Reliability optimization of heterogeneous cloud computing systems considering K-mixed redundancy strategy

Redundancy allocation is an effective technique for enhancing the reliability of cloud computing systems. For the reliability optimization of cloud computing systems considering K-mixed redundancy strategy and heterogeneous non-exponentially distributed virtual machines (VMs), there is a lack of a universal and exact mod eling and solving method. In this paper, we propose a system modeling method based on dynamic fault trees and conditional binary decision diagrams (DFT-CBDD) for the system considering K-mixed redundancy strategy to obtain the exact analytical formulation of system reliability. Then, a novel mathematical model is developed for the redundancy optimization of the VM system considering active, standby, and K-mixed redundancy strategies, and a genetic algorithm (GA) is designed to solve it. Finally, the proposed method is applied to a practical cloud computing system for optimal reliability under limited cost and energy and the strong performance of K-mixed redundancy is verified in reliability enhancement.