HP takes aim at Teradata with Neoview mousetrap

Last week I chatted with Ben Barnes, vice president and general manager of the HP's Business Intelligence Group. The former general manager of IBM's Global Business Intelligence Solutions Division was hand-picked by HP CEO Mark Hurd to take an old friend--Teradata, the leader in enterprise data warehousing.

Prior to taking over HP Hurd spent 25 years at NCR, including three years as president and COO of the company's Teradata division (Teradata was spun off as a separate public company in October 2007). Barnes spent time in the 1990s as general manager for marketing at Teradata.

HP's answer to Teradata, which generated $439 million in revenue in Q3 2006, is Neoview.

HP's solution was to convert its NonStop SQL database (which came via Tandem, part of HP's Compaq acquisition) into a query-based, decision support database for business intelligence and enterprise data warehousing that can digest and analyze all of the data within large corporations. According to Barnes, 350 HP developers are working on database and management tools, and have added more than 1.5 million lines of code.

HP is running the NonStop database on its big iron, such as Itanium-based HP Integrity rx2620 servers with 128 or 256 processors. List price for a 16-processor system with 4-terabytes of storage is about $600,000. For the data analysis piece, HP integrates with the usual suspects--Cognos (IBM), Business Objects (SAP), Hyperion (Oracle), SAS and others.

Barnes cited "top down interest " from Mark Hurd in taking on Teradata and others in the enterprise warehousing space. "Its a long sales cycles, but it will ramp up significantly," Barnes said. "Mark and I are convinced its a big business."

Teradata, IBM, Oracle and Microsoft own the bulk of the enterprise datawarehouse software market, which IDC sizes at $5.2 billion today and expects to grow at 9.5 percent annually through 2010.

HP plans to double its salesforce in this category, Barnes said, and leverage its financing organization, which recycles existing equipment at customers sites into secondary markets. In addition, HP acquired Knightsbridge, a high-end enterprise datawarehouse consultancy. HP has staffed up Knightsbridge from 700 to 2800 consultants in the last year.

So far HP doesn't have much evidence to show that it will eat Teradata's lunch. It's market share in the enterprise data warehousing space is close to zero. HP hasn't yet appeared on Gartner's Magic Quadrant for Data Warehouse Database Management Systems (below).

Barnes contends that HP will gain market share because it has developed a better mousetrap than competitors.

"If all we do is what Teradata has done over the last 20 years, it's not enough. The world is changing, and we made a big investment in software with Neoview, leveraging $1 billion in storage and server R&D and big investments in ISVs, investments in services businesses," Barnes continued. "Since we have better a mousetrap, there is no need to buy Teradata."

Randy Lea, vice president of product and service marketing at Teradata, isn't convinced that HP has a new or even better mousetrap.

"We haven't seen anything that shows they have a better architecture or functionality than we do," Lea said. "We spend every waking moment at Teradata is focused on datawarehousing, period. Oracle and IBM have thrown the R&D argument around for years, but we have led the Gartner Magic Quadrant seven straight times. We would be glad to benchmark against Neoview, but they have declined to do so."

Luke Lonergan, CTO and Co-founder of Greenplum, another player in enterprise datawarehousing, said acknowledged HP's Neoview has some strengths. "It has some mature code and good people, and a lot of investment and mindshare from HP, but it's basically a proprietary system, resurrecting the Tandem MX project [NonStop SQL] and using proprietary interconnects," Lonergan said. "Many of our architects came from Tandem. I don’t begrudge HP competing with Teradata--I wouldn’t be surprised if HP bought Teradata, and is using Neoview as a way to drive Teradata's price down. Neoview is coming out a bit hot in the sense it is not totally mature."

HP doesn't anything to say about acquiring Teradata, but it is touting Wal-Mart as an early customer. Barnes said the giant retailer is moving some workloads from its Teradata systems to Neoview.

Teradata's Lea told me, "Wal-Mart is still a strong customer and we are providing the vast majority of mission critical workload. At this time, we don't know of any Teradata system that has been replaced by an HP Neoview system."

I would guess that HP is beginning to make some inroads at Wal-Mart versus Teradata. The company has a good connection into Wal-Mart--it's CIO Randy Mott was formerly the CIO of Wal-Mart and spent 22 years at the company before job hopping to Dell prior to HP. "Mott is the number one CIO when it comes to enterprise datawarehousing," Barnes said. He also said that Wal-mart has a 128-processor Neoview machine in production and another one on order.

But HP's biggest Neoview customer to date is Mott. HP is getting rid of 750 datamarts, consolidating its enterprise datawarehouse to a single build on Neoview that will be accessed by 50,000 employees, Barnes said. Other early major customers cited by Barnes include Bon-Ton, 3M and Canon.

HP is working on a version 2.4 of Neoview, due this year, which will offer suggestions to users on how to better organize data and make it easier to tune the query optimizer. In addition, Barnes said that his team is working to shorten the time required to build models and deliver more real-time analysis. "We have a strong potential for leadership in this market, even if we are late," Barnes concluded. "Think of us a startup up with benevolent funder."

Indeed, with its large footprint and deep pockets, HP will make an appearance in the Magic Quadrant, but as Barnes admitted, it's a long sales cycle. And maybe Lonergran is right. Acquiring Teradata would be a way for HP to shorten its sales cycle, but it seems that Mark Hurd is focused on Neo, creating his own construct for sprinkling business data with intelligence.