基于仿生学原理的云资源自主监控系统设计与实现

Design and implementation of cloud resource monitoring system based on bionics

为了解决现有监控系统因主控节点功能太过集中而导致某些时刻网络流量过大、系统扩展性差和无法及时应对节点失效的问题,提出了一种基于仿生自主神经系统(BANS)的新型云资源监控系统。首先,系统中引入了多级存储、分批上报的机制,将需要上报的监控信息分批次分时段上传汇总,使得在任何单一时刻系统内不会产生过大的流量和负载,保证了系统的稳定性;其次,系统中加入了类动态主机配置协议(DHCP)的主动发现机制以及定期轮询策略,使得系统在处理新节点加入,节点故障问题时,获得了类似仿生自主神经系统一样的自组织、自修复能力。实验结果表明,基于BANS的云资源监控系统实现了自组织与自修复的功能,并且可以有效降低系统内的通信流量,某些单一时刻能将流量降低到仅有原来的三分之一。

Concerning the problems of heavy network traffic, which is caused by the excessive load of master node, bad expansibility and poor ability to handle node failure in existing monitoring system, a novel Cloud Monitoring System based on Bionic autonomic nervous system（ B-CMS） was proposed. Firstly, B-CMS imported hierarchical storage and batch-wise report mechanism to upload the monitoring information, which can decrease the network traffic at any moment to ensure monitor system＇s stability. In addition, the use of similar Dynamic Host Configuration Protocol（ DHCP） and polling-driven heartbeat checking mechanism enabled the monitor system to get the self-organizing and self-repairing ability which is similar to Bionic Autonomic Nervous System（ BANS） when adding new node and handling node failure in autonomic way. The experimental results show that B-CMS achieves self-organizing and self-repairing, and effectively decreases network traffic. In some special moment, the network traffic is only one-third of the original system.