异构云系统中基于智能优化算法的多维资源公平分配

资源分配策略的研究一直是云计算领域研究的热点和难点，针对异构云计算环境下多维资源的公平分配问题，结合基因算法（GA）和差分进化算法（DE），分别给出了两种兼顾分配公平性和效率的资源分配策略，改进了解矩阵表达式使异构云系统中的主资源公平分配（DRFH）模型转化成为整数线性规划（ILP）模型，并提出了基于最大任务数匹配值（MTM）的初始解产生机制和使不可行解转化为可行解的修正操作，以此提高算法的收敛速度，使其能够快速有效地得到最优分配方案。实验结果表明，基于GA和DE算法的多维资源公平分配策略可以得到近似最优解，在最大化最小主资源份额目标值和资源利用率方面明显优于Best-Fit DRFH和Distributed-DRFH，而且针对不同任务类型的资源需求，具有较强的自适应能力。

Fair allocation of multi-dimensional resources based on intelligent optimization algorithm in heterogeneous cloud environment

Resource allocation strategy has been a hot and difficult research topic in cloud computing field. In view of the fair distribution of multi-dimensional resources in heterogeneous cloud computing environment, two resource allocation strategies were proposed by combining Genetic Algorithm (GA) and Different Evolution (DE) algorithm and taking into account both fairness and efficiency in heterogeneous cloud environment. The solution matrix was improved to convert the Dominant Resource Fairness allocation in Heterogeneous systems (DRFH) model into Integer Linear Programming (ILP) model, a Max Task Match (MTM) based algorithm was used to generate initial solutions, and a revising operation was brought to change infeasible solutions into feasible solutions, which can accelerate the convergence to acquire the optimal solution quickly and effectively. Experimental results demonstrate that the multi-dimensional resources fair allocation strategies based on GA and DE algorithm can obtain near-optimal solutions; and in aspects of maximizing the value of minimum global dominant share and resource utilization, it is superior to Best-Fit DRFH and Distributed-DRFH, and has higher environmental adaptability to the resource requirement of different task types.