Research on the key problems of virtual machine deployment in cloud Platform

Research on the key problems of virtual machine deployment in cloud Platform

Jilin University Dong Yu Double

This paper analyzes the current popular virtual machine deployment strategy and its related improvement methods, and analyzes the current research work of cloud computing researchers on virtual machine deployment mechanism. Based on this, this paper studies the customization method of virtual machine image, designs the location selection strategy of virtual machine deployment based on performance energy consumption ratio as evaluation standard, and studies the key problems of virtual machine deployment mechanism. The main contributions of this article are: (1) customizing virtual machine mirroring via LFS (Linux from Scratch). To provide users with customized interface, cloud providers can dynamically update the services and applications provided to users, cloud platform according to the needs of users to generate the corresponding requirements profile, and generate a script to achieve virtual machine image automatically generated. The LFS custom virtual machine mirroring significantly reduces the size of the system image, increases significantly at boot speeds, and reduces system overhead and communication overhead for subsequent deployment processes. (2) The location selection strategy in the process of virtual machine deployment is studied. In this paper, a distributed parallel genetic algorithm is proposed to implement the location selection strategy of virtual machine deployment in cloud Platform DPGA (distributed parallelgenetic algorithm of slide-up Rogue deployment on Cloudplatform). For the purpose of minimizing energy consumption in the premise of guaranteeing system performance, in order to improve the coverage rate of initial population and to select initial population loosely in solution space, the DPGA crossover method with adaptive crossover and the multiple point mutation method of adaptive mutation rate are adopted, Genetic algorithms are used in two stages to obtain better and more accurate solutions. The deployment strategy implemented by DPGA can significantly improve performance energy consumption compared to other methods and is more stable than other methods. (3) The mechanism of virtual machine deployment is studied. This paper presents a deployment method for a custom virtual machine DCVM (deployment of customized virtual Machine). Deploying virtual machine mirroring, which was customized by the previous LFS method, radically reduces the overhead and communication overhead of the deployment process. DCVM combines the advantages of some methods in the virtual machine dynamic migration mechanism, uses the pre-copy and post-copy algorithm and combines the incremental compression mechanism to deploy through the multicast mechanism. DCVM significantly shortens overall deployment time, reduces the amount of data transfer during deployment, and enables rapid deployment of virtual machines in cloud computing environments.

Research on the key problems of virtual machine deployment in cloud Platform