Business intelligence vendor Teradata is unfazed by Oracle's and IBM's recent saber rattling, saying the vendors' respective technologies remain fundamentally unsuited for high-performance analytics.
Oracle announced version 2 of its Exadata data warehousing appliance last month, with CEO Larry Ellison reportedly saying that Oracle is installing the faster, cheaper Sun Microsystems-based appliance "within the Teradata installed base."
[ Teradata this week targeted the cloud with data warehousing announcements. | Get the no-nonsense explanations and advice you need to take real advantage of cloud computing in the InfoWorld editors' 21-page Cloud Computing Deep Dive PDF special report, featuring an exclusive excerpt from David Linthicum's new book on cloud architecture. ]
Darryl McDonald, chief marketing officer for Teradata, which is hosting its annual Partners conference in Washington this week, claimed that Oracle's alleged customer wins have been mostly confined to "uncontested" situations, such as customers running the Oracle database on Hewlett-Packard servers that are being switched or upgraded to the Exadata appliance.
"There hasn't been a whole lot of head-to-head. In the few times we have competed, we've had a very high win rate," he said.
The reason, McDonald said, is that the Oracle database was built for transaction processing, not analytics. Real application clusters (RACs), intended to enable Oracle database to scale out, remain a Band-aid solution, he said.
"Oracle tries to throw a lot of hardware and CPUs at the problem, but it's fundamentally still a transactional database," McDonald said, meaning it has "I/O problems and resource contention."
IBM earlier this month announced a data warehousing bundle involving its DB2 database software running on top of its Power servers in conjunction with the new pureScale clustering feature for faster, larger grids.
Whatever IBM's marketing may say, DB2 still has "very similar" problems as Oracle, McDonald said, due to its heritage in online transaction processing (OLTP). These are problems not faced by Teradata's massively parallel platform (MPP), he said, and evidenced by customers operating data warehouses with more than a petabyte of data.
McDonald said there are several new members of Teradata's elite club of "Petabyte Power Players," which he declined to name, as well as double the 35 customers with data warehouses in the 100 TB-plus range it had last year.
While the co-opetition between IBM, Oracle, and Teradata may be heating up, Teradata isn't worried that it will result in a falling out.
"Many Fortune 500 firms are dependent upon us for mission-critical apps. I don't foresee how Oracle or IBM would not want to continue to certify and port their products onto Teradata," McDonald said.
Teradata also faces competition from a fleet of startups and smaller companies offering specialized data warehousing appliances, often sporting extremely low prices.
Michael Koehler, CEO of Teradata since its spinoff in fall 2007 and the senior vice president for the division since 2003 under NCR, said Teradata can match those firms on price-per-terabyte with appliances such as its Extreme Data Appliance 1550, which today sells for about $15,000 per TB.
At the same time, he decried the overemphasis on this metric, which he said fails to capture real-world performance problems, such as the number of concurrent users or queries companies can run.
Teradata opened its annual conference Monday by announcing that it is bringing its data warehousing technology to private corporate clouds as well as public ones, such as Amazon.com Inc.'s EC2.
Despite the move, Teradata remains ambivalent about the cloud. Like another on-premises player, Microsoft, which has crafted a " software plus services" strategy, Teradata argues that its move is simply to broaden its offerings, rather than to push customers in a direction they don't want to go.
"We think it's going to be a number of years before the Fortune 1000 leverage public clouds in a big way," McDonald said.
This story, "Teradata: We still rule over Oracle, IBM in analytics" was originally published by Computerworld.