Extreme Data Workshop 2018

Workshop, 18.-19.9.2018, Jülich Supercomputing Centre, Forschungszentrum Jülich

Exascale computing will require exascale data handling. Earth system sciences and other application areas (e.g. astrophysics, genomics, photon science) are facing rapidly growing challenges due to the fact that observation systems and numerical models are producing data of unprecedented volumes and at unprecedented rates. Weather and climate models must be further developed to resolve the small scales that are relevant for extreme events and the assessment of their impacts. New telescopes, medical imagers, satellites, and other instruments produce data in finer resolutions, at higher frequencies, with increasing numbers of channels. Scientific progress and the development of services (e.g. environmental prediction) hinge on the ability of the next-generation compute and data infrastructures to cope with these data streams. Present workflows often decouple data generation, data processing and services, and data archiving and curation. The anticipated dimension of extreme data will require development of new hardware technology, new workflows and service architectures, and a better integration across various storage systems. The workshop aims at identifying common demands across research fields and at developing prospects how future extreme data systems may unfold.

The Extreme Data workshop is organized in three half-day sessions:

• Extreme data demands
• Extreme data technology
• Extreme data services

Each session will consist of 2-3 invited and up to 4 contributed presentations taking stock of the currently anticipated developments and identifying the major challenges on the way to exascale data handling. Each block of presentations will be followed by a half-hour discussion session, and a final discussion will take place at the end of the workshop. A proceedings volume will be published after the workshop.

The workshop shall focus primarily on HPC-related extreme data. However, topics such as data integration and sharing between HPC and cloud systems will also be covered. Big data analytics concepts which can help managing exascale data, are also in scope. With regard to earth system data demands, technologies, and services, this event complements a workshop on extreme HPC to be held at ECMWF (Reading, UK) also in September 2018.