The Israeli startup announced that it created a platform that could analyze 18 billion anonymized call data records on a single GPU accelerator
Israeli startup SQream Technologies announced that it was able to execute 18 billion anonymized call data records on a single NVIDIA GPU accelerator, showing that a GPU-accelerated database is able to exceed the performance of leading vendors in the field.
SQream, along with European telecommunications company Orange, was able to analyze the equivalent of four months worth of multimedia IP traffic, according to the announcement made at the 2014 NVIDIA GPU Technology Conference in San Jose, Calif.
SQream’s GPU-accelerated solution consists of a single server with one NVIDIA Tesla K40 GPU accelerator, making its solution cost less than existing platforms. The company said its data platform will allow enterprises to achieve analytics once considered unattainable.
Faster, with less hardware
“We have basically created a data-center in a box and we are currently making our beta product available to more customers,” Ami Gal, CEO of SQream, said in a March 26 statement. “Orange Silicon Valley is an important innovation partner in exploring technology frontiers in Big Data. We are proud to have such an esteemed telecom’s innovation center working with us and we look forward to working with other leading enterprises.”
Founded in 2010, SQream is a high tech startup with 25 employees specializing in big data and GPU technology.
In Wednesday’s statement, SQream said that worldwide IP traffic is approaching two exabytes per month and that number is expected to grow every two years and reach 10 exabytes per month by 2019.
Orange Silicon Valley said using SQream’s DB, it was able to analyze 18 billion call detail records in a shorter time, and with less hardware.