Fault tolerance management in IaaS clouds

Fault tolerance, reliability and availability in Cloud computing are critical to ensure correct and continuous system operation also in the presence of failures. In this paper, we present an approach to evaluate fault tolerance mechanisms that use the virtualization technology to transparently increase the reliability and availability of applications deployed in the virtual machines in a Cloud. In contrast to several existing solutions that assume independent failures, we take into account the failure behavior of various server components, network and power distribution in a typical Cloud computing infrastructure, the correlation between individual failures, and the impact of each failure on user's applications. We use this evaluation to study fault tolerance mechanisms under different deployment contexts, and use it as the basis to develop a methodology for identifying and selecting mechanisms that match user's fault tolerance requirements.