基于CSP的能耗高效云计算资源调度模型与算法

CSP-based resource allocation model and algorithms for energy-efficient cloud computing

利用约束满足问题对异构云数据中心的能耗优化资源调度问题建模,通过求解建立的约束模型可以获得能耗最优的资源分配方式,并在此基础上提出了能耗优化的资源分配算法dynamicpower(DY)。与已有的算法MinPM、FFD、BFD相比,算法DY考虑了资源的异构性,能够降低云数据中心物理服务器的能耗。最后,利用Choco实现了提出的算法DY,并将DY与MinPM、FFD、BFD进行实验比较,实验结果表明,提出的算法在能耗上有明显优势。

The resource allocation for energy-efficient cloud computing in a heterogeneous environment was modeled as a constraint satisfaction problem （CSP）. By solving the constraint satisfaction problem, the optimized allocation scheme minimizing energy consumption in virtualized cloud data centers was also obtained. Based on the optimized allocation i scheme, an energy-efficient resource allocation algorithm, dynamicpower （DY）, which takes into account the heterogene- ity of resources, was proposed. The performance of algorithm was evaluated using Choco. Experimental results show that, compared with first-fit decreasing （FFD）, best-fit decreasing （BFD） and minimizing the number of physical machines （MinPM）, the proposed algorithm （DY） has less energy consumption.