Big Data presents itself as a double-edged sword: both promising and problematic. While large-scale data collection and analytics are revolutionizing scientific discovery, they also pose serious challenges for infrastructure in cost-constrained environments. NASA’s Center for Climate Simulation named data infrastructure the biggest problem facing climate science, with the growth of storage needs outstripping the growth of necessary computing power by 6.67x. New advances in intelligent software-defined data centers (SDDCs) have the potential to combat these challenges, but they must incorporate the ability to optimize data dependence constraints quickly and performably. We formally define data dependence constraints for Big Data storage architectures, present automated methods for encoding the system states of such architectures into constraints suitable for SMT solving with Z3, and describe methods for efficiently resolving our data dependence constraints to enable SDDCs. We include proofs of correctness of our new constructions and optimizations inspired by experimental evaluation.