Cloud computing in recent years has become very popular and, indeed, many current applications are served from the cloud. Because the cloud architecture is based on virtual machines, VMs and VM management are quite important. Monitoring a large number of virtual machines is an important research topic. Since the establishment of the Personal Information Protection Act, there has been more attention focused on cloud security. Among cloud security topics, log analysis allows insight into virtual machine operation. If there are incidents, they need to be reported instantly in order to maintain information security. Consequently, log collection and real time warning systems are necessary. In this paper, we propose a distributed management architecture and evaluate its system performance. We also discuss the effects on the system due to bandwidth and background traffic ratios, VM numbers and transmission time using a set of simulations. The results show the proposed architecture not only saves time, but also reduces the load of log transmission for monitoring cloud services.