In-Depth

Teradata Cozies Up to SAP

Partnership highlights Teradata’s unique analytic strengths

• By Stephen Swoyer
• 02/04/2004

When you’re shopping for a data warehouse for use with your SAP ERP applications, Teradata probably isn’t the first name that comes to mind.

That could change, however. Teradata and SAP last week entered a partnership intended to promote Teradata as an SAP data warehouse solution for customers with high data volume requirements. In addition, Teradata will become an SAP Global Technology Partner.

Teradata, a subsidiary of NCR Corp., does not market a relational database for use with operational data, says Mike Schiff, a senior analyst with consultancy Current Analysis Inc. Teradata has limited its uptake as a data warehouse for use with ERP applications. “What’s surprising about it from my perspective is when you’re running SAP, you’re running it on some database, and that database is obviously not Teradata, because Teradata is not meant for operational systems,” he points out. “But you would assume you’d probably run [your data warehouse] on the same database that you had your operational [data] on.”

The irony, of course, is that Teradata is one of the most respected names in high-end-data warehousing, and supports some of the largest data warehouses in existence. SAP, for its part, is the world’s largest purveyor of enterprise applications. For companies that maintain extremely large SAP implementations, Schiff says, Teradata would be an ideal analytics engine. Until now, however, customers that have wanted to make Teradata’s analytic software play nice with SAP’s operational applications have largely been on their own, he observes. “This move will greatly reduce, and in some cases eliminate, the custom development and integration costs currently experienced in joint SAP and Teradata environments,” Schiff suggests.

Wayne Boyle, chief technology officer of applications strategy for Teradata’s Development Division, says that integration between Teradata and SAP has to this point largely been one-off, on a customer-specific basis. “That was all done as a custom implementation. A lot of times that would be a roll your own solution, sometimes a third-party ETL tool, but all very much one-off stuff,” he observes.

The two partners plan to execute on the alliance in a phased approach, the first part of which will ratchet up integration between Teradata and SAP to facilitate seamless access from SAP’s Business Information Warehouse to data residing in Teradata. “There’s a series of standard interfaces as part of the NetWeaver architecture, and we will be certifying select interfaces up and down the NetWeaver stack,” says Scott Gnau, vice-president of product strategy and portfolio management with Teradata’s Development Division.

This is a compelling proposition, Schiff says, because native access to Teradata means customers don’t have to resort to the kludge of summary data. “You can use stuff that’s generated for the Business Information Warehouse and go after data stored on Teradata, but you don’t need to store summary data [in the Business Information Warehouse] first,” he says, noting that the use of summary data can in some cases preclude the possibility of further analysis.

“Once you summarize it, you summarize it, so if you want to go from costs by department level to costs by employee, you can’t do it,” Schiff points out. “Most data warehouses summarize data at some level in order to reduce the amount that must be stored. One of Teradata’s advantages is its ability to handle massive amounts of data without the need to summarize.”

Both companies expect that the integration between SAP and Teradata will primarily be of interest to customers in specific vertical markets, such as, for example, aerospace and defense, pharmaceuticals, and telecommunications. As a result, the alliance’s second phase will attempt to delivery industry- or function-specific analytic applications, officials say.

Company representatives say they plan to push Teradata and SAP integration wherever it makes sense, and not just in specific verticals.

“We’re not limiting ourselves to those specific industries or those specific target markets,” says Gnau . “The objective is to get the certification to work for any high-data volume type of application.”