Teradata builds on traditional strengths while positioning itself for growth through services and analytics.

Throughout the changes of three decades, Teradata has remained a steadfast presence in the data world, with a large and loyal following. Spun from a research program at the California Institute of Technology, Teradata launched itself from a Brentwood, California, garage in 1976 on $150,000 of seed capital. Three years later, the company incorporated, began to grow and kept growing even as thousands of companies collapsed during the technology bubble. Built on an industrial-strength database and massively parallel processing (MPP) that remains an industry standard, Teradata became synonymous with big iron and big data. Twenty-six years after incorporation, Teradata is now a $1.5 billion company that resembles its old self - and then again doesn't. Currently aimed more at the software sector, Teradata has augmented an ongoing emphasis on enterprise data warehouse (EDW) decision engines with a newer focus on analytics. DM Review Editorial Director Jim Ericson recently spoke with chief technology officer Stephen Brobst and senior vice president and general manager Mike Koehler for an update on the company's strategy and direction. 

DMR: Teradata always seems to have a steady stream of big customer announcements, but is the market for traditional data warehousing becoming saturated? 

Stephen Brobst: I'm not saying that everybody's done all that can be done in pure reporting, but the growth area in data warehousing is around operational intelligence and enabling the real-time enterprise with enterprise-wide business intelligence. We call this "active data warehousing." A passive data warehouse has its focus on reporting, while an active data warehouse builds on that by providing operational intelligence for execution of the business strategy.

Our customers are focused on financial management, not financial reporting. If I'm going to miss my monthly revenue projection, I want to know it on day five, not on day 30. On day five, I can take corrective action and be proactive in influencing outcomes. That means setting up triggers and understanding patterns while I can still do something about it. Although we enable financial reporting, the focus is on enabling management, getting down to detail and figuring out the actions they need to take right now.

Michael Koehler: Our market momentum is strong. Our business grew 12 percent in 2004 and nine percent in 2005. The EDW has become a business requirement today - and growth in the EDW space is being driven by the increasing number of decisions that businesspeople make each day and the expanding amount of data they must manage. Surveys back this up. Integration and the availability of data is critical. Our customers grow their data warehouses because there is a return on investment each step of the way. The best evidence for the business value of building out an enterprise data warehouse is the fact that the majority of our customers start small - a terabyte or less - and they grow. We serve companies of all sizes. Not all our customers are in the Fortune 1000. The growing demand for the EDW is why we continue to win new customers and expand with current customers.

DMR: How does Teradata enable the idea of operationalizing data versus simply amassing data?

MK: The value of the data warehouse really comes when you can integrate data from across the enterprise and expose it to all users. Companies that are moving forward with an enterprise data warehouse are scaling up and building out to actively enable more users. 

SB: One of our battle cries right now has to do with decisioning services and making decisions actionable, meaning we're integrating into the business process. When you map out the business process, you get involved in exceptions. We provide the information to enable decisioning services, but those have to integrate with the action-taking systems which are part of the core workflow and that are handled largely by the OLTP [online transaction processing] systems. Our goal is to provide assistance capabilities that back up decisioning processes - which are real business processes.

DMR: Tell us about your strategy shift from hardware and databases toward software.

SB: In the current business environment, scalability is important, so having Teradata "inside" is important. But scalability is more than just the size of the database; it has to do with the number of concurrent queries you can run and the complexity of the data model. For business it's the ability to ask more complex questions - which you couldn't ask before. We do provide a total solution, but as a software company we've grown our stack around analytic applications - our own, and with our partners'. Ours include analytic CRM [customer relationship management], finance and performance management, profitability analytics, demand chain management and supply chain analytics. We also have a decisioning suite specially tuned for retail. We have implemented applications that use an operational intelligence model with workload prioritization, much more sophisticated than any other database has, with a re-engineered I/O model that allows traditional full table scan workloads to coexist with the tactical decision-support workloads. One of the things people worry about is, if you're doing all operational intelligence, it might mean your traditional stuff is going to suffer, and we can't allow that to happen. So we've re-engineered the underlying I/O model of the database to allow the co-existence of those two workloads in a very sophisticated way.

MK: We invest in solutions for industry and subindustry segments. We build industry-specific logical data models and EDW roadmaps. These help customers understand the data they have, figure out new ways to use that existing data, and then determine data needs for new applications. The whole idea is to get more value out of the EDW, faster.

SB: The difference with Teradata is we transform and store the data only one time, so you can leverage it multiple times. If you have the right technology that doesn't force you to break things up into redundant pieces and have a different database for every single analytic application, you get much lower total cost of ownership. For organizations without EDW scalability, each new analytic application becomes another data mart.

DMR: Let's face it though, a lot of organizations are looking at point analytic applications that address transactional or federated data sources versus the old monolithic view of the data warehouse.

SB: You have to look at architecture versus methodology. The EDW architecture supports centralized, shared information. The methodology should not be the big-bang theory, which is what "monolithic" implies. You want to produce business deliverables, not infrastructure deliverables, incrementally with measured values at each point along the way. The EDW does this. To your question, we have a retail customer that was dealing with a fraud issue and needed access to certain data. Their original plan was to build an ODS [operational data store], but someone on the steering committee who was part of the data warehouse team said, "Wait, we have all the data already in the EDW." They already had the workload management to prioritize it to the proper level. They saved months and millions of dollars and got what our CMO, Bob Fair, likes to call a "freebie." That's pragmatic.

MK: Another good example is a financial institution outside the U.S. that is working through Basel II. After they got six months down the road, they realized 70 percent of the data they required already resided in the enterprise data warehouse - and the project they had scoped out required one-third of the investment and time they originally thought. At the end of the day, multiple groups of people working with siloed infrastructure as it relates to analytics is just more costly and delivers less insight than when you integrate it.

DMR: Do you sell this kind of proposition to IT or business, or both? Is the CIO your usual internal salesperson?

SB: In reality, we have to sell both IT and the business. Often it is the business that is selling to IT. There is some unique capability they need, and in many cases, Teradata is not the "IT standard" in an organization that is coming from an OLTP world.

MK: If an organization is not using a centralized EDW, the Teradata approach represents a change in terms of architecture and direction. Many times the interest comes from the business side, based on an understanding of the possibilities and the value of an integrated, enterprise approach. Other times, if a company has hit the wall technically with a large data warehouse implementation, then we're working with the CIO.

DMR: The big enterprise resource planning vendors have all adjusted and componentized their products in support of the demand for services. How has Teradata had to adjust to the new service paradigm?

SB: We used to think of our stack as the data warehouse stack, but we don't anymore. All you have to do is look at our customers. We have retailers doing vendor-managed inventory, and all their suppliers are accessing their data warehouse. A large number of our customers actually have more nonemployees than employees using the data warehouse! The model is to allow customers to deploy Web services for decisioning services that integrate with business processes using middleware such as TIBCO, NetWeaver and others. We have re-implemented our own analytic applications and componentized and standardized them. This includes Teradata Linux implementations, LDAP implementation and J2EE compliance for all of our analytic applications. To play in that synchronous environment, we need to deliver sub-two-second response times for certain classes of decisions. That's not your grandfather's data warehouse.

MK: We've been investing heavily in R&D. Our core engineering headcount has grown more than 20 percent in the last two years. We've also grown our business and technical consulting headcount by 20 percent over the same time. We understand that the way our customers derive value from detailed enterprise data varies by industry segment, and that's why we go to market by industry segment, such as telecommunications, retail, travel and transportation, and so on. Each segment has unique business improvement opportunities and we build analytical solutions to address those. As Stephen has said, we have the best technology, but in customer engagements we don't lead with technology. We lead with business value and enterprise fit. 


Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access