“We’re going to see a total revamping of the architecture that supports the decisions that users are trying to make,” said Schmarzo, Global Enterprise Information Management Competency Lead at EMC.
Speaking to Wikibon’s Dave Vellante and SiliconANGLE’s John Furrier from the floor of EMC World 2011, Schmarzo said the volume and type of data users are looking to analyze today cannot be processed with yesterday’s data warehouses.
Traditional data warehouses, Schmarzo said, aggregate data in a central location. Not only is aggregation not feasible in a Big Data environment, “the minute you aggregate data, you wipe out the nuances that are interesting from an analytics perspective,” Schmarzo said.
“It’s like going wine tasting and slabbing Vaseline on your tongue,” he added.
The new breed of data warehouse takes a different approach to analytics. Greenplum’s massively parallel data warehouse, for example, has linear scalability, meaning data does not have to be aggregated first. This gives Greenplum users new capabilities that they may never have thought f before
“The challenge we have with our customers is not getting them to think better, but to think different,” Schmarzo said.
EMC, which acquired Greenplum in December, is helping its customers think of new and different ways of analyzing data. For example, retail shops could track and analyze customer behavior as customers roam brick-and-mortar stores via an RFID chip embedded in loyalty cards.
“We’re working with our customers to help them envision what this can do for them,” Schmarzo said.
During his time on theCube, Schmarzo also gave his views on the future of Hadoop, and the trials and travails of Cisco and Yahoo.
As for Hadoop, Schmarzo said digital media companies have largely perfected their use of Hadoop to analyze log files and other data, but the big moment for the open source framework will come when traditional business intelligence and data warehouse practitioners recognize what Hadoop can do for them.
He expects that to begin to happen in next 9 months as Hadoop spreads to the Wells Fargos and Proctor & Gambles of the world.
Further, he doesn’t view the current commercial Hadoop vendor landscape as being on a war footing. EMC announced its entry to the market yesterday with its Greenplum HD appliance. Cloudera and IBM also have commercial Hadoop offerings, and Yahoo is considering spinning of its Hadoop engineering division.
“There is so much opportunity out there,” Schmarzo said. “It’s a land grab.”
When it comes to Cisco, “I think Cisco made a classic tech mistake in thinking that they were a packaged goods business,” Schmarzo said. The networking giant strayed too far from its core business, getting into the consumer market.
And Yahoo, which had the opportunity to buy Facebook for under $1 billion a few years ago, has similarly lacked focus on their core business.
“Even if they could have bought Facebook, for example, for $800 million, who’s to say that Facebook would have gone on to be successful,” Schmarzo said. “Facebook might have gotten buried by the Yahoo lack of focus.”