From its beginning, the Globus vision was to make research easier for scientists, taking on the burden of the annoying and tedious data tasks to make more time for the actual science. Gradually, the service has increased its toolbox from transfer and sharing to analysis and discovery, tapping the potential of the cloud to handle the tricky business of research data management. At Globusworld 2017, the spotlight was appropriately on “labs” -- both the internal working on new features and the external using Globus services to boost their science.
The seventh installment of the Globus annual meeting again brought researchers, IT professionals, and computer scientists to a downtown Chicago hotel for two days of tutorials and talks. Co-founder and project director Steve Tuecke’s keynote, titled “Beyond Big Data,” emphasized that the data features of Globus are just the foundation, not the finish line of their work. Besides announcing Google Drive integration, user interface changes, beefed-up search capabilities, and other new or coming-soon features, Tuecke also connected these and older features to the scientific projects they enable.
“Our mission is not data transfer,” Tuecke said. “It’s data management throughout the research lifecycle.”
Globus co-founder Steve Tuecke updates the Globusworld 2017 crowd on recent usage statistics.
The power of those capabilities was made clear by several lightning talks from Globus users, representing a variety of scientific areas. Andrew Gustafson from the University of Minnesota talked about how Globus services underpin a new platform for sharing data on agroinformatics -- which uses field data and simulation to predict crop yields, climate performance, pest resistance, and other important agricultural factors. Globus makes it easier for academic and corporate partners to collaborate on research results, Gustafson said, but also makes it possible to limit sharing in cases where businesses don’t want to give data to competitors.
Other speakers talked about Globus features such as user authentication and identity management on massive national supercomputing infrastructure such as Compute Canada and XSEDE. Representatives from the Universities of Michigan and Minnesota described the incorporation of Globus features into research computing centers, simplifying the movement of data to HPC resources or backup storage. Presenters from Oak Ridge and Lawrence Berkeley National Laboratories talked about integrating Globus into science DMZs and research data portals.
Later, a session from Globus Labs provided early glimpses of some fresh ideas that could be the subject of future outside testimonials. A new testing ground for students, postdocs and researchers to develop ideas around new tools for data-driven discovery, Globus Labs encompasses the internal and collaborative projects that look for the next step after data access, Globus co-founder and CI Senior Fellow Ian Foster said.
“We’re always looking forward, trying to probe the boundaries and develop new ideas,” Foster said. “If the mission of Google is to make all the world’s data accessible, our mission is to make all the world’s research data reliably and securely accessible.”
CI Fellow and Research Specialist Kyle Chard discusses new approaches to data discovery at Globusworld 2017.
Early results on display at Globusworld included a Netflix-like “recommendation” algorithm to help scientists find published datasets that might be useful for their research, new approaches to extract metadata features from unstructured data to enable discovery, or responsive storage that automatically kicks into action when needed, such as when a telescope starts collecting and transmitting images. Other new services are already hard at work, such as the tools behind the Materials Data Facility, an effort to create a unified repository for materials science research results that has already taken in terabytes of data.
For more on Globus services, visit globus.org, and for more slides from Globusworld 2017 talks, visit https://www.globusworld.org/program.