The Research Of Cloud Platform’s Maintainability

Because of the prosperity and popularization of cloud computing, more and more enterprises choose cloud computing technology to reform their original data centers. Some enterprises begin to lease cloud services of large manufacturers. At the same time, the researchers keep themselves focused on the security and utilization rate of the cloud platform. But very few pay attention to the maintainability of cloud platform. Thanks to the practice and thinking of building the DartCloud cloud platform, this paper makes a comprehensive exposition on it. In this paper, the maintainability of cloud platform is described mainly from the following aspects: rapid deployment of cloud platform and the establishment of monitoring system, image management and the support of business expansion.The work is funded and supported by the 863 project "The development of Internet language translation system". The main work and contributions are summarized as follows:· Making a comprehensive exposition on the maintainability of cloud platform, mainly focusing on the following aspects: rapid deployment of cloud platform and the establishment of monitoring system, image management and the support of business expansion.· Analyzing how OpenNebula and Xen work, and give the guide to build a cloud platform. Aiming at the problem of low efficiency of the existing deployment, proposing and implementing the operating system batch deployment and automation deployment to improve the scalability and deployment efficiency of cloud platform. The relevant patent has been accepted and publicity.·Implementing a cloud resource monitoring and management system based on Ganglia, providing a basis for performance tuning by showing the CPU utilization and other indicators.· Proposing and implementing an image management solution, including the access control, image version management and image automation management based on the incremental replication strategy. The relevant patent has been accepted and publicity· Through achieving the collection of experimental data by DartCloud resource monitoring system, analyzing the performance of parallel computing application in big data with different resource expansion method, including scale-out and scale-up. Providing a basis improving resource utilization and reducing the cost of performance tuning management. The research results have been written and published in a conference paper.