数据中心中路由编码的可行性研究

发布时间：2018-06-01 01:10

本文选题：数据中心 + 物理网络　；参考：《电子学报》2017年11期

【摘要】：修复带宽最优并不代表修复通信量也是最优的,后者与物理网络拓扑有着密切联系.本文基于路由编码的思想减少修复通信量.首先,基于信息流图对物理网络中数据的传递过程进行建模,证明得出了满足路由编码可行的充要条件,并发现路由编码可以基于再生码实现.然后,针对数据中心网络设计的特点,为Fat-tree设计了一个工作在应用层的协议来生成修复树,为CamCube设计了一个启发式算法来生成修复树.关于最小存储再生码的数据修复过程的仿真实验表明,路由编码可以有效地降低修复通信量,2种修复树生成方案在各自适合的网络中均有较好性能:在帮助节点数较小时,Fat-tree和CamCube中的修复通信量分别降低了大约50%和30%.
[Abstract]:The optimal repair bandwidth does not mean that the repair traffic is also optimal. The latter is closely related to the physical network topology. This paper reduces the amount of repair traffic based on the idea of routing coding. Firstly, the data transfer process in the physical network is modeled based on the information flow graph, and the necessary and sufficient conditions for the routing coding are proved to be feasible, and it is found that the routing coding can be realized on the basis of the regenerative code. Then, according to the characteristics of data center network design, a protocol working in the application layer is designed for Fat-tree to generate repair tree, and a heuristic algorithm is designed for CamCube to generate repair tree. The simulation results on the data repair process of the minimum storage regenerative code show that, Routing coding can effectively reduce the repair traffic and the two repair tree generation schemes have better performance in each suitable network: the repair traffic in Fat-tree and CamCube is reduced by about 50% and 30% respectively in the small number of help nodes.
【作者单位】：空军工程大学防空反导学院;国防科学技术大学计算机学院;
【分类号】：TP308

【相似文献】