Robustness improvement of component-based cloud computing systems

With the increasing popularity of Cloud computing systems, the demand for highly dependable Cloud applications has increased significantly. For this, reliability and availability of Cloud applications are two prominent issues for both the providers and the users of Cloud. However, ensuring these two properties in Cloud applications is often very difficult. This is especially because of the characteristics of the Cloud computing paradigm, which is a combination of hardware and software components in a dynamic setting. In spite of the challenges, it is often a key objective to ensure reliability and availability of such applications to guarantee the expected quality of service (QoS). Many methods, strategies and approaches have been proposed in the existing literature; however, as far as we have investigated, these works do not provide a global solution that could provide reliability, availability and high margin of QoS at the same time (for such systems). In this paper, we propose a novel formal framework for constructing reliable and available Cloud components using the DRB (distributed recovery block) scheme. The aim is to provide a strategy that can enhance Cloud dependability through the uniform treatment of software and hardware faults by constructing fault-masking nodes. A fault-masking node is suitable for handling (i.e., detection and tolerance of faults) software, hardware, and response time faults using both the acceptance test and try blocks to ensure safety and liveness properties at the same time.