Research And Application Of Monitoring System For Cloud Computing Platform

Cloud computing is more complex than grid computing because it is virtual, dynamic with multiple levels. Monitoring is an important part of the cloud computing platform, which plays an important role to improve its service quality. Therefore, it is of great significance to study the monitoring system of cloud computing platform.Firstly. I did research on the data acquisition methods of the existing monitoring system. Based on the advantages of the widely used push and pull mode of data collection, I suggested design a kind of push-pull hybrid data acquisition algorithm to increase the data acquisition efficiency. Experiments show that this algorithm can reduce the times of acquisition and the intervention of system.Then, we need to build a prediction model based on the action learning as to the lagging problem of the data real-time procession of the monitoring system. According to the characteristics of the data monitored and time limits of predicting, we subdivided the prediction model into performance data long-term prediction model, short-term threshold cross-border model and accident to predict in advance model. We designed algorithm for the model in combination with the data mining technology and machine learning techniques. In addition, we need to design accuracy evaluation equation of the prediction model and research the dynamic adaptability of the prediction model.Finally, combined with the characteristics of cloud computing platform, we came up with cloud computing platforms component model which is hierarchical and extensible. After that, considering scalability, low coupling, real-time, accuracy, low intervention and generality, I designed and implemented the monitoring system, validated the feasibility of push-pull hybrid data acquisition algorithm.