Research On Resource Monitoring For Cloud Data Center

After years of development, datacenter has become a cloud datacenter with cloud computing features. As the core information system, cloud datacenter use virtualization technology to encapsulate resources for providing flexibility and scalable services. However, a lot of problems occur in this field such as large distribution, heavy computing load, high energy consumption and so on. In order to ensure cloud datacenter efficiency and stable operation, cloud datacenter must have the resources to monitor the system for its own characteristics. The existing resource monitoring systems are mostly focused on physical resources monitoring, or certain features monitoring. They do not apply to huge and highly virtualized cloud datacenter.A research of resource monitoring in cloud datacenters has been carried out in this paper. It involves three aspects.(1) Based on the characteristics of distributed structure of cloud datacenters, we design a monitoring structure of elasticity and a monitoring pattern of real-time monitoring and state monitoring. Therefore, a balance between the real-time of data and network load control has been achieved.(2) An optimization strategy has been designed to further reduce the network load of the monitoring system according to the state monitoring characteristics. This optimization strategy requires only a small amount of data transmission and computing, minimal overall impact on the system, the experiments show that the optimization strategy can be well reducing the condition monitoring data transmission, and it is a lightweight optimization strategies.(3) A new prediction algorithm of CPU idle rate has been designed. Based on the relationship between CPU idle rate and energy consumption of servers, it can monitor and predict the servers’ energy consumption indirectly, which has high practical value.At last, according to the monitoring pattern designed in this paper, we constructe a monitoring system. This system can monitor physical and virtual resources in cloud datacenters, and it can reduce the network load using state monitoring, at the same time, the energy consumption monitoring provides a more practical function.