Katana Graph is developing the next generation of graph computing
Datasheet

The Katana Graph platform creates competitive advantage by providing actionable insights from massive unstructured data sets, using high-performance graph algorithms. Graph adoption is accelerating at a rapid pace as graphs provide richer insights from context and connections represented in the data when compared to traditional analytics on tabular data formats. Knowledge Graphs are powering a new era of rich applications and insights never before possible in domains such as Pharma, Security, FinTech and Manufacturing.

Product Overview

Katana Graph is a cloud 3.0 data platform offering elastic performance at massive scale for Graph Queries (contextual search), Graph Analytics (path finding, centrality and community detection), Graph Mining (pattern discovery), and Graph AI (prediction). Katana Graph achieves best-in-class scale-up and scale-out performance by incorporating disruptive technologies like hardware acceleration (GPUs, ASICs, etc.), storage (non-volatile memory), and infrastructure (Kafka, OpenMPI, etc.) along with the ease of programmability.

“Katana Graph has leveraged the innovative architecture of the 3rd Gen Intel Xeon Scalable processor in its state-of-the-art graph analytics engine, which is exciting news.”

Wei Li
Vice President and General Manager of Machine Learning Performance in the Architecture, Graphics, and Software group at Intel

Datasheet Venn Graph

Value Proposition

Katana Graph is a partner that enables businesses to tackle evolving business challenges through innovation and flexibility. Katana Graph shortens the time to insight by outperforming competing technologies in programmability, scalability and performance to extract meaningful, contextual answers that power data-driven decisions.

Performance

  • 10-100x faster when compared to other leading solutions

Developer Friendly

  • Wide array of programming support for the broadest set of developer use cases
  • C/C++ (Core), OpenCypher (Query), Python (Analytics, Mining)

Scalabilty

  • Validated at >256 machines
  • On-prem supercomputing and Public cloud support (WS, GCP, and Azure)
  • Supporting complex workloads on petabyte datasets representing >trillion edge graphs without sacrificing performance
  • Verified on WDC12: 3.5B nodes, 128B edges

Flexibility

  • Hardware Optimized CPU, GPU, Optane memory, Heterogeneous execution and Distributed clusters
  • Graph Partitioning – Supports custom (including 1D, 2D, and hybrid vertex and edge cuts) & user-defined policies
  • Optimized partition-aware communication and compute
  • Storage and compute separation allowing for data workload–driven deployments

Technology Overview

At the heart of Katana Graph’s solution is the Katana Graph Engine with its accompanying customizable partitioner, communication, hardware virtualization and storage technology modules, which are the culmination of more than a decade of advanced research in graph technology and high- performance computing.

System Architecture
Icon: Graph Query

Lightning-Fast Graph Query

The query engine supports OpenCypher, a popular graph query language that allows users to search for complex graph patterns, update the graph and perform analytic operations like aggregations across matched portions of the graph. 

Icon: Graph Analytics

Graph Analytics

An extensive library of graph analytics routines that provide global insights from graph relationships. Use cases include: Finding well-connected nodes in a graph (influencers), ranking nodes (webpages), community detection (identifying clusters of similarity for fraud or segmentation), and path finding (routing).

Icon: Graph Mining

Graph Mining

Powerful library of graph mining algorithms to quickly identify explicit or implicit patterns. Applications include frequent pattern mining, k-motifs, cliques, etc.

Icon: Graph AI

Graph AI with Graph Neural Networks

Graph neural networks (GNN) are a new powerful approach for feature learning on graphs. Katana provides easy-to-use and scale-out packages for learning large-scale knowledge graph embeddings. Use cases include applications such as node classification, link prediction, recommendation systems in industries like bioinformatics and cheminformatics. 

Icon: Cloud Deployment

Automated Cloud Deployments

Push-button cluster provisioning and management of cluster resources allow ease of use.  Katana Graph is cloud agnostic and works in on-prem and off-prem environments. 

Performance Highlights

Katana is 10-100x faster than competing systems.

Input: Friendster (66M nodes; 1.8B edges);

AWS: 8 cores; 64 GB each

Benchmarks: Page Rank (10 iterations)

Input: Friendster (66M nodes; 1.8B edges);

AWS: 8 cores; 64 GB each

Benchmarks: Page Rank (10 iterations)

Input: LDBC Scale-1 (Nodes=3.2M, Edges=17.3M);

AWS: 8 cores; 64 GB each

Benchmark: 3-Node query

Driving the next generation of graph computing

We believe in the power of data. At Katana Graph we are developing technologies to help people and businesses unleash the immense potential of their large scale irregular and unstructured data.

CONTACT US