UW Information Technology

Defining a New Sky

Andrew Connolly

Astronomy Professor Andrew Connolly relies on the UW’s high-performance network to handle massive data sets.

If you tried to describe the southern half of the observable universe in zeroes and ones, how much data would you pile up?

UW Astronomy Professor Andrew Connolly and his colleagues plan to find out. They’ll use the eight-meter Large Synoptic Survey Telescope (LSST) being built in Chile, with a 3.2 billion-pixel camera, to survey the sky.

“Where you see one object with the naked eye, we’ll see 10 million galaxies,” Connolly said. “Within a year from the start of the survey, we’ll have collected more information about the sky than in the entire history of astronomy.”

So much data will be collected, in fact, that the LSST project describes its goal as “defining a new sky.” Helping researchers transfer, store, share and access the overwhelming immensity of that data is UW-IT’s job.

“Fundamentally, it comes down to plumbing,” said Clare Donahue, UW-IT’s Associate Vice President for Networks, Data Centers & Telecommunications. “You need big pipes. And you need them properly maintained, always available and secure. It’s all part of the core infrastructure that UW-IT provides.”

To help prepare for this flood of data, UW’s infrastructure is on the verge of a major upgrade, thanks in part to a grant from the National Science Foundation. By 2015 UW-IT plans to complete installation of a 40G “big pipe” to serve as the UW’s data backbone. That’s four times the capacity of UW’s existing 10G network.

As part of this ambitious upgrade, UW-IT is also creating a 100G High Speed Research Network (HSRN) spanning the UW data centers and connecting to the outside research Internet through the Pacific Northwest Gigapop and Internet2. The combined backbone design and HSRN will offer dedicated data paths for researchers, including enhancements to UW’s “Science DMZ,” a network design promoted and partially funded by federal agencies that allows collaboration with peers at other institutions without being slowed down by firewalls. This, when used with large-scale storage and compute systems at UW like Hyak and lolo, will help support big data science projects.

“If you don’t have the underlying pipes, dealing with all this data can take weeks or even months,” Connolly said. “With a thousand times more data than today’s surveys, waiting a month to analyze it changes the sort of questions you’re going to ask and changes the exploratory nature of science.”

It’s not only astronomy that needs this capacity, Donahue said. Other UW researchers are working on decoding the human genome, unraveling the mysteries of ocean currents and tectonic plates, and unearthing clues to discover core principles at the interface of mathematics, biology, and medicine, and they need it too.

In the end, many people will benefit from discoveries that emerge from the masses of data generated by UW researchers and supported by UW-IT, Donahue said. It might come in the form of previously undreamed-of cures for diseases or ways to prolong life, early warning of devastating tsunamis or fundamentally new understandings of the nature of the universe.

“UW-IT is a good partner, open to thinking about providing resources we’ll need five years down the road,” Connolly said. “Having this infrastructure helps UW get the best scientists, the best researchers and the best students.”

Originally published in the UW Information Technology 2013 Annual Report. View our Annual Reports archive.