基于链路带宽的分布式存储系统框架及动态负载均衡技术

针对在节点间进行数据传输、备份等操作时，不同节点间链路带宽将限制分布式存储系统的性能的问题，借鉴软件定义网络及云存储技术的基本思想，通过对控制流与数据流的分离，提出了一种基于链路带宽的分布式存储系统框架及动态负载均衡技术.所提出框架中，在分布式数据节点之外引入一个中心控制节点，该节点掌握全局网络视图，全局网络视图中既包括数据存储表记录存储数据的分布，又包括当前链路带宽情况.在用户读取数据时，根据数据存储表确定需要的数据存放于哪些数据存储节点，指派相应的数据存储节点根据路由表发送数据包，并实时监测网络中的链路带宽负载情况，及时调整数据的传输路径.仿真验证所提架构和算法可以有效解决海量数据在不同位置的分配调度，以及缩短用户对大量数据提取的响应时间及提升存储速率等问题. 

Framework of Distributed Storage System and Dynamic Load Balance Technology Based on Bandwidth Condition

In the wide-area distributed storage network, the data are stored in different nodes which are extensively distributed. When the data need to be transferred between the nodes, the link bandwidths between different nodes limit the performance of the distributed storage system. To solve this problem, a distributed storage system framework and a dynamic load balance technology were proposed based on link bandwidth conditions. This idea was formed based on the principles of the cloud storage technology and the software defined network (SDN) to realize the separation between the control flow and the data flow. Under this framework, a central control node was established to manage the global network view, including data storage table to record the stored data distribution and current bandwidth situation. When a user application was sent for accessing data, the storage table was used to determine the locations of data. Then the corresponding data nodes were assigned to send data packets according to the routing table. The central control node was used to monitor the bandwidth load situation and dynamically adjust data transmission path. The simulation results verify that the proposed framework and algorithm can effectively solve the scheduling with huge data, reduce the response time, and promote the storage rate.