The Smart Grid paradigm is influenced primarily by the need to integrate renewable energy from wind and solar resources. Two main tools that have been proposed to carry out integration are (i) decision and control that makes use of all available information via a cyber-physical infrastructure that includes communication, and computation, (ii) Demand Response (DR), the concept of controlling loads using smart meters and devices as well as economic signals. Given that the pertinent information is available at multiple time-scales and from multiple sources, decision and control algorithms need to necessarily have a distributed, hierarchical structure. In this paper, we propose a distributed cyber-physical control architecture to match energy supply to energy load at the sub-transmission and distribution levels. A hierarchical model of the overall cyber-physical energy system is introduced, and includes the dynamics of the grid at the primary, secondary, and tertiary levels. With a goal of ensuring frequency regulation using optimal allocation of resources including renewable energy resources (RER), a distributed control methodology is presented and numerically evaluated in the presence of intermittency in the RERs.