Internet applications have been increasingly and widely used in every field in the society. On the other hand, the disruptions of services caused by software fault and hardware fault are more and more widespread. To improve the survivability of the Internet applications, we have conducted a study of virtualized clustering technology and software rejuvenation that follows a proactive fault-tolerant approach. In this paper, we present a virtualized two node cluster recovery model to increase the survivability level of Internet applications in which software and hardware failures occur. The system is modeled under Markov chains. Two different clustering systems are studied, traditional clustering system and virtualized clustering system, and compared in order to determine the best configuration which provides the highest level of survivability and availability.