A Bio-inspired Algorithm for Energy Optimization in a Self-organizing Data Center

The dimension of modern distributed systems is growing everyday. This phenomenon has generated several management problems due to the increase in complexity and in the needs of energy. Self-organizing architectures showed to be able to deal with this complexity by making global system features emerge without central control or the need of excessive computational power. Up to now research has been mainly focusing on identifying self-* techniques that operate during the achievement of the regular functional goals of software. Little effort, however, has been put on finding effective methods for energy usage optimization. Our work focuses specifically on this aspect and proposes a bio-inspired self-organization algorithm to redistribute load among servers in data centers. We show, first, how the algorithm redistributes the load, thus allowing a better energy management by turning off servers, and, second, how it may be integrated in a self-organizing architecture. The approach naturally complements existing self-management capabilities of a distributed self-organizing architecture, and provides a solution that is able to work even for very large systems.