A Study of Risk Evaluation Framework for Cloud Computing System

Virtualization asks for safer and better quality to service-oriented cloud computing system suppliers. Most of the traditional researching is focus on the risk assessment of the information system and DDoS, but lacking of researching on cloud computing in deep. So that the service-oriented cloud computing system risk evaluation researching is very essential. In this paper, we build a service-oriented cloud computing system risk assessment framework that using distributed dynamic status monitoring of virtual machine system and making risk prediction value to summarizing the final risk assessment level of the system as a whole. The model can be used in monitoring, identifying, predicting and evaluating for cloud computing security risk that is having effect on the risk evaluation of virtual machine node and whole cloud system. First we proposed a new method to identifying risk named LSA-GCC that used LSA for analyzing log files and cluster for classifying. The method doing the feed forward risk identification by collecting the log of the operating system and Web Service that are on VM node level,and the LSA-SAM method can improved the detecting accuracy for abnormal events. Then the risk predictiton method had been proposed that combined with RBFNN(Radial Basis Function Neural Network) and AHP. This method build a risk predicting model by AHP(Analytic Hierarchy Process) that consists of four indicators, that are P (performance), T (Time), A (event statistics) and R (risk identification). Each indicator include some monitoring parameters that can compute the risk indicator value by weight matrix. Experimental results show that MRPGA-RBF risk prediction method can predictive risk value accurately.