Optimized Deployment of Data Intermediation Services

Data intermediation services are an emerging paradigm of services aiming to provide accountability for sharing data. Even though significant effort is committed to conceptualizing the functionalities and business models of these services, optimization approaches for deploying such services efficiently are still unexplored. In this paper, we introduce a system for deploying intermediation services, and we propose a resource management mechanism that minimizes the deployment costs using reinforcement learning and optimization. To evaluate this approach, we conduct experiments over a widely distributed computing infrastructure, and we show that, compared to alternatives, the proposed mechanism decreases resource utilization by 57% and lowers the pricing cost by 63%.