Dependability modeling and performability analysis of server virtualized systems (SVSs) are today a key challenging problem to evaluate cost-effectiveness and savings in green data centers. This paper tackles various issues related to performability analysis of server virtualized system (SVS) handling workload-aware power management (PM) mechanism and subject to software aging, unplanned failures and Migrate-VM rejuvenation. In this work we develop a modeling approach based on stochastic reward nets (SRNs) to investigate dependencies between several SVS modules including virtual machine monitor (VMM), virtual machine (VM), data intensive applications and power-manageable component (PMC) with workload-aware timebased PM mechanism. We show through numerical analysis how availability, power usage and power-performance trade-off, of SVS, are impacted by aging and workload burstiness.