The most efficient version of the high-performance Energy Sciences Network, or ESnet, was unveiled Tuesday at Lawrence Berkeley National Laboratory.
ESnet is a high-performance network managed by Berkeley Lab and funded by the U.S. Department of Energy, or DOE, according to its website. Its newest version, ESnet6, allows scientists to exchange data sets across the country at 100–400 gigabits per second.
“ESnet operates as the central circulatory system connecting all of our National Labs and DOE’s nationwide facilities rapidly moving data between these sites so that scientists can reach instruments, computers, experiments and people around the world seamlessly,” said Carol Burns, deputy director for Research at Berkeley Lab, at the unveiling event.
David McCallen, director of the Center for Civil Engineering Earthquake Research at the University of Nevada, Reno, explained that ESnet6 has allowed him to “miraculously” exchange large amounts of data at great speed with other scientists across the country.
McCallen characterized ESnet as a fast data pipeline, using optical fibers to transport data from one site to another. With ESnet6, McCallen can transport more than 9 billion data sets from Tennessee to Berkeley in just 10 minutes. Previously, this data transfer would have taken hours or days to complete.
“The process of science is changing,” said Inder Monga, executive director of ESnet and division director of scientific networking at Berkeley Lab. “There are big collaborations across different geographies. They use instruments and collect data, so if you want to tackle the problem of climate or environment, you need data globally to understand how the weather is changing, how the weather patterns are changing, how the air and the currents are changing.”
Monga said ESnet’s mission is to “remove boundaries” between scientists to help them more easily collaborate across long distances. He added they aim to have a network “reliable and resilient” to network failures or cyberattacks.
Monga said the data will make fewer stops before reaching its intended destination, allowing ESnet6 to provide better speeds and data throughput compared to previous versions.
“What we have been seeing so far is that this upgrade is to handle what we call this data tsunami that is coming at us,” Monga said.
McCallen noted ESnet6 would be “transformational” to earthquake risk and hazard research when paired with DOE supercomputers. He said the two allow for earthquake simulations across an entire region, showing, for example, how an earthquake would affect Bay Area communities.
The use of increasingly large amounts of data in science has kept the network in a constant state of development, McCallen noted.
“The demand and the pull for ESnet6 is the fact that we are using models of ever higher fidelity,” McCallen said. “There will always be a development cycle for improvement of our data transfer.”