Energy-Aware Ant Colony Based Workload Placement in Clouds
Eugen Feller, Louis Rilling, Christine Morin
With increasing numbers of energy hungry data centers energy conservation has now become a major design constraint. One traditional approach to conserve energy in virtualized data centers is to perform workload (i.e., VM) consolidation. Thereby, workload is packed on the least number of physical machines and over-provisioned resources are transitioned into a lower power state. However, most of the workload consolidation approaches applied until now are limited to a single resource (e.g., CPU) and rely on simple greedy algorithms such as First-Fit Decreasing (FFD), which perform resource-dissipative workload placement. Moreover, they are highly centralized and known to be hard to distribute. In this work, we model the workload consolidation problem as an instance of the multi-dimensional bin-packing (MDBP) problem and design a novel, nature-inspired workload consolidation algorithm based…