IaaS And SaaS Layer Reliability Research In Cloud Computing

Cloud computing provides information services with an effective business model. It simplifies storage and computation as easy-to-use resources, which are provided for the users by the network in the pay-on-demand pattern. Cloud computing has characteristics of flexibility, convenience and economy, etc. However, these characteristics also make cloud computing more complex compared with the traditional computing model, and are more likely to fail. The reliability of cloud computing is directly related to the quality and performance of cloud services, and is of great significance to the development strength of cloud computing industry. This paper respectively does some research into fault type and the test method of reliability in IaaS layer and SaaS layer. And propose a snapshot method of virtual network in IaaS layer. In addition, build fault tree model of SaaS layer based on fault tree analysis method. Specific work is as follows:(1) Summarize the research actuality of cloud computing reliability and existing problems of the cloud computing. And put forward the corresponding solutions according to the problems of various types.(2) Respectively research reliability of IaaS layer in physical layer and virtualization layer. Summarize fault scenarios, fault detection and mitigation methods of hardware in physical layer. At the same time, for the virtualization layer, propose snapshot method of virtual network in IaaS layer. Improve the performance of a snapshot from both the dynamic migration and the similarity of snapshot images. As a result, effectively reduce the system downtime caused by the snapshot, and reduce consumption of the flow when the snapshot transmission.(3)Build Jelastic cloud platform. And analyze and summarize various fault types and logical relationship between faults in SaaS layer, by monitoring CPU、RAM、HDD、Network in real time. Build fault tree model in SaaS layer, including 10 major failure events and 84 basic failure events, and obtained the probability of occurrence and structure importance degree of each major event through experiments, providing references for fault handling and improvement of SaaS layer.