Dryad and DryadLINQ: Academic Accelerators for Parallel Data Analysis
By Derick Campbell, Microsoft External Research
Releases such as the academic accelerators code named Dryad and DryadLINQ, currently available for free download, are great examples of what can be achieved when members of the global research community collaborate to develop technology. The result is availability of relevant tools that enhance discovery and tackle challenges. Working together, Dryad and DryadLINQ support quick and efficient parallel data analysis, a critical capability in today’s accelerated, data-driven research environment.
Geoffrey Fox, professor of informatics at Indiana University, is using the releases in his quest to leverage data gathered via radar. His goal: to learn more about the earth’s past and its present in order to make more informed, potentially life-saving predictions about its future. Using Dryad and DryadLINQ, Fox is currently analyzing radar data focused on glaciers in Greenland, where the ice sheets are melting more quickly than they have previously. “There is a hurry-up effort to gather data and make reliable predictions of the future of ice sheets, which have an impact on global climate,” he says. “The melting of the ice caps is established, but what isn’t established is how the melting correlates with the overall environmental situation, one part of which is the ice sheets.” For Fox, that’s where technology like Dryad and DryadLINQ come into play: they make it possible and relatively easy for researchers to conduct analysis that is separate but parallel, making it possible to fit various pieces of the puzzle together more quickly, efficiently and accurately than before.
Fox says he is confident that Dryad and DryadLINQ will prove themselves applicable to all kinds of discoveries, ranging from genetic research to developing a deeper understanding of fault lines. He is currently involved in discussions about undertaking research in response to last month’s crippling earthquake in Haiti, a region on which very little historical data has been captured, preserved or analyzed. “Radar data can be used to measure the stress on the earth, which in various ways, including predicting aftershocks,” he says. While he doesn’t foresee predicting actual earthquakes in the near future, the ability to do so could very well lie in the data that catalogs smaller movements along fault lines over time. And being able to analyze that data is an undertaking made less daunting by technology such as Dryad and DryadLINQ. “We have now high-performance ways to analyze data,” he says. “Software helps us do this in an efficient way, and the advances in computing make it cost effective, so while we’re doing lots of science, some of it has a social impact.”