论文部分内容阅读
为了解决高速网络入侵检测系统(n IDS)的性能瓶颈问题,提出了可用于n IDS的负载均衡策略和算法。在阐述基于多引擎并行处理的n IDS框架的基础上,提出和分析了3种实用的n IDS负载均衡策略,重点论述了一种基于流的动态负载均衡算法——FDLB算法。该算法依据通过动态反馈和预测机制得到的当前引擎负载情况,以一个会话为分配单位,将新的网络数据包分发给当前负载最小的引擎。实验结果表明,在大流量多引擎情况下,FDLB算法的负载均衡效果要比轮转算法好得多。
In order to solve the performance bottleneck problem of high-speed network intrusion detection system (n IDS), a load balancing strategy and algorithm that can be used for n IDS is proposed. Based on the n IDS framework based on multi-engine parallel processing, three practical n IDS load balancing strategies are proposed and analyzed, and a dynamic flow-based load balancing algorithm called FDDLB is discussed. Based on the current engine load obtained through dynamic feedback and prediction mechanism, the algorithm uses one session as the allocation unit and distributes the new network packet to the engine with the least load. The experimental results show that the FDLB algorithm has a better load balancing effect than the round robin algorithm in the case of high-flow multi-engine.