There has been increased focus on improving the security and efficiency of the electricity grid resources. Many solutions have been proposed. Demand Response (DR) and Demand Side Management (DSM) are presented as popular solutions in studies towards achieving these goals. This paper studies an implementation of Model Predictive Control (MPC) for a group of buildings with thermal and electric demands. The MPC formulation solves a global problem by decomposing it into smaller decentralized problems. The results are then compared against a centralized MPC implementation for the same site and for the same period of time as the previous MPC implementation. This comparison is able to compare and contrast the advantages and limitations of different MPC implementations applied on real systems. Through this, the study aims to provide clear guidelines on what practices could prove useful when providing DR for groups of buildings utilizing predictive algorithms.