Advances in storage are becoming crucial as HPC and cloud systems handle ever-increasing data, a trend expected to accelerate. For decades, parallel file systems like Lustre and GPFS have seen few structural changes, with limited integration of I/O support technologies that leverage hierarchical storage layers, such as node-local burst buffers. Recently, however, there has been a surge in innovations in data storage, processing systems, and I/O support software for several key reasons:

• Technology: The growing availability of persistent solid-state storage and storage-class memory technologies, capable of replacing both memory and disk, is opening up new possibilities for the design of storage systems.

• Performance requirements: Disk-based parallel file systems are no longer sufficient to meet the performance demands of high-end systems. However, the optimal use of solid-state storage and storage-class memory to achieve the required performance remains unclear. As a result, new approaches for integrating these technologies into HPC systems are being actively developed and evaluated.

• Application evolution: Data analysis applications, such as graph analytics and machine learning, are becoming increasingly critical in both scientific and commercial computing. I/O often presents a major bottleneck for these applications, particularly in cloud and HPC environments, where rapid turnaround or the integration of intensive computation and analysis is required. As a result, data storage, I/O, and processing demands are evolving, driven by the emergence of complex workflows that integrate computation, analytics, and learning.

• Infrastructure evolution: In the future, HPC technology will extend beyond dedicated supercomputing centers. Concepts such as “Embedded HPC,” “HPC in a Box,” “HPC in the Loop,” “HPC in the Cloud,” “HPC as a Service,” and “near-real-time simulation” will drive the need for new, small-scale HPC deployment environments. To enable a seamless “computing continuum,” a federation of systems and functions will be needed, with unified mechanisms for managing I/O, storage, and data processing across all participating systems.

• Virtualization and disaggregation: As virtualization and disaggregation become more widely adopted in cloud and HPC computing, the importance of virtualized storage is growing. Increased efforts will be required to understand its impact on performance.

Our goals for the ESSA Workshop are to bring together leading researchers and developers in data-related fields—such as storage, I/O, processing, and analytics—on extreme-scale infrastructures, including HPC systems, clouds, edge systems, and hybrid combinations of these. We aim to discuss advancements and potential solutions to the new challenges we encounter.

Workshop Chairs

• Chair: Chen Wang, Nanyang Technological University, Singapore

• Co-Chair: François Tessier, Inria, France

Program Chairs

• Chair: Jakob Luettgau, Inria, France

• Co-Chair:  Ahmad Tarraf, TU Darmstadt, Germany

Web and Publicity Chair

• Chair: Amit Samanta, University of Utah, USA