Tom Peterka, a computer scientist in Argonne’s Mathematics and Computer Science Division, will chair a U.S. Department of Energy (DOE) workshop on In Situ Data Management, to be held in North Bethesda, Maryland, Jan. 28–29, 2019.
Traditionally, data from large-scale simulations has been analyzed via postprocessing. As scientists produce increasing greater volumes of data, however, this approach is facing several problems: the available storage may be insufficient; the data transfer across the network may encounter bandwidth limitations; and the ability to adjust experiment parameters during simulation may be delayed. Researchers therefore have been exploring new techniques for in situ data management.
“Progress has certainly been made,” Peterka said. “But in situ data management on next-generation computing platforms will require advances in areas including programming and execution models, machine learning, provenance and reproducibility, and analysis across ensemble members.”
The two-day workshop will explore these topics and others, beginning with a plenary session on science applications and followed by six intensive breakout sessions. The objective is to identify future opportunities and capabilities needed for in situ data management and the associated research challenges that must be addressed in order to meet DOE mission needs.
The workshop is sponsored by the U.S. Department of Energy, Office of Advanced Scientific Computing Research. Registration is now closed. More details about the workshop can be found at https://www.orau.gov/insitudata2019/default.htm.