基于动态副本技术的云存储负载均衡研究

以系统总响应时间最小化为目标,以文件热度为依据,提出了一种多时间窗负载均衡策略。在计算文件热度时,不仅考虑了访问的次数和大小,还将I/O访问时序引入到文件热度统计中,该方法能有效控制短时间突发性数据访问导致的不必要副本创建。在多时间窗负载均衡策略中,设置了三种不同大小的时间窗口,分别实现了存储节点负载均衡、文件副本的负载均衡以及低热度文件多余副本的删除工作。实验数据表明,多时间窗负载均衡策略能显著降低I/O访问响应时间。

Load balancing study in cloud storage based on dynamic replica technology

For the load balancing of cloud storage system, this pape provided a multi-time windows load balancing strategy which had three time window with different size, this strategy could achieve the load balancing of storage nodes and replicas. It designed new method to statistic the heat of files, it associated the heat of a file with time sequence, which could effectively avoid the creation of replica because of burst of file access in a brief moment. The experiment indicates that multi-time windows load balancing strategy can effectively reduce the response time of I/O access.