Over the past two decades the rise of information technologies (IT) has enabled businesses to communicate, coordinate, and cooperate in unprecedented ways. However, this did not come without a price. Today, IT infrastructure accounts for a substantial fraction of the national energy consumption in most advanced countries. Subsequently, research turned to finding ways of making IT more sustainable and lessening the environmental impact of IT infrastructure. In our previous work we developed LINFIX, an innovative method for handling the scheduling problem in data centers, which substantially reduced the total energy consumption compared to commonly used practices. Due to the computational complexity of the scheduling problem, we were, however, unable to estimate the cost reduction of LINFIX compared to what is theoretically possible. In this work we employ a genetic algorithm to provide a benchmark to better assess the quality of the LINFIX solutions. While the genetic algorithm frequently finds better solutions, the additional average cost reduction when compared to LINFIX is less than 0.1 percent. Taking the computational speed into account, this confirms our hypothesis that LINFIX provides very energy efficient scheduling plans in short time.
Brandt, Tobias and Bodenstein, Christian, "EVALUATING SCHEDULING METHODS FOR ENERGY COST REDUCTION IN A HETEROGENEOUS DATA CENTER ENVIRONMENT" (2012). ECIS 2012 Proceedings. Paper 103.