Proactive maintenance with variant workload under Distributed Multimedia Application Systems

Distributed multimedia applications such as video on demand (VoD), require dynamic quality of service (QoS) guarantee from service servers for their continuous multimedia streams. In this paper, we build the service availability Markovian model for unified failure-recovery mechanism under variant load scenarios, by calculating local and global kernels of the Markovain model, we get the steady-state availability and unavailability probabilities. Numerical results show that there exist differences between the system availability and the request perceived availability under variant load state (the nature of this kind of system's scenario). This provides strategies for improving VoD system's availability