Gold Sponsors
Silver Sponsors
Bronze Sponsors
Featured Speakers
Lustre Tutorial – Introduction for New Users
SE Technical Manager
DDN
This training session is targeted for end-user who are familiar with parallel file systems and their use in high performance computing but are new to the Lustre file system. The aim of the session is to acquaint end-user to Lustre’s HPC pedigree then describe the key components, the architecture, understanding file striping and their use to acceleration application performance.
Lustre at Cambridge: Past, Present and Future
HPC Systems Specialist
University of Cambridge
This presentation highlights the evolution of Lustre at Cambridge University, covering its growth, technical improvements, and future plans. It touches on past deployments, current operational strategies, and upcoming goals focused on performance, scalability, and secure research environments. Key challenges like system migrations and storage demands are also discussed.
Weathering the “Data” Storm: The Met Office’s New Lustre Filesystems
Expert HPC Analyst
Met Office
The Met Office is moving toward a new supercomputer service that is six times more powerful than the existing one. The new Lustre filesystems installed need to be able to withstand peaks of TiB/s of duplex bandwidth and thousands of IOPS, while managing a volume of data of multiple petabytes per day. This talk presents the Lustre configuration used by the Met Office to manage the operational and research workflows, and the changes made to the atmospheric model to take full advantage of Lustre.
LNet Routing Between Omnipath and Infiniband - Experiences at UCL ARC
Principal Research Infrastructure Developer
UCL Arc
This presentation shares UCL’s experience managing Lustre file systems across Omnipath and Infiniband networks. It discusses system setup, performance observations, and practical issues encountered with LNet routing.
Lustre 2.17 and Beyond
Lustre CTO
Whamcloud
The presentation explores Lustre's roadmap through version 2.17 and beyond, emphasizing scalability, performance, and adaptability for modern HPC and AI workloads. It covers key advancements like hybrid I/O, client-side compression, metadata caching, and improved usability. Continued innovation supports exascale readiness, multi-tenancy, and storage efficiency, with evolving tools and features for users and administrators alike.
Charting the Future of EXAScaler
Senior Systems Engineer
DDN
This presentation explores the future of DDN EXAScaler, emphasizing performance, scalability, and simplicity for AI and HPC workloads. It introduces features like client-side compression, hot/cold tiering, local NVMe caching, tenant isolation, and encryption. Enhanced management tools and online upgrades reduce complexity. EXAScaler continues to evolve for maximum efficiency and AI-ready storage performance.