The avalanche of data currently flowing through the enterprise world has created its own vocabulary. Categories of size are now necessary to explain how information technology systems handle massive amounts of information. In an effort to challenge conventional thinking around how file storage is handled, one company has branded itself as WekaIO Inc., named after the wekabyte or the equivalent of 1 trillion exabytes on the binary powers of 10 scale. In other words, it’s an unfathomable amount of data.

WekaIO has developed a next-generation massively parallel file system optimized through NAND flash and NVMe (non-volatile memory) to confront problems raised by small file and metadata-heavy workloads. “We’re the first file system to now bring parallel algorithms over NVMe,” said Liran Zvibel (pictured, right), co-founder and chief executive officer of WekaIO. “We get you lowest latency and highest throughput either on-premises or in the cloud.”

Zvibel visited the set of theCUBE, SiliconANGLE Media’s mobile livestreaming studio, and spoke with co-host John Walls (@JohnWalls21) and guest host Justin Warren (@jpwarren) during the AWS re:Invent conference in Las Vegas, Nevada. He was joined by Maor Ben Dayan (pictured, left), chief architect and co-founder of WekaIO, and they discussed the company’s management of workflows using GPU resources, a recently successful test versus a competitor, and WekaIO’s tiering functionality. (* Disclosure below.)

Feeding hungry GPUs

As graphics processing units and field programmable gate arrays, or FPGAs, have become instrumental for general purpose machine learning workflows, complex projects run into compute-bound issues. WekaIO solves that by bringing GPUs back into the picture, saving customers time and money, according to Zvibel.

“If you are working with GPU servers that are costly, they are usually just idling around waiting for the data to come to them,” Dayan explained. “We just unstarve these GPU servers and let you get what you paid for.”

Speed remains a key differentiator in the storage space. WekaIO’s technology is geared for customers with millions of small files, such as companies in life sciences or media/entertainment, and the firm recently engaged in a proof-of-concept demonstration with a competitor that involved testing speed and storage solutions on millions of files.

“The leading competitor finished the work in six and a half hours,” Zvibel recalled. “We finished the work in just under two hours.”

In November, WekaIO announced a partnership with Hewlett Packard Enterprise Co. to integrate its technology in the Apollo Gen10 high-performance computing platform. The company has been emphasizing its ability to let customers tier between flash and object storage, creating a better way to manage system economics.

“We take that tiering functionality and we couple it with our highest-performance snapshotting,” Zvibel said. “We enable our customers to scale their clusters to sizes they couldn’t even imagine prior to us.”

Watch the complete video interview below, and be sure to check out more of SiliconANGLE’s and theCUBE’s coverage of AWS re:Invent. (* Disclosure: WekaIO Inc. sponsored this segment of theCUBE. Neither WekaIO nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE