基于队列理论的云资源分配收益最大化算法  

PROFIT MAXIMIZATION ALGORITHM FOR CLOUD RESOURCE ALLOCATION BASED ON QUEUE THEORY

作　　者：郑宇超[1] 夏学文[1] 艾冬梅[2] 

机构地区：[1]华东交通大学软件学院,江西南昌330013 [2]北京科技大学数理学院,北京100083

出　　处：《计算机应用与软件》2017年第11期252-257,共6页Computer Applications and Software

基　　金：江西省教育厅科研项目(GJJ150539)

摘　　要：为了优化资源分配收益,提出基于时间服务因子TSF资源定价机制下的资源分配算法。首先,算法将资源分配问题形式化为队列模型,并构建了收益最大化函数。然后,通过Lagrange乘子法求解最优化函数的解。利用时间服务因子,算法分别通过安全满意度因子ASF和响应满意度因子RSF定义了资源的定价方式,并同步考虑定价、请求到达率、资源服务率及可用资源量,得到了使收益最大化的资源分配方式。实验结果表明,与传统的启发式资源分配算法比较,该算法得到的收益更高,且尤其在云资源数量较稀少的场景下,算法将更加具有优势。For optimizing the profit of resource allocation, a resource allocation algorithm in resource pricing mechanism based on time service factor is proposed. First, the resource allocation problem was formalized as the queue model, and the profit maximization function was constructed. Then, the profit maximization function was solved by Lagrange muhiplier method. Using the time service factor, our algorithm defined the pricing mode respectively by the assurance satisfaction factor and responded satisfaction factor. In addition, resource allocation was maximized by considering pricing, request arrival rates, resource service rates, and available resources. Experimental results show that compared with the traditional heuristic allocation algorithm, our algorithm yields higher returns, especially in the context of the scarcity of cloud resources. The proposed algorithm will have more advantages.

关 键 词：队列理论 资源分配 云计算 收益最大化 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]