In-memory databases are not a new technology, but with data sets getting bigger and businesses looking for insights faster, real-time data analytics has reached a new level of interest. And executives at Kognitio, which has been in the in-memory field since 1989, feel the market has come around to their way of thinking.
Kognitio has been building its massively parallel processing (MPP) system for 23 years. CTO Roger Gaskell said the company, which started off as Whitecross Systems, first relied on its own custom-built hardware. Doing this, Gaskell said, meant the company learned lessons about parallel processing that sped up performance before commodity hardware arrived in the last decade. In essence, he said, Kognitio was developing its own software and building its own blade servers before blade servers came on the market.
At the Hadoop Summit in San Jose on June 13, Kognitio committed itself to the data management platform. Hadoop integration will be an “integral part of the Kognitio Analytical Platform,” the company said, and Kognitio will embed its software agent inside Hadoop clusters to enhance the speed of ad-hoc queries, loading as much as 15 terabytes an hour. During its barnstorming tour of the U.S., Kognitio announced partnerships with high-profile businesses to work as their in-memory partner, like MetaScale, AVS, and Hortonworks.
About the technology: Businesses want real time access to their ever-increasing data for quick analysis, but the industry’s hottest technologies, like Hadoop, still take dozens of minutes to process queries.
An in-memory platform is database management system that relies on main memory for storage instead of using disk storage, greatly reducing I/O reading and increasing speed. For analytics, it provides a much faster environment for crunching data than traditional data warehousing.
Kognitio offers three layers of its in-memory product: an in-memory analytical platform, software licenses so companies can run the MPP system on their own hardware, or cloud-based services so companies can crunch data quickly and pay per project.
The mechanics of in-memory analytics are well-known—data accessed in memory means faster query results than data accessed from disk. Philip Howard, a research analyst at the Bloor Group, said while the cloud and in-memory are no more well suited for each other than any other warehouse technology, data crunched in memory is definitely faster and, “you’d really like all relevant data in-memory at all times.”
He said not all in-memory platforms are created equal. “Kognitio has been doing it longer than anyone else so it has more experience,” he added.
Use cases: Kognitio’s existing customers include global organizations processing queries against large data sets. Aimia, a company that manages customer loyalty and rewards programs for large retail chains like CVS, uses Kognitio’s software to analyze customer data for its clients. British Telecom has used Kognitio’s analytics since 1993 to provide an intranet cloud-like service for its employees.
Companies that use Kognito’s cloud-based analytics software can access the technology through Amazon Web Services.
More on the Hortonworks partnership: The partnership with Hortonworks Represents an important step, Gaskell said. It should help users looking to match Hadoop’s batch processing capability with Kognitio’s in-memory analytics expertise.
“What we’re doing is producing a fairly tight integration between Kognitio and Hadoop,” Gaskell said, who was employee number two at Kognitio. “We’re not just providing connectors so you can pull data sets out. What we’re doing is providing a connector that actually passes part of the workload down to Hadoop. We get Hadoop to do what it’s good at, and then we do what we’re good at, the crunching of the data.”
About the company: Privately-held Kognitio was called Whitecross Systems before merging with a consulting company with its current name in 2005. The company has done most of its business in the U.K. and continental Europe until now.
Email Staff Writer Ian B. Murphy at email@example.com.