Currently, cloud services are rapidly developing and are becoming more widely used. One of their most important attributes is availability that determines the need for research in this area. In the past, dozens of publications focused on the availability of cloud services. However, some of them provide only qualitative consideration, others do not consider whole infrastructure of cloud services from end to end. This paper clarifies the definition of service availability and gives quantitative evaluation of end-to-end availability of cloud services for typical situations taking into account all components of the service infrastructure. It shows that achieving high availability requires redundancy for such components as network connections and data centers.