Cloud computing is a novel paradigm for the provision of service on demand. Through the use of virtualization, physical resources are converted into “resources pool” which provides service on demand. Due to the complexity and diversity of user requests of the cloud computing system, its performance is difficult to model and analyze. Although there are some researches on cloud service performance, very few of them addressed the issues of resources sharing among VMs and its impact on service performance. This paper presents a model for performance analysis of cloud services. The model considers the resources sharing among VMs. In addition, various types of failures, such as VMs failures, physical servers failures and network failures are also considered. The service requests are also relaxed compared with prior research. A service request is divided into many subtasks and each subtask consists of a series of data processing and transmission. The average service time of service requests is obtained. And a numerical example is presented.