Aiming at the problem of massive data parallel processing in high speed network for intrusion detection system,a data distribution method based on capability and task completion time was proposed.Load status of nodes was evaluated by monitoring the resource utilization of nodes.When data distribution was conducted periodically,data set allocated to each node was adjusted to make sure that data set was processed and completed by each node in the cluster at almost the same time,by analyzing the relation between the task completion time under the real-time load and the task execution time under the no-load.Results of test show that the proposed method can effectively narrow the gaps among the task completion times of nodes in the cluster and it has good load balance.%针对入侵检测系统对高速网络中海量数据并行处理的问题,提出一种基于能力与任务完成时间的数据分配方法.通过监测结点的资源利用情况,评估结点的负载状态;在周期性地进行数据分配时,通过分析结点在实时负载情况下的任务完成时间与无负载情况下的任务执行时间的关系,调整分配给结点的数据量,使集群内各结点对数据处理的完成时间趋于一致.仿真结果表明,该方法有效缩短了集群内各结点间任务完成时间的差异,具有较好的负载均衡性.
展开▼
