SQream Technologies raises $39.4 million for GPU-accelerated databases


Data set analytics startup SQream Technologies today announced it has raised $39.4 million. The bulk of the new funds will go toward talent recruitment, product R&D, and enhancing the company’s customer delivery platforms, according to a spokesperson. SQream aims to lower the barrier to big data optimization with better performance, a reduced footprint, and cost savings over its competitors.

Forbes found that just over 50% of companies are adopting big data analytics, with 95% citing the need to manage unstructured data. This uptick is also likely attributable to ROI adopters report seeing — Entrepreneur notes that businesses using big data experience a profit increase of 8% to 10%, on average. But obstacles remain, with 40.3% of respondents to a Statista survey suggesting big data usage was held up by a lack of organizational agility.

SQream’s column-oriented database product — SQream DB — is a GPU-accelerated data warehouse capable of handling complex queries with features cherry-picked from relational database systems. Using on-premise and cloud acceleration, SQream claims its customers analyze trillions of records and load up to 3 terabytes per hour per GPU out-of-the-box.

SQream DB decouples storage from compute, eliminating the requirement to copy, redistribute, or repartition data when teams grow. It integrates with business intelligence tools via a wide selection of drivers and connectors, including ODBC, JDBC, Python, Node.JS, Spark, R, Java, and C++. SQream DB’s interface layer serves as a collection of services that control the data warehouse, while the compute layer is where the actual data processing tasks are run. The last layer — the storage layer — is split into a metadata module where all routine database objects are stored and a persistent bulk data component that’s heavily optimized for raw table-scan performance.

VB Transform 2020 Online – July 15-17. Join leading AI executives: Register for the free livestream.

SQream DB can read data directly from external sources with the external table syntax, and its columnar storage system is partitioned horizontally and vertically for analytic operations like joins, aggregations, summarizations, and sorting. The columnar engine allows selective access to the required subset of columns, reducing disk scan and memory I/O versus standard storage. Meanwhile, hyper-partitioning splits up storage horizontally into manageable chunks, complementing AI-assisted compression that matches compression schemes with given corpora.

Admins can control access with SQream DB’s role-based permission system, and the platform scales automatically with the addition of storage and compute nodes. It’s designed to work on any Linux and Nvidia CUDA-capable server. Growing SQream DB in either direction doesn’t affect data availability or integrity, the company says.

While GPU-accelerated databases like SQream are well-suited to certain workloads, it’s unclear whether they’ll move into the mainstream anytime soon. That’s because they tend to perform poorly on database operations that can’t be parallelized or that don’t involve floating point and other numeric processing. Plus, they struggle to stand out — SQream competes with vendors like BlazingDB, Kinetica, and OmniSci (formerly MapD), among others.

The company says it’s specifically targeting AI and data analytics applications requiring high database throughput. One customer — AIS, one of Thailand’s largest mobile phone operators — deployed SQream DB to reduce queries of millions of raw data records from one hour to less than 50 seconds, SQream claims. Another, Sheba  Cancer Research Institution, reportedly uses the platform to analyze up to 1 petabyte of genome sequences.

Mangrove Capital Partners and Schusterman Family Investments co-led this latest raise, with participation from existing investors Hanaco Venture Capital, Sistema.vc, World Trade Center Ventures, Blumberg Capital, Silvertech Ventures, and Alibaba Group. The series B+ brings SQream’s total raised to over $50 million. (SQream partnered with Alibaba in 2018 to offer the former’s GPU-accelerated database as a service to Alibaba Cloud customers.) As a part of the round, Silvertech Ventures’ Charlie Federman and Mangrove’s Roy Saar will join Tel Aviv-based SQream’s board of directors.

Sign up for Funding Weekly to start your week with VB’s top funding stories.



Source link

Previous post The US-China Battle Over the Internet Goes Under the Sea
Next post After OTF takeover, USAGM CEO Michael Pack faces pressure from lawmakers and courts