Caching at the edge of wireless networks is a key technology to reduce traffic in the backhaul link. However, a concentrated amount of requests during peak-periods may cause the outage of the system, meaning that the network is not able to serve the whole set of demands. The outage probability is a fundamental metric to take into account during the network design. In this paper, we derive the analytical expression of the outage probability as a function of the total amount of users requests, library size, requests distribution, cache size and capacity constraints on the backhaul resources. In particular, we focus on a scenario where end-users have no direct connection to the master node which holds the complete library of content that can be requested. A general formulation of the outage is derived and studied for two relevant caching schemes, i.e. the random caching scheme and the most popular caching schemes. The exact closed form expressions presented in this paper provide useful insights on how requests, memory and resources can be balanced when the parameters of a cache-enabled network have to designed.