Singapore University Enhances AI Research with High-Performance BeeGFS Cluster Powered by xiRAID
Infrastructure includes 2xserver nodes equipped with 24 NVMe drives each, protected and accelerated by xiRAID.
This is a Press Release edited by StorageNewsletter.com on December 30, 2024 at 2:00 pmA cutting-edge storage solution powered by Xinnor Ltd.’s xiRAID technology has propelled a leading university in Singapore to new heights in AI research and HPC.
The solution, architected by On Demand Systems Pte Ltd (ODS), showcases unprecedented data access speeds, saturating 2x100Gb InfiniBand ports with a measured sequential read performance of 24.7GB/s.
The university’s deployment, built to support advanced AI projects in fields like healthcare and natural language processing, features a BeeGFS cluster. The infrastructure includes 2xserver nodes equipped with 24 NVMe drives each, protected and accelerated by xiRAID. This configuration not only delivers exceptional throughput but also ensures data protection and scalability for future growth.
“xiRAID’s ability to achieve near-perfect resource utilization and maximize NVMe performance is a game-changer for academic research,” said Davide Villa, CRO, Xinnor. “Our collaboration with ODS demonstrates how software-defined RAID can meet the rigorous demands of AI and HPC workloads while optimizing costs and simplifying deployment.”
ODS’s leadership in deploying high-performance systems was key to the project’s success. “With xiRAID at the heart of this solution, we have enabled the university to push its boundaries in research and innovation,” said Rakesh Sabharwal, founder, On Demand System Pte Ltd. “The seamless integration of xiRAID with BeeGFS has ensured a reliable, high-performing storage backbone that matches the university’s ambitions.”
This deployment highlights the critical role of xiRAID in advancing data-centric research. Its RAID-6 configuration ensures data reliability even amidst multiple drive failures, while its dynamic use of CPU cores minimizes overhead. As a result, researchers can access data faster, drive GPU-accelerated computing tasks efficiently, and focus on groundbreaking discoveries.
The collaboration between Xinnor and ODS illustrates the transformative potential of combining innovative technology with domain expertise. As AI and HPC demands continue to rise, this solution sets a benchmark for academic institutions WW.
Resource:
Case study (PDF)