Indiana University's Jetstream
02 Jun, 2020

The NSF-funded Jetstream system offers cloud-based, on-demand computing and data analysis resources to researchers in a number of fields. Photo by Emily Sterneman, Indiana University. Courtesy Indiana University.

National Science Foundation $10M Award to 'Jetstream 2' Brings New Opportunities for CyVerse

The cloud infrastructure for science will enhance and expand Jetstream's role in data science training and research with more computational capacity and GPU support for analyses and workflows.

Indiana University (IU) announced this week the receipt of a $10M grant from the National Science Foundation (NSF) to deploy Jetstream 2, a distributed cloud computing system to support on-demand research in a range of fields, such as artificial intelligence (AI), social and life sciences, and enhanced large-scale data analyses for the nation, including COVID-19 research.

The award presents exciting opportunities for CyVerse, as Jetstream 2 provides new options for cloud-based data science training, GPU resources and support for machine learning workflows, and opportunities to enhance security for research datasets.

"A primary goal of CyVerse is to make complex leading edge technologies readily usable by scientists," said Nirav Merchant, co-principal investigator of CyVerse and Director of the Data Science Institute at the University of Arizona (UArizona), and one of the investigators of the Jetstream 2 award. "Jetstream 2 represents a big step in that direction."

Jetstream 2's signature innovation is its ability to make high performance computing and software easy to use by researchers who have limited experience with supercomputing and cloud systems. This is especially helpful for smaller academic communities with little previous access to such resources.

"Jetstream 2 builds on the tremendous success of the original Jetstream system at IU and with our partners," said Brad Wheeler, IU vice president for Information Technology. "It bundles computation, software, and access to storage for individuals and teams of researchers who span hundreds of areas of research across the nation and at the frontiers of scientific inquiry."

Jetstream 2 will give projects such as The Carpentries, which frequently partners with CyVerse to teach foundational coding and data science skills to researchers worldwide, a training platform that is easily and broadly accessible for students and educators alike.

The system will offer CyVerse AI users especially new capabilities to improve and scale their analyses, including GPU and CPU resources and support for machine learning workflows.

Additionally, a portion of the physical infrastructure for Jetstream 2 will be housed at Arizona State University (ASU), which brings Jetstream partners at ASU and UArizona closer toward their goal of developing a regional computing cloud for the state.

And, Merchant added, UArizona team members will explore possibilities for extending HIPAA-compliant secure computing from CyVerse onto Jetstream as a means of offering expanded security for researchers who require it for their analyses.

The current Jetstream system was funded and led by IU in 2014 as the NSF's first production science and engineering research cloud system for the nation, offering cloud-based, on-demand computing and data analysis resources. Jetstream is a deployment of CyVerse's Atmosphere cloud computing platform within XSEDE, the national Extreme Science and Engineering Discovery Environment.

"The initial vision of Atmosphere was to make accessing the cloud as convenient as possible for scientific users," said Edwin Skidmore, director of infrastructure at CyVerse. "Atmosphere was easily adapted for Jetstream's purpose to serve science and engineering communities."

Over the years, the Jetstream system has given thousands of U.S. researchers access to a powerful cloud-based environment that complements other NSF systems – all from a laptop or iPad – allowing them to explore and understand immense amounts of data.

Jetstream 2 builds on the classroom success of Jetstream, which was used in classes to teach computational biology and chemistry, and in student projects on AI approaches to biological field research, veterinary medicine, and textual analysis.

The project team has a goal of serving more students than any other NSF-funded cyberinfrastructure resource, leading to a diverse pool of graduates entering the STEM workforce with robust training in computational science.

Jetstream 2 is funded under NSF award number 2005506.