The main goal of this paper is to make a step towards the development of a unified approach that incorporates both reliability and traffic engineering aspects into a common, unified model. The approach is motivated by a simple fact: if the traffic flow is blocked between two nodes, then for the end user it is irrelevant why it is happening (whether it is due to failure or congestion). What really matters is that the network should provide sufficient connectivity and bandwidth between the end-nodes, at least with a prescribed probability, taking into account that failure and congestion may both happen. Existing methods do not support such a joint analysis of network reliability and traffic, since the two scenarios are addressed by different models. The proposed approach can provide a tight and unified estimate of the probability of any event in the network that is expressible via the availability of sets of links. It takes both reliability and traffic engineering aspects into account in...