Despite HPC is the base of many research projects involving numerical simulations, generally its scope within education of the applied disciplines needs improvement. We bring together I/O experts from data centres, users and application workflows to share current practices for parallel I/O and storage with application to scientific workflows, issues, and obstacles for both hardware and the software stack, and R&D to overcome these issues. The meeting series fills this gap by providing a forum to experts, students and the interested public with presentations from external guests about HPC storage systems. This also facilitates discussions with the goal to train essential skills for research in HPC. The event takes place every month both during winter and summer semesters regardless of lecture free time.

Our regular meetings typically revolve around a selected topic, have at least one invited speaker, provides opportunity for discussion (>= 15 min) regarding this topic and an open discussion at the end. We decide about some upcoming topic and an initial speaker.

It is supported by the NHR and GWDG. Powered by VI4IO, the HPC IODC workshop, and the MCSE, DECICE and KISSKI projects.

Attendance is free of charge.

We send the regular invitation to the Mailing list.

Please note that we plan to record special sessions with the intent of providing the recordings via BBB to other students but also to publish and link the recordings on YouTube for future terms. If you appear in any of the recordings via voice, camera or screen share, we need your consent to publish the recordings. See also this Slide.

• 2024-01-10 - Informal Meeting about the monthly storage talks
  • Goal is to identify the next topics, interest in this area…
  • Welcome – Patrick Höhn, Julian Kunkel slides
• 2024-02-07 - Best Practices in Organizing I/O – Moderator: Patrick Höhn
  • Best Practices in Organizing I/O for ML Projects – Ali Doost Hosseini slides
• 2024-03-06 - Using GPU Direct Storage – Moderator: Patrick Höhn
  • Using GPU Direct Storage on the different HPC storage systems at GWDG - performance possibilities and challenges - Sebastian Krey slides
  • GPU Direct in the IO500 - Julian Kunkel slides
• 2024-04-03 - Metrics for performance and procurement – Moderator: Sebastian Oeste
  • Metrics for performance and procurement – Sebastian Oeste slides
  • Procument and performance experience at GWDG – Christian Köhler slides
• 2024-05-08 - Staging tools – Moderator: Patrick Höhn
  • BeeGFS Approach – Philipp Falk (ThinkParQ)
• 2024-06-05 - File systems – experience from operation
  • Vast, Ceph, NextCloud, NFS, BeeGFS, Lustre, Spectrum Scale, NFS, WekaFS, GeckoFS, PNFS
  • Everyone brings their experience - 1 slide per FS would be great (pros/cons)
• 2024-07-10 - Scalable Storage Competetion – Moderator: Patrick Höhn
  • Scalable Storage Competetion – Marcus Merz (GWDG)
• 2024-08 - Vacation
• 2024-09-18 - Storage Semantics – Moderator: Patrick Höhn
  • Memory Centric Storage for Exascale (MCSE) – Patrick Höhn
• 2024-10-02 - High performance all-flash filesystem using xiRAID, Lustre and commodity hardware – Moderator: Patrick Höhn
  • High performance all-flash filesystem using xiRAID, Lustre and commodity hardware – Davide Villa (Xinnor), Markus Hilger (MEGWARE)
• 2024-11-06 - Modeling and Emulating Parallel Filesystems for Research and Benchmarking – Moderator: Patrick Höhn
  • Modeling and Emulating Parallel Filesystems for Research and Benchmarking – Philipp Martin (RWTH Aachen)
• 2024-12-11 - Utilizing Transfer Learning for Solving the Puzzle of Understanding I/O Performance – Moderator: Patrick Höhn
  • Utilizing Transfer Learning for Solving the Puzzle of Understanding I/O Performance – Radita Liem (RWTH Aachen)