Finding the right balance of cloud computing resources

Researchers in the School of Management have developed a solution to manage cloud computing resources more cost-effectively.

Forthcoming in Information Systems Research, the study develops strategies for virtual resource management in cloud data centers by integrating concepts and methods from computer science, machine learning, operations management and statistics into algorithms that can be applied to a number of different virtual computing environments.

“Our findings are most relevant to infrastructure-as-a-service providers, who provide the physical hardware like servers and networking equipment for cloud computing, but they can easily be extended to platform-as-a-service environments as well,” says Ram Ramesh, professor of management science and systems.

To validate the performance of their model, researchers developed use cases based on actual pricing and service data from Amazon Elastic Compute Cloud (Amazon EC2). Based on these cases, their framework can achieve up to 30% cost savings compared to the traditional resource management model.

The researchers say their findings will have the greatest impact for cloud service contract administrators, managers for cloud computing products and services, and engineers for cloud computing technology.

“Cloud resource management is a complex issue, with multiple performance metrics to consider,” Ramesh says. “Administrators in cloud data centers can use these algorithms to develop efficient service-level agreements and cost-effectively manage the computing resources involved in those agreements.”

Ramesh collaborated on the study with Zhiling Guo, associate professor of information systems at Singapore Management University, and Jin Li, assistant professor at Xidian University School of Economics and Management.