queries or voluminous data updates overnight. The system can also handle more complex queries with SQL statements up to 1 Mbyte in size--a 16-fold increase--with as many as 2,046 variables, compared with 254 in older versions. Teradata 7.0 also offers new data-compression functions, as well as improved security and privacy-protection capabilities.
"Teradata is a couple of years ahead of everybody else," says Richard Winter, president of consulting firm Winter Corp. He estimates that only 5% to 10% of all data warehouses need the scalability and performance enhancements Teradata 7.0 offers.