Team members exchange information on data with goal of benefitting users
HPC staff members from seven national laboratories came together last month to consider various ways of treating one of today’s key topics: data.
The 3-day workshop—the Joint Facilities User Forum on Data-Intensive Computing—brought together users and HPC center staff to discuss approaches to handling data, best practices, and the future of data-driven scientific discovery.
The co-organizers of the Oakland, California, event were the OLCF, National Energy Research Scientific Computing Center (NERSC), Argonne Leadership Computing Facility (ALCF), Sandia National Laboratories, Lawrence Livermore National Laboratory, and Los Alamos National Laboratory. Day one of the forum focused on advances in managing, analyzing, and visualizing data. The second day featured a discussion of successes, failures, best practices, and lessons learned. The topic for the third day was working with data and practical ways to get things done.
Fernanda Foertter, an HPC user assistance specialist, chaired one of the sessions on the first day of the forum.
“This is a continuation of what we started at Oak Ridge in 2013 with the workshop on analysis of very large datasets. NERSC staff attended and expressed interest in producing a joint meeting the following year. Before we knew it, six laboratories were interested in participating,” she said. “It was a case where labs came together very fluidly to organize a user-centric workshop. We enjoyed working with each other very much in the planning stages.”
Chris Fuson, who gave a talk on transferring data into and out of the OLCF, said the small group sessions provided a great opportunity for the exchange of information. Fuson’s presentation featured tools available to OLCF users for transferring large datasets.
“We continue to provide users with the ability to create increasing amounts of data,” he said. “It’s good that we are discussing data management and transfer with users and other facilities.”
Several other OLCF staff members took part in the workshop. Ashley Barker, group leader of User Assistance and Outreach, was a session chair during the second day’s meetings on best practices, and ORNL‘s Rangan Sukumar delivered a presentation entitled “Machine learning for data-driven discovery.” Other ORNL participants included Stuart Campbell, Clay England, Doug Fuller, John Harney, and Norbert Podhorszki.
Fuson said the opportunity to share best practices with other centers was extremely helpful.
“It is interesting to hear how the other centers do things, and it was a good experience to talk about these topics,” he said. “Our goal is to help our users benefit from this collaboration.”