Software aging refers to the phenomenon that long-running software shows signs of increasing failing rate, overmuch resource usage, and performance degradation. Software rejuvenation is a proactive approach to dealing with this problem. However, commonly used rejuvenation methods involve a relatively larger overhead. An alternative is to reduce the severity of software aging by online adjusting the settings of related parameters of the system. In this paper, we conduct controlled experiments to analyze severity of software aging under different settings of related parameters. Based on the experimental data, a metric is defined to measure the severity of software aging. A multiple-input and multiple-output (MIMO) model is then constructed to formulate the relationship between severity of software aging and related parameter settings. The proposed MIMO model gives us a way to control the severity of software aging at runtime.