Methodology and software for integrating high-performance computing resources into a distributed scientific data processing system managed by PanDA WMS

Seminars

Laboratory of Information Technologies


Date and Time: Wednesday, 23 June 2021, at 3:00 PM

Venue: Online seminar on Webex, Laboratory of Information Technologies




Report topic: “Methodology and software for integrating high-performance computing resources into a distributed scientific data processing system managed by PanDA WMS”

Speaker: Danila Oleynik

Abstract:

The talk will present the results of research and development carried out by the applicant within the BigPanDA project. A methodology for integrating high-performance computing (HPC) resources (supercomputers) into distributed high-throughput computing (HTC) systems for processing scientific data will be described. The implementation of this methodology will be shown as both a prototype and a full-scale software package in the frames of the PanDA workload management system (PanDA WMS). A set of recommendations for managing high-throughput data processing on HPC infrastructures are formulated as one of the research results. The possibility of increasing the efficiency of using leadership computing facilities is demonstrated.
(based on the PhD thesis)