Who IBM’s Server Group Turns To for Machine Data Analytics
IBM’s engineering prowess is second to none, and its Systems and Technology Group builds the computers that run the world’s biggest companies. But when IBM’s STG unit went looking for a way to predict failures by analyzing log data returned by its customers’ servers and storage arrays, it looked externally to a little-known machine data analytics startup from Santa Clara.
Glassbeam got its start five years ago, before the Internet of Things (IOT) became the industry’s hottest buzzword and out-hyped even “big data.” Since then, the small company has carved a niche for itself by providing an end-to-end machine data analytics service for device manufacturers in the IT, networking, and medical fields.
“We have customers from storage, networking, and medical device verticals, and these manufactures then sell devices to their customers,” explains Puneet Pandit, CEO and co-founder of Glassbeam. “They are using it for all kinds of log data from these devices, which is extremely useful to understand the supportability of these devices, what’s breaking, what’s working, and how to predict certain device failures.”
Glassbeam’s cloud-based service, called SCALAR, runs on Amazon and is based on the Cassandra NoSQL database. After the loosely structured machine data is uploaded into Glassbeam’s system, it’s parsed and transformed into more highly structured data with programs written in its proprietary Semiotic Parsing Language, or SPL, a high-level language designed specifically for working with machine-generated data.
SPL is about 10 times more efficient than writing in Perl or Python, Pandit says. “So what might take a week to write a [program] to parse log data from a Cisco router, we can do that in less than an hour because we have high-level constructs to understand machine data,” he tells Datanami. “Once the SPL is defined, then the whole system just wakes up and starts running analytics on the data.”
The company’s Glassbeam Explorer interface lets users explore and search raw data and events using a Lucene-powered search engine. It then provides two stages of analytics, including reports built from historical data, which is useful for capacity planning and license usage, as well as more in-depth data science engagements, where the company aims to predict devices failures.
IBM’s STG group started using Glassbeam a few years ago. The multibillion-dollar group builds and sells high-end hardware, including System z mainframe and Power servers that run Linux, AIX, and IBM i operating systems, and various storage devices, such as the enterprise-class DS8800 storage array, which scales well into the petabyte range.
“IBM is one of our largest customers,” Pandit says. “When they sell devices to top enterprise accounts, like Citibank or JP Chase Morgan, they’re under constant pressure to make sure the uptime is high, there’s no unplanned downtime, they are proactive in front of customers, telling them what features are coming down the path, and they should upgrade features.
“We help IBM in a big way because the machine data that comes from every device every day–we parse it in our cloud and provide the analytics and reports back to the IBM [customer service] organization,” he continues. “Five hundred people within IBM use Glassbeam solutions on a daily basis. They can figure out what customers are using what models, and what firmware and what OS, what kind of errors are happening in these environment, in 24-hour, one-week, one-month slices. All that makes them more proactive and predictive in front of their customers.”
IBM, of course, sells its own big data analytics software solutions under the InfoSphere HDInsight brand. IBM Software Group sells Hadoop solutions and real-time streaming solutions and is very much attuned to the IoT opportunity. Many of these solutions run on the Power platform, as well as the mainframe too.
Just the same, when STG needed an end-to-end solution for analyzing the phone-home data coming from its customers’ machines, it didn’t mind looking outside the Big Blue box. “At end of the day we are a best of breed analytics provider,” Pandit says. “When we go to big companies like Siemens, GE, and IBM, we’re talking to different business units, and many times, as you can imagine in these companies, the business units have the budgets and the pain points.”
“They are primarily in the IT data center market,” he says. “We are not in the IT market. We don’t sell solution to the IT managers or CIOs or operations people. We sell our solutions to the manufactures of those devices. Our buyers are typically VPs of support or engineering in product companies. We have seen certain instances where our customers have played with Splunk or Sumo Logic, but they come away after a few weeks saying the analytics they can provide is not as deep or structured as we can provide.”
Glassbeam yesterday announced a $2 million round of venture funding, bringing its total venture funding to more than $8 million. The company plans on using that software to continue funding R&D and also to expand sales and marketing, which is critical considering the immense interest right now in anything connected to the IoT.
The company appears well-poised to capture its share of the emerging market for machine data analytics. “Many manufactures think they can build their own analytic solution in house. But the reality of the situation is, it may look easy to start a project, but it becomes a pretty enormous project as they start doing that in-house,” Pandit says.
In addition to $2 million, the company announced that Kumar Malavalli, a Glassbeam co-founder, has taken an expanded role as its chief strategy officer. Malavalli has worked in the region for years and is in the Silicon Valley Hall of Fame. He was a co-founder at Brocade Communications before its IPO, and was widely considered to be the visionary behind Fiber Channel storage fabrics. Most recently, he led InMage, which he founded in 2001, to a successful acquisition by Microsoft.