数据中心网络拓扑感知型拥塞控制算法.docxVIP

数据中心网络拓扑感知型拥塞控制算法 　　摘要：　　针对数据中心网络的链路拥塞问题，提出了一种拓扑感知型拥塞控制算法。首先，根据广义超立方体拓扑多维正交和单维全连接的结构特点，结合网络流的最大流最小割定理，提出了拓扑感知地选取分布流量请求的不相交路径策略；然后，根据带宽需求自适应选取不相交路径；最后，利用已选取路径的剩余带宽为权重动态调整每条路径的流量分配比例，从而达到缓解网络链路拥塞、均衡网络负载和减轻目的节点侧数据重组压力的目的。实验结果表明，与链路关键性路由算法、多路径健忘路由算法、最小割多路径路由算法和免拥塞路由策略相比，TACC算法在均衡链路负载和优化算法部署时间方面有良好的表现。　　关键词：　　数据中心网络；拥塞控制；拓扑感知；最大流最小割定理；自适应路由；动态权重分配　　中图分类号：　　TP393　　文献标志码：A　　Abstract：　　To solve the congestion problem of links in Data Center Network ， a TopologyAware Congestion Control algorithm was proposed in data center networks. According to the properties of multidimensional orthogonality and singledimensional full mesh in the generalized hypercube， a topologyaware strategy was put forward to find the disjoint routes of distributing the request of flow by the maxflow mincut theorem. Then， the number of disjoint routes was adjusted adaptively for satisfying the bandwidth requirement. Finally， the weighted value of selected routes was utilized to adjust dynamically the distributing rate of per routes flow. The algorithm can relieve the network congestion， balance the load of links and reduce the pressure of recombining data in the destination. The experiments show that the load of links are balanced perfectly and the deployment of the algorithm is realized easily in comparison with Link Criticality Routing Algorithm ， Multipath Oblivious Routing Algorithm ， MinCut MultiPath and CongestionFree Routing Strategy .　　To solve the congestion problem of links in Data Center Network ， a TopologyAware Congestion Control algorithm was proposed in data center networks. According to the properties of multidimensional orthogonality and singledimensional full mesh in the generalized hypercube， a topologyaware strategy was put forward to find the disjoint routes of distributing the request of flow by the maxflow mincut theorem. Then， the number of disjoint routes were adjusted adaptively for satisfying the bandwidth requirement. Finally， the residual bandwith of the selecte