Heterogeneity aware dominant resource assistant heuristics for virtual machine consolidation

Power consumption is a critically important issue for data centers. Virtual machine (VM) consolidation is fundamentally employed to improve resource utilization and power optimization in modern data centers. In general, VM consolidation is formulated as a vector bin packing problem, which is a well-known NP-hard problem. Hence, heuristic algorithms, such as single dimensional heuristics (e.g., first fit decreasing (FFD)) and dimension-aware heuristics (e.g., DotProduct), are usually deployed in practice for VM consolidation. However, all of these previous heuristic algorithms did not sufficiently explore the heterogeneity of the VMs' resource requirements. In this paper, we propose several heterogeneity aware dominant resource assistant heuristic algorithms for VM consolidation. The performance evaluations validate the effects of the proposed heterogeneity aware heuristics on VM consolidation. The proposed heuristics can achieve quite similar consolidation performance as dimension-aware heuristics with almost the same computational cost as those of the single dimensional heuristics.