Many systems design, configuration, runtime and management decisions must be made from a large set of possible alternatives. Ad hoc heuristics have traditionally been used to make these decisions, but they provide no guarantees of solution quality. We argue that operations research-style optimization techniques should be used to solve these problems. We provide an overview of these techniques and where they are most effective, address common myths and fears about their use in making systems decisions, give several success stories and propose systems areas that could benefit from their application.