摘要
针对传统的面向终端的粗粒度流量控制方法易引起数据中心网络拥塞问题,将软件定义网络(SDN)引入大数据应用体系架构,提出了基于SDN的MapReduce调度优化方案。该方案中,SDN控制器可由多个MapReduce应用程序共享,并根据控制(master)节点的需求对网络进行拓扑管理和资源调度,降低了网络拥塞发生的概率,同时提升了MapReduce计算速度。
The traditional data transmission methods, which adjust the protocol parameters and transmission paths based on network state estimation, can result in the data center network con- gestion. The softwawre-defined network （SDN） is introduced to the big data application frame- work. A MapReduce scheduling optimization method based on SDN is proposed. Several MapRe- duce applications can share a SDN controller in the method. The SDN controller can allocate the resource and the manage physical connectivity based on requirements of master nodes. There- fore, the network congestion can be alleviated and the computation speed of MapReduce can be improved.
出处
《指挥信息系统与技术》
2015年第2期49-52,共4页
Command Information System and Technology
