Methods with low complexity for evaluating cloud service reliability

The critical significance of cloud computing lies in its ability to deliver high performance and reliable calculation on demand to external customers over the Internet. Because of heterogeneous software/hardware components and complicated interactions among them, the probability of failures improves. The services reliability arouses more attention. Cloud reliability analysis and modeling are very critical but hard because of the complexity and large scale of the system. The connectivity of subtasks and data-resources can affect the system reliability. This paper proposes the analysis methods of cloud service reliability based on a simple manner on two conditions: independent failures and correlated failures, which uses Graph Theory, Bayesian networks and Markov models. Simulation results show that time complexity of our proposed method has been greatly improved than traditional algorithms. Our new methods ensure the precision of reliability calculation.