Software Maintenance Optimization Based on Stackelberg Game Methods

Application servers (AS) of virtualized platform may suffer from software aging problem. In this paper, we first formulate the system model including three virtual machines. Two of them act as the main servers, and the third machine acts as the backup node. The motivation of our formulated model is that the relationship between the service provider and the service maintainer is collaborative as well as having different goals between them, the service provider as a leader wants to maximize his system availability, while the service maintainer wants to minimize his maintenance cost. Thus, the problem of maximizing availability and minimizing cost between the service provider and service maintainer is Stackelberg game based. Next, we assume that the AS degradation is caused by resource consumption due to memory leaks for the AS on the active VMs, and we present the system degradation states based on Markov renewal processes. We give the analytical definitions of threshold levels for R_alert at each VM, which are used to determine the optimal rejuvenation schedules. In addition, we obtain the steady-state availability expressions for the system and the mean maintenance cost. Finally, we give the Stackelberg strategy with the open-loop information and the solutions for the game theory by a numerical illustration.