Towards the user-centric analysis of the availability in IaaS

Availability is a key property in computational services and, therefore, is guaranteed by Service Level Agreements (SLAs) from the majority infrastructure services, such as virtualization (Amazon EC2, Windows Azure, Google Cloud, Joyent, Rackspace, ...) and storage (Ama zon S3, Google Cloud Storage, ...). These SLAs describe availability in natural language and there are important differences in the scope and penalties that each service provides. Furthermore, descriptions use spe cific domain terms so they are difficult to understand by service cus tomers. These circumstances make that availability analysis is a tedious, error-prone and time-consuming task. In this paper, we describe in de tail this problem and provide a first approach to deal with these SLAs supported on current SLA analysis techniques.