Key to nimble network recovery is gaining leverage over network operation. Traditionally networks are managed at a low level, making recovery a challenge. With the right tools, we can manage them at a higher level, gaining leverage like that achieved over hardware with high-level programming languages. Good programming language design revolved around identifying appropriate high-level abstractions, which required understanding the objectives of programmers. The situation is almost identical for networks. The networking community is trying to understand the objectives of network managers, specifically, what constitutes network management. To traditional language designers, this looks like a familiar problem. It is solved by new abstractions for managing the network hardware directly. This is the standard view within industry, as evidenced by software-defined networking. However, the view is myopic. A different view of the problem can lead to even greater leverage for the Navy. Abstractions must be designed around managing network applications and the requirements of mission-critical flows, not hardware. Such abstractions can be translated into lower-level ones for hardware management so that the network ultimately realizes them. The focus of our research is designing appropriate abstractions and algorithms for their translation.
