In the new era of the Extreme Data Economy, companies across many industries are utilizing data as an asset, above and beyond any product or service they offer. Yet unprecedented agility is required to operationalize artificial intelligence, keep business in motion and succeed in today’s post-big data economy. To enable this level of responsiveness, we continue to innovate at Kinetica and are excited to announce significant enhancements to our Insight Engine with the new version 6.2.
As businesses try to move from traditional batch-driven SQL analytics to more advanced augmented analytics using algorithms and AI on all types of data (structured & unstructured, static & streaming, human & machine-generated, long-lived & perishable), they are looking at adopting technologies that can power this analysis and these complex workloads in a single engine, in a single platform. This was pretty much unachievable until GPUs were made more general purpose.
Combining the brute force of GPUs, the speed of memory and the optimization of a columnar structure, Kinetica’s GPU database was born to allow users to run these complex workloads in a single engine.
With Kinetica 6.2, we take this a step further. 6.2 not only adds capabilities critical for operational analytical applications, but also eases machine learning (ML) workloads by offering faster performance and broader support for unstructured data. With 6.2, Kinetica is the only GPU-accelerated technology that supports a full range of SQL analytics.
The core features added include:
Database & SQL enhancements
- Expanded SQL language capabilities including those required for Online Analytical Processing (OLAP) and advanced analytical workloads (support for Rank, Partition, Window functions)
- Advanced view creation via SQL or Native REST to conquer high throughput aggregate workloads
- High-speed distributed key value lookups for operational analytical applications similar to what NoSQL databases provide for web apps
Machine Learning improvements
- Introduction of Kinetica File System to analyze, train models and run inference on unstructured data like images and videos more effectively
- Improved performance of machine & deep learning models within the engine
Even though businesses try to operationalize AI, it can be very difficult to drive data analysis and deeper insight with existing open source and AI tools. Data scientists typically rely on complex and specialized tools and hardware. They frequently need to copy large volumes of data into these specialized environments in order to build and train their models. Then, once it is working in the lab, it can be difficult to make that functionality available to business users.
With Kinetica, and now version 6.2, users can simplify and accelerate the entire deep learning pipeline and augment and accelerate their traditional OLAP workloads.
This is the first blog of a three-part blog series on the new capabilities in Kinetica 6.2. We’ll follow up soon with a deeper dive into some of the key capabilities and examples of applications that can be built with them. In the meanwhile, if you are interested in learning more, download our trial or try out our demo.
Resources to help you get started: