SYSTAP: Empowering Businesses with Cost-effective, Highly Scalable Graph Solutions

Graphs are a powerful means of representing information that has nodes and relationships. They are found across industries from social networks to telecommunications to biological networks. Enterprises face new analytic challenges in exploiting graph data at scaleto create business advantage and deliver new information services. Existing technologies for managing data within enterprises typically do not scale effectively when applied to graph problems. With the growing commercial demand for building analytics on graphs, SYSTAP has developed efficient graph solutions ranging from embedded graph databases to high performance graph analytics running across HPC clusters. SYSTAP’s solutions make graph analytics faster and cheaper by using fewer hardware and power resources, which dramatically improves the price to performance ratio. “We are building software solutions for big graphs that are accessible to a full range of businesses from very large scale HPC clusters to on-demand SaaS customers.” says Brad Bebee, CEO, SYSTAP.

The company’s flagship product is BlazeGraph™. A platform with both Semantic Web and Property Graph APIs with High Availability (HA) and horizontal scaling, BlazeGraph™ powers enterprises including EMC, AutoDesk, DARPA, and others within finance, pharmaceuticals, and bioinformatics. It is also widely implemented within public sector and research organizations. As a case study, EMC was in need of graph capabilities to launch their next generation Storage Resource Management (SRM), called ProSphere. SYSTAP provided the graph technology to EMC’s team of over 300 engineers. EMC experienced an exceptional growth in the ability to manage storage resource information by embedding SYSTAP’s software.

At the 2014 IEEE Bigdata conference, SYSTAP announced MapGraph™ HPC, an innovative and new technology that serves as a key solution to process very large graphs in real-time using GPU acceleration. MapGraph™ HPChas been demonstrated to achieve performance increases up to 10,000X faster than other CPU-based technologies on data-parallel graph analytics on GPU Clusters.“We believe it takes Facebook 10+ minutes using 200+ racks of servers to traverse their graph over a single iteration. With our GPU technology they could do it in seconds on a cluster with similar RAM,” remarks Bebee. This technologyhas immediate applicability to real-time fraud detection, social and communication network analysis, as well as many others.