9‎ > ‎

x

Big Data

The Big Data Challenge to Legacy Data Management Companies

By Steve Lohr
April 7, 2014 4:00 amApril 7, 2014 4:00 am PhotoMainframe computer sales — once the centerpiece of I.B.M. — now account for an estimated 3 percent of the company's revenue.Mainframe computer sales — once the centerpiece of I.B.M. — now account for an estimated 3 percent of the company's revenue.Credit I.B.M.

Scott Gnau, president of Teradata Labs, seemed a little irritated by all the attention being showered on the big data marketplace. Big-data exuberance has surged with the recent news about how much money was raised by Cloudera, the frontrunner among start-ups distributing Hadoop, open-source software used for storing and parsing huge volumes of data. The total — mainly from Intel, but also venture capital firms — was $900 million, putting a value of $4.1 billion on the young company.

Mr. Gnau called the size of the investment “astounding,” a view shared by many. He leads the research and development arm of Teradata, a leader in data warehouse software, founded in 1979.

In an interview last week, before a Teradata conference this week in Prague, where the company is making product announcements, Mr. Gnau took issue with what he characterized as the prevailing “misinformation” that big data-style software could substitute for Teradata’s technology. “It’s not one size fits all,” he said.

Mr. Gnau is correct, of course. But the challenge to old-line data software companies like Teradata is not that their programs will ripped out and replaced, but that its business erodes and it loses pricing power, as customer spending and technical talent migrates to a new generation of technology.

Look at the story of the mainframe computer. The rise of the microprocessor and lower-cost computing didn’t spell the death of the mainframe. In fact, more mainframe processing cycles are used today than ever before. But less expensive microprocessor-based computing began nibbling at the edges of some of the computing tasks mainframes had done, and the low-cost paradigm undercut I.B.M.’s pricing power. The mainframe technology lived on. But its high-margin economic model eventually collapsed, nearly taking I.B.M. down as a company in the mid-1990s.

Is a similar fate awaiting Teradata, and parts of the businesses of Oracle, I.B.M. and others? Not immediately, but as more software is built to run on top of Hadoop, and this next-generation of data software becomes faster, more reliable, and capable of handling more data-handling tasks?

I called Richard Winter, an independent technology consultant, and asked him that question. First, Mr. Winter said, the new technology is a real challenge. “Hadoop,” he said, “isn’t threatening to replace something like Teradata’s data warehouse software, but it is part of the competition for new data workloads.”

Yet Teradata and other legacy data software suppliers, Mr. Winter said, are moving to embrace the new technology rather than merely resist it. Teradata, for example, has a partnership with Hortonworks, another start-up distributing Hadoop. On Monday, Teradata introduced Query Grid, essentially a software dashboard for tapping into data whether it resides in a data warehouse or in Hadoop.

This is a familiar tactic from the incumbents’ playbook. Your customer base is your strongest asset, and your message and technical strategy is to bring your customers into the new world without destroying the past — their investment and yours. Faced with the Internet threat, Microsoft, the dominant software company of the personal computer era, famously adopted a campaign known as “embrace and extend.”

But Mr. Winter makes a further point. Teradata is the champ of high-end data warehousing. The data warehouse, Mr. Winter said, is where your most valuable data is held. It houses customer records, information about products, operations and transactions. It makes sense, he explained, for that data to be pampered and curated, because it will be used thousands and millions of times over many years.

Hadoop, Mr. Winter said, is ideal for rapidly and inexpensively storing vast amounts of data of different kinds — web browsing and sensor data, for example — that might be of interest. Clever algorithms then mine that unruly data for patterns that could yield insights for saving money and finding customers. But that data, Mr. Winter said, doesn’t need the careful handling of the corporate information in the data warehouse.

“Hadoop,” Mr. Winter said, “will be this huge thing that grows alongside. But it’s a very different kind of model and capability than that of a data warehouse.”

#auto

Subpages (4): 2 4 n u
Comments