E4 Computer Engineering and Xinnor Equip University of Pisa with Advanced BeeGFS-Based Storage for AI Workloads

The University of Pisa, in partnership with E4 Computer Engineering S.p.A. and Xinnor Ltd., has implemented a groundbreaking high-performance storage solution to support its advanced AI workloads.

By integrating the company’s xiRAID with BeeGFS, the university has successfully met the demanding requirements of its NVIDIA DGX systems, critical to AI research.

NVIDIA DGX systems, designed for cutting-edge AI tasks, posed unique storage challenges, including high bandwidth demands, low latency requirements, and support for massive datasets with random I/O patterns. To overcome these obstacles, the University of Pisa collaborated with E4 and Xinnor to build a next-gen storage system. The solution includes dual storage nodes, each powered by xiRAID in RAID-6 configuration, delivering a fail-safe environment for large-scale AI operations.

“The collaboration between Xinnor, E4, and the University of Pisa exemplifies innovation in action,” said Davide Villa, CRO, Xinnor. “By integrating xiRAID, we addressed the university’s complex AI storage challenges, providing a solution that is as reliable as it is high-performing.”

The newly implemented infrastructure demonstrated outstanding performance, with read speeds reaching 29.2GB/s and write speeds of 25.8GB/s in tests involving up to 128 processes/node. This performance ensures the University of Pisa can efficiently handle data-intensive AI tasks, maintaining its competitive edge in research and development.

Maurizio Davini, CTO, University of Pisa, highlighted the impact of the new system: “The integration of xiRAID with BeeGFS has transformed our storage infrastructure, enabling us to push the boundaries of AI research. The system’s performance and reliability have exceeded our expectations, allowing our researchers to focus on innovation without being hindered by storage limitations. This collaboration with E4 and Xinnor has been instrumental in maintaining our leadership in computational research.”

As a long-term partner of the University of Pisa, E4 Computer Engineering played a pivotal role in designing and implementing the solution. By combining E4’s deep expertise in HPC and AI with Xinnor’s advanced storage technology, the project achieved both performance excellence and operational reliability.

“This project highlights the importance of tailored solutions,” stated Cosimo Gianfrida, CEO, E4 Computer Engineering, “The synergy between Xinnor’s xiRAID and BeeGFS, along with our custom monitoring tools developed by E4, ensures that the University of Pisa has a future-proof storage infrastructure that meets its evolving needs.”

The successful implementation of Xinnor’s xiRAID by the University of Pisa demonstrates how innovative storage solutions can drive progress in research and industry.

Resource:
Case study