In-Depth

Teradata Stays Hungry

Teradata's new DW appliances -- and, especially, its new appliance pricing -- highlight a new limberness (some might call it agility)

Teradata Corp. -- the new spin-off of the former NCR Corp. -- recently kicked off its own foray into the data warehouse (DW) appliance segment with its new Teradata 550 and Teradata 2500 offerings.

Teradata's new DW appliances -- and, especially, its new appliance pricing -- highlight a kind of limberness (some might call it agility) on the part of the venerable data warehouse giant. That's how vice-president of product and services marketing Randy Lea characterizes the product introductions.

The Teradata of old wouldn't so much have countenanced an explicitly-branded DW appliance, Lea says, let alone have pursued so aggressive a partnering strategy with industry players -- and nominal competitors -- such as SAS Institute Inc. and Relational Solutions Inc. (RSI).

"It's part of a new Teradata. We are a little more aggressive," Lea acknowledges. "Part of it is that our [financial] results are our results now. We aren't part of a bigger parent company. That makes you more aggressive."

Teradata's official entry into the appliance segment helps close a loop, Lea continues: it used to be that Teradata concentrated on the data warehousing high-end -- e.g., what it calls Active Data Warehousing (i.e., a DW methodology that focuses on traditional user constituencies -- power users or analysts -- as well as frontline consumers of operational intelligence) -- at the expense of low-end customers who were trying to address lingering capacity and performance issues, particularly with respect to ad hoc query and analysis.

Teradata excels with ad hoc queries, Lea stresses, but -- prior to the introduction of its new appliance systems -- couldn't match the price points targeted by Netezza Inc., DATAllegro Corp., Kognitio, Dataupia Inc., and others.

With a sub-$100,000 appliance ($67,000 for the 550, according to Teradata marketing collateral), it can now do just that, he says. "This is where the appliance vendors quite frankly were leveraging our exit from the market, but we now have products -- the 550 and the 2500 -- that are designed specifically for this environment," Lea indicates. The 2500, in particular, gives Teradata a competitive leg-up vis-à-vis its appliance competitors, he maintains: "This functionality of the 2500 is for our traditional data warehouse; this is for our Active Data Warehouse, so there's clear differentiation in performance capabilities versus what our competitors offer [at similar price points]."

The new Teradata seems less locked into monolithic -- or enterprise-data-warehouse-or-bust! -- thinking. Lea, for example, acknowledges the unthinkable: that most customers start with data marts (independent or dependent) and grow their data warehouse infrastructures from there.

It used to be that Teradata tried to pigeonhole Netezza and other appliance players as data mart-only plays. So to have Teradata acknowledging the foundational importance of data marts seems like a Very Big Deal.

This isn't quite the reversal-of-course that it might seem, at first glance, anyway: to be sure, customers start with data marts -- and to the degree that appliance solutions can help them quickly roll-out data mart solutions, they're perfectly acceptable. As customers grow, Lea argues, they find that DW appliances are less suitable for enterprise data warehouse (EDW) -- or even for departmental- or business-unit-level DW-type -- implementations.

With its new appliance offerings, he argues, Teradata can offer customers affordable data-mart-only solutions on the bottom end (with the new 550 server), an entry-level foray into Active Data Warehousing (with the 2500), and -- of course -- the full-blown EDW experience, too.

"You always start with a data mart and then you grow it. Where we were, we'd gotten to a single price point where it was hard to bring that price point down" to address the data mart-specific needs of customers," he explains.

If Teradata's insistence on enterprise-data-warehousing-at-all-costs seems a bit self-serving, so, too, does at least one popular riposte from the appliance crowd: namely, that the importance of the EDW itself has been overblown. EDWs are moving targets, appliance players like to argue, so much so that EDW adopters won't ever fully populate their warehouses with all of their enterprise data.

Lea dismisses this as a strawman argument even as he effectively agrees with it: No, Virginia, an enterprise won't ever populate its EDW with all of its information -- but does that mean it shouldn't even try? More to the point, Lea argues, companies can reap windfalls of analytic insight by making information from different business units available to consumers (in other business units) across the enterprise. Even if only a fraction -- say, two-fifths -- of that data is available, isn't that better than a largely siloed scheme?

"Almost every company out there is trying to build an enterprise data warehouse, but due to organizational constraints or what-have-you, no one has [yet put] 100 percent of their data [in the warehouse]," Lea acknowledges. "The biggest we've seen is one of our customers that claimed they had 92 percent of their data in the warehouse, but we don't expect everyone to [achieve] that. What we're talking about is the enterprise view. Think about it. For every subject area you decide to commit to the warehouse, you have everybody in the enterprise looking at that data. You may only have 40 percent of your operational data in the warehouse, but now the whole enterprise is looking at it."

If anything, he maintains, Teradata's new appliance offerings should make it easier for organizations to wade into the waters of EDW-dom: they can start small, he says, and build out over time.

"We're the data warehouse architects. We always have been. When you have one product, they look at you as being a little inflexible. Not being able to meet their needs. But these [new system offerings] give us that platform opportunity. You can start small. You don't have to update the application [when you go bigger]. You don't have to change the SQL. You don't have to change anything. It's all a Teradata 12 database [under the covers], so all of your ETL, your application development -- all of your training is portable all the way across."

About the Author

Stephen Swoyer is a Nashville, TN-based freelance journalist who writes about technology.

Must Read Articles