Agenda
The workshop will be held in Santa Fe, NM, USA on Tuesday, October 31, 2023 in Canyon room.
Note: All presenters and attendees must register at the IEEE Cluster website.
Agenda: (all times are in Mountain Daylight Time / MDT)
14:00 - 14:10 – Welcome Message & Speed Introduction
14:10 - 15:00 – Keynote: HPC Storage: Adapting to Change [Slides]
Speaker: Phil Carns, Argonne National Laboratory, USA
15:00 - 15:30 – Work-in-Progress Talks (10 minutes per talk, including Q&A)
WiP Talk I: Accelerating I/O in Distributed Data Processing Systems with Apache Arrow CHFS, Sohei Koyama, Kohei Hiraga and Osamu Tatebe. University of Tsukuba, Japan. [Slides]
WiP Talk II: Does Varying BeeGFS Configuration Affect the I/O Performance of HPC Workloads?, Arnav Borkar, Joel Tony, Hari Vamsi K. N, Tushar Barman, Yash Bhisikar, Sreenath T. M. and Arnab K. Paul. BITS Pilani K K Birla Goa Campus, India. [Slides]
WiP Talk III: DAOS as HPC Storage: Exploring Interfaces, Adrian Jackson and Nicolau Manubens. The University of Edinburgh, UK and European Centre for Medium-Range Weather Forecasts, Germany. [Slides]
15:30 - 16:00 – Coffee Break
16:00 - 17:15 – Research Paper Talks (25 minutes per talk, including Q&A)
Paper I: I/O-Aware Flushing for HPC Caching Filesystem, Osamu Tatebe, Kohei Hiraga and Hiroki Ohtsuji. University of Tsukuba, Japan and Fujitsu Ltd., Japan. [Slides]
Paper II: Mango-IO: I/O Metrics Consistency Analysis, Radita Liem, Sebastian Oeste, Jay Lofstead and Julian Kunkel. RWTH Aachen, Germany, TU Dresden, Germany, Sandia National Laboratories, USA and Goettingen University / GWDG, Germany. [Slides]
Paper III: An I/O Performance Evaluation of Varying CephFS Striping Patterns, Debasmita Biswas, Sarah Neuwirth, Arnab K. Paul and Ali R. Butt. Virginia Tech, USA, Johannes Gutenberg University, Germany and BITS Pilani K K Birla Goa Campus, India. [Slides]
17:15 - 17:30 – Open Discussion and Closing Remarks
Keynote: HPC Storage: Adapting to Change
Phil Carns (Argonne National Laboratory)
Abstract: Exascale computing has arrived, bringing with it a remarkable, and still growing, collection of data-intensive application use cases. This is an exciting opportunity for I/O researchers and practitioners, but it also increases pressure for us to keep pace and deliver on the promise of HPC technology. The community cannot afford to rest on its accomplishments; expanding access to new systems, problem domains, and user communities calls for us to adapt to new technologies, data management needs, and end-user expectations. This presentation will highlight illustrative examples in the field and discuss strategies for adapting to changes in HPC storage.
About the speaker: Phil Carns is a computer scientist in the Mathematics and Computer Science Division of Argonne National Laboratory. He is also an adjunct associate professor of electrical and computer engineering at Clemson University and a fellow of the Northwestern-Argonne Institute for Science and Engineering. He received his Ph.D. in computer engineering from Clemson University in 2005.
Phil has worked at Argonne since 2008, acting as technical lead, principal investigator, and developer for influential HPC research projects including Darshan (application I/O characterization), TOKIO (platform I/O characterization), Mochi (composable data services), PVFS (parallel file system), CODES (storage system simulation), and the Exascale Computing Project (data libraries and services for exascale platforms). Phil is a recipient of multiple R&D 100 awards. His research interests include HPC storage architectures, HPC system software, and I/O workload analysis.