Survivability is the ability of system to continue operating in the presence of failures or malicious attacks [4]. We present an original method for performing probabilistic analysis of survivability of network systems. We can simulate failures and intrusion events in our method and then observe the effects of the injected events. Our model is based on Markov Decision Processes which are generalization of Markov Chains and provides the analysis of probabilistic measures for network systems, such us: probability that a service that has been issued will be finished or the expected time it takes a service to finish. We illustrate the idea of our technigues by a simply example.