The impact of hierarchical structure on efficiency of Cloud monitoring

Cloud computing systems are often seen as dynamic pools of Virtual Machines (VM) installed on provider side physical machines to be offered to Cloud users. Cloud customers could use these Virtual Machines as services, platforms or as a whole infrastructure. However, in practice the infrastructure of a computing Cloud includes several levels, such as virtual gateways, virtual clusters and virtual nodes. In this paper, we pursue a study of the impact of a hierarchical structure, formed of three levels, on the process of monitoring the system with the main goal of discovering symptoms of malicious behaviors in Clouds. We address in this paper two major questions. First question refers to optimize the number of clusters in the hierarchical structure to guarantee efficient monitoring. The second question, posed in some previous papers in this area, concerns efficient distributed implementation of the monitoring process; Namely, how to choose locally the next VM to be visited by a Forensic Virtual Machines (FVM) in a light and local way.