论文部分内容阅读
【目的】随着云计算平台越来越庞大并且呈现出动态化特点,对如何满足多个用户的计算请求而进行有效资源分配,同时保障整个网络的性能进行探索。【方法】首先,提出一种集群虚拟网络(CVN)策略和动态堆排序算法。其次,将联合多个pod构建跨pod集群虚拟网络(CP-CVN),同时保障网络的性能。进而将此网络优化问题公式化为线性规划问题(LP),并提出一种近似对偶算法解决此线性规划问题。【结果】理论分析表明所提出的对偶算法是可行的,实验结果验证对偶算法解决了软件定义网络中的资源分配问题,同时保障网络性能。【结论】理论和实验分析表明所提出的对偶算法是可行的。
[Objective] As the cloud computing platform becomes more and more large and dynamic, it is necessary to allocate resources effectively to meet the computing demands of multiple users while ensuring the performance of the entire network. 【Method】 Firstly, a cluster virtual network (CVN) strategy and dynamic heap sequencing algorithm are proposed. Second, the pod will be combined with multiple pod cluster virtual network (CP-CVN), while ensuring network performance. Then this network optimization problem is formulated as a linear programming problem (LP), and an approximate dual algorithm is proposed to solve this linear programming problem. 【Result】 Theoretical analysis shows that the proposed duality algorithm is feasible. The experimental results demonstrate that the dual algorithm solves the problem of resource allocation in software-defined networks while guaranteeing network performance. 【Conclusion】 Theoretical and experimental results show that the proposed dual algorithm is feasible.