In this paper, an optimization approach for the allocation of remote-controlled switching devices in real-scale electrical distribution systems is presented. The proposed approach uses the metaheuristic Tabu Search to solve the switch allocation problem with the following three objectives: improving the grid's reliability indices SAIDI and SAIFI, minimizing the average energy not supplied, and reducing the acquisition, maintenance and operational costs of the installed or reallocated switches. In order to properly calculate the reliability indices of each allocation proposal, an additional optimal restoration process is solved for each zone of the network. The restoration process determines the status of the remote-controlled switches that minimizes the unsupplied demand due to a permanent fault, and that isolates the faulty section of the network. Finally, all tests were executed using realistic, unbalanced, three-phase EDS, with thousands of nodes and consumers, and numerous interconnected feeders, to demonstrate the scalability and efficiency of the proposed approach.