云数据中心集群间网络性能优化的探讨

发布时间：2018-02-21 19:28

本文关键词： 集群网络性能优化吞吐量　出处：《电信科学》2015年05期 　论文类型：期刊论文

【摘要】：现代的大规模云计算任务往往需要多个集群协作完成,因此,规划连接集群的网络、保证网络的性能具有重要意义。通过介绍一次集群间网络的性能问题及其解决过程的真实案例,对数据中心集群间网络性能优化进行了探讨。这个案例来源于一个用于大规模实时数据处理的数据中心,在一次大规模并发任务中发现,在带宽并未完全使用的情况下发生了严重的分组丢失现象。基于对拓扑结构的分析,对性能瓶颈进行了定位。通过搭建测试环境并进行实际测试,对性能瓶颈的来源进行了发掘,发现是某交换机在链路未满负荷时由于处理能力和缓存不足造成的。最后,基于模型,分析了各个要素对于性能的影响,并基于分析结果设计了基于增加帧长度的解决方案。
[Abstract]:Modern large-scale cloud computing tasks often require multiple clusters to work together, so plan the network to connect to the cluster, It is of great significance to ensure the performance of the network. By introducing the performance problems of a cluster network and the real cases of its solution, This paper discusses the optimization of network performance among data center clusters. This case comes from a data center for large-scale real-time data processing, and it is found in a large-scale concurrent task. Serious packet loss occurs when bandwidth is not fully used. Based on the analysis of topology, the performance bottleneck is located. The source of the performance bottleneck is found to be caused by the insufficient processing capacity and cache when the link is not fully loaded. Finally, based on the model, the influence of each element on the performance is analyzed. Based on the analysis results, a solution based on increasing frame length is designed.
【作者单位】：中国石化石油勘探开发研究院;
【分类号】：TP308;TP393.06

【相似文献】