Grant to Fund Research into Dramatically Improving the Performance of Big Graph Databases
Lucata, provider of a next-generation server platform for accelerating and scaling graph analytics, AI and machine learning (ML), today announced it has been awarded a National Science Foundation (NSF) Small Business Innovation Research (SBIR) Phase 1 grant. Lucata will use the grant to fund research into how to dramatically improve the performance of large graph databases through leveraging the power of the Lucata next-generation computing architecture. The research will focus on enabling write updates to a graph database without locking the entire database, allowing database reads at the same time new data is being written to the graph.
Read More: SAI360 Placed Leader In 2021 Gartner Magic Quadrant For IT Vendor Risk Management Tools
A major limitation of today’s graph databases is the need to lock the entire database while new data is being written, which prevents database reads during the writes. Correspondingly, when data is being read, new data cannot be written to the graph. This locking/unlocking process significantly limits the overall performance of graph databases. The SBIR grant will enable Lucata to explore the feasibility of and market interest in a high-performance graph database architecture that allows simultaneous reads and writes of data. Leveraging the company’s patented Migratory Threads technology, Lucata will develop a system that locks only a local area of the graph database while data is written to that area, allowing reads to continue on the remaining unlocked data. For large graph databases where the likelihood of a simultaneous read and write to the same area of the graph are low, locking only a limited area of the graph for writes will allow reads to continue with little or no delays due to writes.
“We are extremely pleased the NSF has recognized the potential of our innovative graph database architecture. This grant will enable us to explore how to solve one of the major obstacles to optimizing large graph database performance for a range of increasingly critical use cases,” said Marty Deneroff, Lucata CTO. “Our patented Migratory Threads technology can handle data locks much more efficiently than traditional technologies, potentially making Lucata the ideal platform for solving the thorny performance challenges due to write locking/unlocking that currently limit the performance of large graph databases.”
Read More: QNET Wins Six New Industry Awards For Innovation And Social Impact
The Lucata computing architecture leverages patented Migrating Thread technology to enable organizations to leverage massive pools of physical memory to accelerate and scale graph analytics and AI and ML model training by orders of magnitude beyond the capabilities of existing approaches. The solution enables high-performance exascale graph analytics, including exhaustive breadth-first search (BFS), on unpruned, unsharded large graph databases. Lucata can be used with open source or commercial graph software or with custom-written graph solutions that leverage LAGraph, GraphBLAS, or the Lucata library of search queries, enabling organizations to use their existing software to uncover much deeper connections within much larger graphs than possible today. These unique capabilities allow organizations to reimagine the capabilities of graph analytics, AI and ML and address previously intractable challenges in fraud detection, cybersecurity, blockchain, risk assessment, healthcare and many other fields.