This project focuses on the issues of network design, provisioning, and traffic engineering for dynamic backbone and metropolitan access networks needed to support a "follow the wind/sun" computing paradigm for green cloud computing. This paradigm is based on the observation that it is more energy efficient to move data and computations to a site with local green energy than it is to connect green energy sources to a power grid and then transmit the green power to distant data centers.
The project will study network design and traffic engineering when placement of the elements generating the traffic load is a possible degree of freedom. The sub-problems include 1) placement of data centers to both be close to green energy and aware of existing network deployment, 2) network capacity planning and rate capacity assignment for networks where data centers may be dynamically powered up/down based on energy availability, 3) routing of user service requests to data centers based on network considerations and estimates of green energy availability.
Broader Impact: This project will explore and quantify better use of green-renewable energy by data-centers, a significant consumer of electric power. The project will produce insights that assist industry in the future design, placement and operation of green data centers. The PIs will train undergraduate (via the REU program) and graduate students in the interdisciplinary area of energy management. As part of the training the PIs will hold highly-interactive weekly meetings in which students actively learn from each other.