NMC Scientist John Bent Presents on TASSI – A New Method of HPC Storage Systems

NMC Scientist John Bent Presents on TASSI – A New Method of HPC Storage Systems

NMC Scientist John Bent Presents on TASSI – A New Method of HPC Storage Systems

New Mexico Consortium scientist, John Bent, recently presented his work on a new method of HPC storage systems at the Lustre User Group (LUG) Conference held May 7-8, 2024 at Texas Tech University in Lubbock, Texas.

The LUG Conference is the main forum for discussing the advantages and exciting changes coming to the open-source Lustre file system and related technologies in High Performance Computer Industry.

Bent’s presentation was titled,TASSI and Automated AI-Analysis of the Lustre-Development Mailing List.

In his presentation, Bent discusses the current challenges with HPS Storage Systems. Ideally, research solves questions of practical importance such that publication helps solve real-world problems faced by production sites. Unfortunately, finding realistic real-world problems can be quite challenging. Typically, researchers read ‘Future Work’ sections in other academic publications which creates a potential waste of effort if impractical research follows previous impractical research. Alternatively, researchers find projects anecdotally through conversations with various production practitioners. Unfortunately, this might result in working on real-world problems which are only narrowly relevant.

In this project, they describe a third method and share early results. Specifically, Bent and colleagues have used the OpenAI API to automatically analyze the full historical archive of the lustre-devel mailing list to help identify which problems are most broadly relevant. In addition to sharing some preliminary results, they also share other insights. The presentation concludes with a brief overview of their recent TASSI project – a project directly motivated by this analysis.

For  the slides to John’s presentation and learn more see: https://www.depts.ttu.edu/hpcc/events/LUG24/slides/Day2/LUG_2024_Talk_09-TASSI_John_Bent_LUG24.pdf