Lustre in HPC, AI and the Cloud: Scope, Community, Accomplishments, Challenges and Roadmap
Wednesday, June 11, 2025 10:15 AM to 11:15 AM · 1 hr. (Europe/Berlin)
Hall G1 - 2nd floor
Birds of a Feather
Community EngagementExtreme-scale SystemsHeterogeneous System ArchitecturesParallel File SystemsStorage Technologies and Architectures
Information
Lustre, the leading open-source and open-development parallel file system for HPC was in use at 9 of the world’s TOP10 supercomputers in November 2024.
Today Lustre is one of the most widely adopted technologies in HPC around the world, from NVIDIA’s Eos system to the LUMI system in Finland, and the Frontier system at ORNL, to the Fugaku computer at RIKEN in Japan and the Leonardo system at CINECA in Italy. It is also in use at around two thirds of the TOP100 systems as well as at the majority of smaller HPC systems.
Lustre was historically driven mainly by U.S. National Labs but has widened its scope beyond its traditional stronghold to include high-end AI/ML and extreme cloud infrastructures. It is a community-developed technology with contributors from around the world. Lustre clients are available for broadly deployed instruction set architectures such as x86, Power and ARM.
Vital to the technology is the Lustre community that continues to drive Lustre forward. As Lustre has an established true open-source and open-development model, end users, developers and solution providers regularly come together through the worldwide OpenSFS and EOFS communities both being present at the BoF.
This community development model has resulted in significant new features, improved stability and broader adoption.
Lustre is also widely used across mid- and small-scale HPC-systems with continued adoption attributable to the stability of the Lustre file system as it has matured.
On the other side we will also discuss emerging use cases like the scalability challenges associated with exascale systems. We will explore these cases and discuss the Lustre roadmap for meeting the new requirements that they present.
At this BoF, Lustre developers, administrators, and solution providers will also discuss technical details of recent developments, such as the recent Lustre 2.16 release, continuing LTS support for Lustre 2.15, as well as issues, new challenges and corresponding opportunities.
A key part related to Lustre operations will be the discussion of best practices, optimizing file system access, metrics, and other Lustre-related items.
Looking beyond the current releases the community will discuss planned features for Lustre 2.17 like Client Data Compression, Metadata Writeback Cache and Erasure Coding as well as RHEL10 Server support and Metadata Redundancy planned for Lustre 2.18
Organizers:
Today Lustre is one of the most widely adopted technologies in HPC around the world, from NVIDIA’s Eos system to the LUMI system in Finland, and the Frontier system at ORNL, to the Fugaku computer at RIKEN in Japan and the Leonardo system at CINECA in Italy. It is also in use at around two thirds of the TOP100 systems as well as at the majority of smaller HPC systems.
Lustre was historically driven mainly by U.S. National Labs but has widened its scope beyond its traditional stronghold to include high-end AI/ML and extreme cloud infrastructures. It is a community-developed technology with contributors from around the world. Lustre clients are available for broadly deployed instruction set architectures such as x86, Power and ARM.
Vital to the technology is the Lustre community that continues to drive Lustre forward. As Lustre has an established true open-source and open-development model, end users, developers and solution providers regularly come together through the worldwide OpenSFS and EOFS communities both being present at the BoF.
This community development model has resulted in significant new features, improved stability and broader adoption.
Lustre is also widely used across mid- and small-scale HPC-systems with continued adoption attributable to the stability of the Lustre file system as it has matured.
On the other side we will also discuss emerging use cases like the scalability challenges associated with exascale systems. We will explore these cases and discuss the Lustre roadmap for meeting the new requirements that they present.
At this BoF, Lustre developers, administrators, and solution providers will also discuss technical details of recent developments, such as the recent Lustre 2.16 release, continuing LTS support for Lustre 2.15, as well as issues, new challenges and corresponding opportunities.
A key part related to Lustre operations will be the discussion of best practices, optimizing file system access, metrics, and other Lustre-related items.
Looking beyond the current releases the community will discuss planned features for Lustre 2.17 like Client Data Compression, Metadata Writeback Cache and Erasure Coding as well as RHEL10 Server support and Metadata Redundancy planned for Lustre 2.18
Organizers:
Format
On Site
Targeted Audience
Providers of commercial and/or community-based Lustre deployments
Developers of features for management and control.
System architects involved with proposals, design, deployment and evaluation of file systems
System administrators in charge of file and storage management
Representatives from academia and industry with a focus on system architecture, operating systems and middleware
Speakers
Frank Baetke
Chairman of the BoardEOFS, Frank Baetke
Jacques-Charles Lafoucriere
Program Manager HPC & Big DataFrench Alternative Energies and Atomic Energy Commission
Pietari Hyvärinen
Senior System SpecialistCSC - IT Center for Science Ltd
Peter Jones
Senior Director of Lustre EngineeringDDN
Andreas Dilger
Lustre Principal ArchitectDDN