Towards Comprehensive Safety Assurance in Cloud-based Systems

When a system malfunctions or a required service is not provisioned in a timely manner, this can lead to human injuries or fatalities. Increasingly, safety-critical system operation relies on offloading of functions into the cloud, including realtime ones. For this reason, the cloud-based systems that are involved must exhibit a high degree of dependability. Thus, to determine a system's dependability, a comprehensive safety assurance process is needed to allow integration in a development process allowing iterative improvement to tackle complexity and changing requirements. Based on the principles of adaptivity and flexibility we propose a 3-leveled safety analysis process for building up a necessary resilience against disruptions and failures of various scale, nature and operation context dynamism. A multilevel genuine combination of traditional and contemporary safety methods is a key to provide necessary system resilience in a cloud. The right composition, expected yield and applicability of methods that will be best suited for cloud context is a subject of our research.