Despite the recent popularity of analyzing epidemic phenomena over networks, the budgeted control and protection of networks from viral propagations is not widely understood. In this paper, we blend methods from control theory and robust optimization to create a framework for network protection from epidemic environments using a limited control budget. As our emphasis is on the application of these methods to realistic contexts, our approach is designed to work with arbitrary networks, and it incorporates both decentralization as well as robustness to uncertainties in network topology. We illustrate tradeoffs between efficiency, robustness, and decentralization with respect to network protection, and we discuss methods that could build upon our framework to mitigate these tradeoffs.