Hadoop Pitched for Business Intelligence
Starting as a Web indexing software, Hadoop is being pitched as a business intelligence tool.
Tue, October 12, 2010
IDG News Service — While it began life as a tool for indexing Web pages, the open source Hadoop framework is being marketed as a tool that could house and analyze vast amounts of data with the kind of proportions that would quickly overwhelm traditional database systems and data warehouses.
Tuesday in New York at the Hadoop World 2010 conference, a number of organizations plan to discuss how the framework could be used within the enterprise. Among the possible uses being discussed: Business intelligence (BI).
"Hadoop is a phenomenal number-crunching engine," said Jake Cornelius, who heads up product management at Pentaho, a BI software provider. He admits it wouldn't be used in all cases of BI, but for really large or complex ones, it could come in handy.
"There really is a small subset of scenarios that we think of as big data problems, where you really have to start looking at Hadoop to solve these big problems," Cornelius said.
Others agree. "If you look at large corporations today, they are dropping data on the floor because they don't have a place to put it," said Eric Baldeschwieler, Yahoo's (YHOO) vice president of Hadoop software development. Running on commodity hardware, a Hadoop cluster could provide a low-cost expansive platform for just such data.
An increasing number of software companies are offering more support for the technology, which could attract more business users. For instance, Yahoo has just released a number of enhancements to make the technology more palatable for enterprise use. On Tuesday, Pentaho released an integration suite for enterprise BI users, called Pentaho for Hadoop.
Yahoo, in fact, is one of Hadoop's biggest users. The company uses the technology in a variety of ways, including as a sort of a very large data warehouse, Baldeschwieler said. Hadoop clusters hold massive log files of what stories and sections users click on. Advertisement activity is also stored on Hadoop clusters, as is a listing of all the content and articles Yahoo publishes.
"It is a hugely varied set of stuff, and the challenge is that when you try to build new products it often makes a lot of sense to ask questions that combine all those different things," Baldeschwieler said.
Recently, Yahoo released a number of enhancements to Hadoop to make it more of an enterprise-ready BI platform. For instance, Yahoo has added security features in its own distribution that would allow Hadoop to span across multiple firewalls.
"Before our engineering, the only way you could put sensitive data onto a Hadoop cluster would be to firewall the cluster and control access," Baldeschwieler said.