I/O and Data Seminar Series: HPC Data Management at LANL
Title of Presentation: “HPC Data Management at LANL”
Presented by: Brad Settlemyer, Los Alamos National Laboratory
The massive storage capacity available on modern massively parallel HPC platforms has led to a series of challenges related to finding and managing large data sets. Storage systems that, by necessity, incorporate multiple media types, scientists searching for needles in haystacks, and emerging storage systems software have led to a series of disruptive changes in how storage systems are architected. In this talk Brad Settlemyer will describe challenges and solutions related to ensuring that file systems with tens of thousands of disks are reliable, efficiently searching and sifting through trillions of files, how fast persistent media types (including Optane DC memory) affect scientific simulation, and how emerging AI workloads may influence storage system design.
Speaker’s Bio: Brad Settlemyer is currently a Senior Scientist in Los Alamos National Laboratory’s HPC Design group. He received his Ph.D in computer engineering from Clemson University in 2009 and works as a researcher in the field of high-performance computing storage systems. He currently leads the storage systems research efforts within Los Alamos’ Ultrascale Research Center and his team is responsible for designing and deploying state of the art storage systems for enabling scientific discovery. He is the Primary Investigator on projects ranging from ephemeral file system design to long-term molecular information storage. He has published papers on emerging storage systems, long distance data movement, network and storage modeling, and storage system algorithms.