6/20/2012 Elise King, Coordinated Science Lab, and Max Tane, ECE ILLINOIS
Written by Elise King, Coordinated Science Lab, and Max Tane, ECE ILLINOIS
Dynamic scalability—the ability of a system to accommodate changing workload without interrupting service—is crucial for cloud computing to be widely adopted. The enormous size of a cloud requires low-complexity algorithms, but if not designed well, these algorithms can cause significant performance degradation as the system grows. Also, existing algorithms often do not scale because of their centralized, high-complexity nature.
ECE Assistant Professor Yi Lu recently received a CAREER Award, given by the National Science Foundation (NSF), to tackle this problem by designing low-complexity algorithms for Web services, data management, and measurement and monitoring in the cloud. The program awards junior faculty members who demonstrate their roles through outstanding research and education.
“It is my first NSF award,” said Lu, a researcher in the Coordinated Science Lab. “I was happy to receive it.”
The algorithms Lu designs will address challenges with dynamic scaling, multi-tenancy and data-intensiveness in the cloud, and challenges with different application workloads, including search, social networks, and map-reduce. The grant, “Scheduling and Resource Allocation in the Cloud Using Graphical Models and Randomized Algorithms,” is for five years.
“We plan to approach these problems with new techniques in the area of randomized algorithms and graphical models,” said Lu. “We already had some very nice preliminary results, which won the Best Paper Award at the 2011 Performance Conference.”
Lu hopes to design and implement algorithms that deliver high-quality cloud services while keeping the cloud infrastructure energy-efficient and scalable. She is also interested in discovering new theories for the new and evolving infrastructure.
Lu is taking her research to the classroom. In the spring semester she taught a graduate-level course, ECE 598YL: Cloud Computing Infrastructures.
“I enjoy the interaction with students the most,” said Lu. “It is great to see an idea getting across. Students are involved at all stages of research, from formulating a problem, trying out different ideas, identifying a solution, and understanding the implications.”