Abstract
Energy efficiency is a major topic in every scientific field, since being energy efficient means producing more for a smaller cost. Data centers are no exception to this rule as their energy use represents a large portion of the global consumption, and it is needless to say that they ought to perform optimally while being eco-friendly in order to preserve natural resources as much as possible while providing a high quality service for the users. In this paper, we propose an efficient algorithm for allocating users to a pool of servers in an energy-efficient way. Our allocation model emphasizes the critical importance of nondominant resource types such as memory, which usually tend to be wasted by homogeneous allocation approaches. We show that the performance of the algorithm makes it worthy of being used in real-time environments where split-second decisions must be made. We compare our algorithm to the most well-known metaheuristics used in operations research and we show that they do not provide a significant improvement in a reasonable time.