News

Kinetica GPU-Accelerated Database Runs on GPU-Powered Amazon EC2 Instances

• By David Ramel
• 05/09/2017

Kinetica's GPU-accelerated, in-memory database is now available on Amazon Elastic Compute Cloud (Amazon EC2) P2 instances, which are themselves powered by graphics processing units (GPUs).

The new family of EC2 P2instance types was introduced last September by Amazon Web Services Inc. (AWS), which called them "the largest GPU-powered virtual machines in the cloud."

With the advent of Big Data and other cutting-edge computing workloads that demand more computing power than can be delivered by traditional central processing units (CPUs), GPU chips have been put to use in ways that go beyond just displaying images and other graphics. Those workloads include deep learning, computational fluid dynamics, computational finance, seismic analysis, molecular modeling, genomics, rendering workloads and so on.

In the data development field, companies such as Kinetica and MapD Technologies (which just open sourced its own GPU-powered database) are leading the way with products such as Kinetica's flagship offering.

Sporting GPU-accelerated analytics, in-memory processing, scale-out capabilities and more, the Kinetica database -- which has been available on the AWS Marketplace since 2014 -- now gets a performance boost from the GPU-powered P2 instances. The company said the new P2-based offering can help customers deploy its database as a turn-key analytics solution without upfront capital investments.

"As customers embrace heavier GPU compute workloads such as accelerated analytics, artificial intelligence, high-performance computing, and Big Data processing, they need higher performance," Kinetica said in a statement today. "P2 instances are powerful GPU instances available in the cloud, providing top-notch performance for compute-intensive workloads such as risk modeling, portfolio optimization, fraud detection, energy exploration and real-time route and inventory optimization. The P2 instances build on Amazon's strengths of offering elastic, secure, and manageable solutions for faster time to value."

The distributed database can scale horizontally or vertically as needed, is integrated with several security solutions and provides location-based analytics, advanced high availability and in-database analytics, Kinetica said.

"A User Defined Functions (UDF) framework to run custom code and open source machine learning libraries such as TensorFlow and Caffe natively in the database with GPU-acceleration for machine learning, deep learning, and fast OLAP," the company said.

Data developers can also leverage many integrations and connectors.

"Kinetica includes industry-standard connectors to make it easy to integrate with existing infrastructures. Kinetica's APIs are fully supported in REST, Java, Python, C++, JavaScript and Node.js. Kinetica ships drivers for integration with industry-standard BI and SQL tools; features full SQL-92 query support through certified JDBC and ODBC connectors. Open source integration components include: Apache NiFi, Apache Spark and Spark Streaming, Apache Storm, Apache Kafka and Apache Hadoop. Also connect with common BI tools like Tableau, Kibana and Caravel."

Along with the AWS announcement, Kinetica also today announced its database was available on Microsoft's N-Series of Azure Virtual Machines with GPU capabilities in the Microsoft Azure cloud.