Lavanya Ramakrishnan |
Lavanya Ramakrishnan has seen her grid computing focus shift from security to scientific applications over the past four years. After finishing her Masters degree at Indiana University in 2002, she continued her work in grid security at MCNC before moving to the newly-formed Renaissance Computing Institute in Chapel Hill, North Carolina, where she jumped feet-first into the world of scientific applications.
Ramakrishnan was responsible for last year's first deployment of the North Carolina Bioportal, and is now heavily involved with grid computing for SCOOP, the Southeastern Universities Research Association's Southeastern Coastal Ocean Observing and Prediction Program, and Linked Environments for Atmospheric Discovery. All these projects bring Ramakrishnan into close contact with application scientists, an interaction she enjoys.
"You realize early on that there's a vocabulary difference, but the partnership works really well once you understand each other a little bit--enough to talk each other's language," she explains.
In SCOOP, Ramakrishnan is working to assure quality of service for a specific storm surge model within the program. For LEAD, she researches dynamic and adaptive workflows. The goal of LEAD is to adapt meteorological resources - everything from radar orientations to data streams - with the weather itself, so workflows must be data-driven and adaptable.
"This is hard because it's not predicable," explains Ramakrishnan. "If you know what you have to run, it's easy. But the number of variables in LEAD is very big, and the user might change his mind, the weather might change, and the workflow and resources need to change to adapt to these conditions."
Ramakrishnan's LEAD work ties in nicely with another project - VGrADS, the Virtual Grid Application Development Software - in which she focuses on fault tolerance strategies for resource scheduling. In yet another project, a collaboration with Duke University called the Grid Resource Oversight Co-ordinator, she is part of a team that investigates using a leasing framework to provide access to Grid resources.
"Just like you lease houses for a certain period of time, we're looking at how can you lease resources the same way," she explains. "This helps resources providers because they can guarantee a higher level of service to each customer, and helps customers because they can make sure they have access when they need it."
As if five grid projects weren't enough, Ramakrishnan is also a part-time Ph.D. student at Indiana University. Having finished coursework and passed qualifying exams over the past two years, she's now delving into her dissertation research on adaptation in workflows in the context of applications like LEAD. She expects to change her part-time status to full-time sometime in the near future, when she takes a leave of absence to finish her doctorate.
When asked what she likes most about her work, Ramakrishnan remarks on the enjoyment she gets from getting to do research and see it deployed in the real grid environment.
"It's very satisfying to know that your research is really being used. And vice versa, that your research is informed by real experience."
And when asked what she finds most challenging, it's no surprise that she mentions one resource that she's having trouble scheduling.
"There's just too little time in the day," she says.
Learn more at the RENCI Web site.
—Katie Yurkewicz
e-mail this article
|
