Splunk: Machine data defines big data hype
While many companies are now "waving the big data banner"--which can mean everything from in-memory computing to making sense of unstructured data--to capitalize on the growing customer interest, one company is looking to focus on another aspect of the market: mining machine data.
Splunk CEO and President Godfrey Sullivan noted that while big data in itself is not a new phenomenon, the emergence of Web transactions, online shopping, smart meters and mobile data has presented new aspects in this space.
Most, if not all, of these data-generating activities are generated by machines, too, which is why machine data is the "biggest component" and "most relevant" aspect of big data, he said.
Splunk defines machine data as those generated by applications, servers, network devices, security devices, and other systems that power one's business operations.
In Singapore recently to meet up with customers, Sullivan pointed out that many of Splunk's customers in Asia are from the telecom, financial services, and government sectors. The public sector, in particular, would use its analytics software for security monitoring and data forensics to keep a tight watch on the integrity of their infrastructure, he explained.
In big data, technologies and architectures are designed to economically extract value from very large volumes of data--including unstructured data--by enabling high-velocity capture, discovery or analysis.
Forced to diversify
While Splunk's strength is in mining heterogeneous infrastructure environments for data and security, with one third of its business security-based, Sullivan said its existing customers are pulling the company into areas such as Web and business analytics.
For example, media house NPR uses Splunk's software to track the number of successful music downloads customers get from its site each day.
Since the company has to pay artistes royalty fees per download, it needs to know how many of actual downloads are incomplete or unsuccessful. These are areas that rival offerings such as Adobe's Omniture tool are not able to track, noted Loo Chiew Hooi, channel sales manager for Southeast Asia at Splunk, who was also present at the interview.
Loo added that while Web analytics tools such as Omniture are great for reporting activities on one's Web site, it is unable to correlate data generated online with those produced offline, such as marketing campaign or sales reports.
Splunk's software is based on a freemium model, where anyone can download the trial software--which comes with a limited storage capacity and limited features--and start mining their company's information and analyze datasets from all machine sources.
The vendor earns its revenue from customers that want additional storage and enterprise functions in the software.
Because Splunk can mine data from any source, Sullivan realizes that one of his toughest jobs as CEO is to determine which areas the company should move into, and which it should avoid.
As such, he said he was prepared to take on only one new business segment per year, in order not to overreach. For 2012, Web analytics would be the company's focus, he revealed, as little customization is needed to bring the product to market.
Business analytics, on the other hand, would likely be 2013's focus, he added. This would require the company to work on tweaking the software's user interface and adapting to the languages enterprise data analysts use, such as SQL, to search for and mine information.
Splunk began shipping its product in 2006 and currently has some 4,000 paying customers in 75 markets. Founded in 2004, the Cupertino-based U.S. company has over 400 employees globally.