ScaleOut Software has introduced a new Analytics Server that it says does data analytics quickly by using in-memory data grid technology to process rapidly changing data.
ScaleOut Analytics Server is a follow-on product to the ScaleOut State Server middleware product that’s been out for about seven years, said Bill Bain, CEO of ScaleOut Software. State Server is deployed across a cluster of servers doing big data analysis in memory rather than on a separate database server. It’s ideal for analyzing rapidly changing data such as shopping carts on an e-commerce site, stock trades at a financial firm or activity on an electric utility’s power grid.
The Analytics Server adds new features including parallel data analysis, which makes it possible to analyze data in place, minimizing data movement and speeding analysis, something that can’t be achieved by pulling information from a database, Bain said.
“Database servers will bottleneck if you throw too much load at them,” he said.
Bain cited Ovum research that the typical in-memory data grid can handle up to 10TB of data and that the mean workload is 3TB.
Also new in the Analytics Server is more simplified object-oriented programming for developers to write applications in familiar languages such as Java and C#. It also adds automated code shipping, which delivers analytics code to grid servers for faster and easier execution.
ScaleOut Analytics Server is also available for deployment in public and private clouds, said David Brinker, chief operating officer (COO) of ScaleOut.
“The bulk of our business is still on-premise, but we are seeing a lot of interest in the cloud, so we have people evaluating us in the cloud at this point,” said Brinker.
ScaleOut Analytics Server will be available in the Amazon Web Services cloud and in Microsoft Windows Azure-based clouds, he said. Azure is the cloud version of the Windows Server data center operating system.
In-memory data grid technology processes data much faster than do previous big data analytics platforms that retrieve data from a file system such as the open source Hadoop and Google’s MapReduce programming model, said Bain.
“We’ve seen an order of magnitude faster analysis than Hadoop by doing memory-based computing,” he said. “If you start to analyze data from a file system, you have to move all that data into memory for analysis so that really adds to the latency of analysis.”
Big data analytics was also the focus of attention at the recently concluded Oracle OpenWorld conference in San Francisco. During a presentation on Oct. 4, Balaji Yelamanchili, senior vice president of the Oracle analytics and performance management products unit, said that Oracle’s Exalogic and Exadata appliances—in which the hardware, processors, operating system and software are all Oracle-branded products—also deliver data analytics results in real time.
In a keynote preceding Yelamanchili’s remarks, Oracle President Mark Hurd shared data points about the rapid expansion of data on networks and within enterprises. He said that 90 percent of the world’s data today was created just within the last two years, that the amount of data is expected to grow 50-fold by 2020 and that the amount of mobile data alone is expected to increase at a compound annual growth rate of 78 percent.
ScaleOut’s offering for data analytics in real time is a new approach to crunching large amounts of data, according to Philip Howard, research director in the data management field for Bloor Research.
“As far as I know, this is the first in-memory data grid to specifically target analytics,” wrote Howard in a blog post. “It is symptomatic of what is going on in the data warehousing and big data space in general, in that it represents yet another approach to an existing problem.”