Research On Anomaly Monitor Syetem Of Linux Kernel Based Virtual Machine Oriented Resources Service

Infrastructure as a Service (IaaS) is one kind of oriented resources service model in cloud computing, it provides user the service by packaging the foundation resources such as hardware equipment. Although cloud computing provides stronger computing ability and more wide bandwidth, it arises much virtual machine anomaly and results in much damage. For example, malicious users can attack by adopting the hardware of supplier. And other users upload system image or programs with malicious code in the absence of information. Anyway, both of these cases can result in damage to the provider of cloud service or the benefit of users. In order to identify the anomaly of the virtual machine, this paper investigates and develops a Xen anomaly monitor system to provide effective recognition to anomaly. And this anomaly monitor system is based on open-source IaaS cloud platform which is Eucalyptus.This paper analyzes the current research condition of cloud monitor. The current research is focus on monitoring the virtual machine in order to implementing the load balance of virtual machine. Due to the application feature of cloud computing, this paper proposes the anomaly monitor scheme of virtual machine under cloud computing.The concrete work is as follows:①We have research on IaaS framework and adopt open-source Eucalyptus to build cloud computing platform. Eucalyptus and Xen based virtual machine anomaly monitor system is proposed, and the data collecting module and data analysis module are designed.②The run features of Xen are studied. Based on Xen, we design and implement a data collecting strategy to collecting multidimensional data, and a storage method to store multidimensional data of Xen.③Study the implement principle of libpcap and implement the libpcap based data collecting functionality to capture the network metric of Xen.④Though comparing with different kinds of method for dealing with data, this article adopts Principal Component Analysis to analyze data which is gathered. In this article, it is a key to analyze how to build the module of PCA and how to check collecting data by two statistical methods called T2 and Q.⑤We use event injection for testing the system with the different kinds of anomaly situations. Finally, we conclude this paper. And we also present the shortcomings and plan the next work.