Intel and Katana Graph work together to accelerate large-scale graph analysis efficiency

Intel and Katana Graph announced a collaboration to port and optimize the Katana Graph engine on Intel’s portfolio of products, including Intel® Xeon® Scalable processors, Xeon-based compute clusters, and the upcoming Intel discrete GPU family, including the GPU codenamed “Ponte Vecchio.” Intel and Katana Graph will leverage high-performance, scale-out parallel computing to help customers solve large-scale unstructured data problems with unparalleled efficiency.

Large-scale, unstructured data sets are often used in social network analysis, security and authentication, electronic chip design tools, biomedical and pharmaceutical applications (gene network analysis and medical knowledge graph mining), and epidemiological studies for simulating the spread of infectious diseases. Intel's broad technology ecosystem can help customers accelerate analysis at every stage of the data cycle. For example, Intel Xeon Scalable processors can analyze large amounts of data at high speed, while Intel® Optane™ persistent memory technology can help customers overcome data movement and storage bottlenecks.

“In order to conduct deep analysis of large unstructured data and use it at scale, it needs to be deployed and executed on both CPUs and GPUs. Our collaboration with Katana Graph will accelerate the adoption of graph analytics on Intel products, including the market-leading Intel Xeon Scalable processors and upcoming GPU products, enabling more customers to benefit from graph computing,” said Wei Li, vice president of Intel Architecture, Graphics and Software and general manager of Machine Learning Performance.

Katana Graph Engine is the leader in scale-up and scale-out analytics, and it can run on a variety of platforms and technology combinations, including large computing clusters of x86 CPUs, large memory systems with Intel Optane persistent memory, single-node or multi-node GPU platforms, or any combination of these technologies. In addition, it can scale to hundreds of machines in production clusters.

“Computing on large, unstructured datasets is the paradigm of the future,” said Keshav Pingali, CEO and co-founder of Katana Graph. “Unlike other companies in this space, Katana has a dedicated high-level programming model and runtime for processing graph and hypergraph applications. That’s why our analytics library is orders of magnitude faster than other vendors’ solutions. We are excited to work with Intel to bring high-performance, scalable graph computing to our mutual customers.”

Katana Graph Engine is already being used by customers including:

A large defense contractor is using Katana Graph to address security issues by deploying a real-time intrusion detection system across computer networks. The system builds an online interactive graph showing how network users interact with each other and with network resources.

Electronic design automation companies are evaluating Katana Graph to implement high-performance parallel modules to solve electronic circuit design problems, including logic synthesis, hypergraph partitioning, placement, and global routing. The graph engine is also being used to build unstructured representations to model contamination and epidemiological research projects on the spread of COVID-19.

Katana Graph's enterprise system is supported across a variety of infrastructure deployment models, such as on-premises, hybrid, and large cloud platforms including AWS and Microsoft Azure. The system provides a full-featured graph database with scalable parallel query, execution for long-running computations, and extensive support for knowledge graphs.