Constrained physical design tuning

Existing solutions to the automated physical design problem in database systems attempt to minimize execution costs of input workloads for a given storage constraint. In this work, we argue that this model is not flexible enough to address several real-world situations. To overcome this limitation, we introduce a constraint language that is simple yet powerful enough to express many important scenarios. We build upon a previously proposed transformation-based framework to incorporate constraints into the search space. We then show experimentally that we are able to handle a rich class of constraints and that our proposed technique scales gracefully. Our approach generalizes previous work that assumes simpler optimization models where configuration size is the only fixed constraint. As a consequence, the process of tuning a workload not only becomes more flexible, but also more complex, and getting the best design in the first attempt becomes difficult. We propose a paradigm shift for physical design tuning, in which sessions are highly interactive, allowing DBAs to quickly try different options, identify problems, and obtain physical designs in an agile manner.