基于SDN的数据中心网络资源调度机制

摘要

With the emergence of Big Data applications in recent years, nowadays data centers are usually deployed with various kinds of Big Data applications and managed by resource manager such as Mesos. Mesos mainly focuses on computing resource and storage resource (CPU and memory), while ignoring network management. But research shows that efficient network resource management is crucial to optimize performance. In this work we present a network management mechanism based on SDN(Software Defined Network). This mechanism can allocate weighted network resource to different jobs based on preconfigured policy. It can allocate more bandwidth to high-priority job to optimize performance while keep the network performance isolation at the same time. We have implemented a prototype system based on SDN controller, floodlight and the experiments show that this prototype has strong usability and high expansibility.%随着大数据应用的不断丰富, 现在的数据中心通常部署着多种集群计算框架, 并由统一的集群资源管理器(如 Mesos)进行管理. 目前的集群资源管理主要关注计算资源和存储资源, 较少的涉及网络资源. 但研究表明高效的网络资源管理对于优化作业性能十分重要.本文提出了一种基于SDN(Software Defined Network)的数据中心网络资源调度机制, 该机制可以根据管理员预设的网络资源分配策略, 加权的进行网络资源调度, 为高优先级的作业分配更多网络资源以优化性能, 并且实现不同作业之间的网络性能隔离. 我们基于开源 SDN 控制器实现了原型系统, 并通过实验验证了该机制的有效性.