This insanely fast big data startup uses only one server – and just got $7.4M in funding
Share on Facebook
Share on Twitter
Share on Google+
Share on Reddit
Share on Email

The SQream team. Photo Credit: SQream's blog

Israeli startup SQream recently announced raising $7.4 million in Series B funding to expand their data processing platform, which uses a single server with just one GPU accelerator – if you don’t know why that’s impressive, you soon will

On Monday, Israeli big data startup SQream Tech announced that it raised $7.4 million in Series B funding to expand its large-scale data processing platform. With applications in fields that need to analyze immense amounts of data, such as genome-related research, cyber security detection and prevention, IoT and telecom data management, and financial predictions and analysis (including detecting fraud), this 5-year-old company’s patent pending technology has unparalleled applications (forgive the parallel computing pun).

What is impressive about their solution is that its platform uses a single server with one GPU (graphic processing unit) accelerator. GPUs were originally used to process render images, but over the past 20 years, have been increasingly used for large-scale big data applications. For companies that analyze data using GPUs, which can process specific kinds of structured data inputs, they typically have to perform actions using a number of GPUs at once in a process that is commonly referred to as parallel computing (now the rest of you can understand the bad pun I made earlier). Since SQream’s solution uses only one GPU, it saves companies enormous costs.

In fact, SQream claims that its SQL (structured query language, a computing language used for relational databases that big data companies use to process large amounts of structured data) software platform can load, compress, and analyze up to 100 terabytes of raw data in close to real-time. Considering worldwide IP traffic is approaching two exabytes per month and that number is expected to reach 10 exabytes per month by 2019, such a solution could not be more timely.

Blumberg brings SQream to the next level

Blumberg Capital led SQream’s Series B round, bringing the company’s total funding raised to date at $11 million. As part of this investment, Blumberg Partner Alon Lifshitz is also joining SQream’s board of directors. CEO and Co-Founder Ami Gal and Co-Founder Kostya Varakin reportedly are also developing compression algorithms that can employ GPUs to “miniaturize the data center,” according to Gal.

As Gal explained to Geektime, this means that “We use many types of compression methods based on the data types. Miniaturization means both on the storage required – compression, decompression, chunkization, no need for extra storage for keys, cubicles, etc. – and on the processing required, meaning GPUs with thousands of cores can process extremely fast.” Suffice it to say, “If one can do so much in one node, the scale of the system can be huge,” he added.

In terms of the company’s next steps, he said that, “We’re focusing on North American markets and continuing the roadmap of the product to tackle even bigger challenges in the healthcare, genome, homeland security, cyber and IoT.”

Share on:Share
Share on Facebook
Share on Twitter
Share on Google+
Share on Reddit
Share on Email
Laura Rosbrow-Telem

About Laura Rosbrow-Telem


I am a social entrepreneurship enthusiast: This is what happens when a former social worker becomes a tech journalist. I mostly write about startups, technology, peace and justice issues, cultural topics, and personal stuff. Before Geektime, I was an editor at the Jerusalem Post and Mic.

More Goodies From Big Data


Google News update is challenge to Facebook on reliability

Google Translate upgrades neural machine translation for Hindi, Russian, and Vietnamese

What exactly is DataViz? Definitions for dummies

  • Kian

    Wow, just a single server, that’s impressive. Wonder how it all will work out.