A Proportional Multi-Resource Scheduling Model in SDCloud

with the growing adoption of virtualized data centers and cloud services, nowadays multiple resource scheduling is increasingly attractive to researchers. Some previous studies achieved progresses in this area. However, these heuristics have obvious limitations in complex software defined cloud environment. A real multi-dimensional model is needed to solve this NP hard problem. Our approach emphasizes energy conservation and load balancing. This model is designed to evaluate the value of a specific resource in the multi-resource scenarios. The corresponding formulation and key points of consecutive algorithm (S-Aware) is discussed in this paper. We innovated upon the multidimensional bin packing approach and all dimensional spaces are consumed in a proportional way. Experiments demonstrate that our algorithm outperforms others both in energy conservation and load balancing.