Fog computing plays an important role to improve the efficiency of time-sensitive applications, cost reduction, and proper data management. In this paper, the above goals are made possible by processing some time-sensitive data and aggregating the rest of the data, by the modules in the fog data manager and associated storage servers, before sending it to the cloud servers. In this paper, we set the optimal workload allocation to the cloud and fog system by considering the compromise between costs of energy consumption of cloud and fog system, transfer bandwidth and revenue losses due to the WAN propagation delay, and time constraints in the Fog and Cloud system. The mixed linear programming obtained by IBM ILOG CPLEX software is solved to achieve the lowest possible cost by optimally allocating loads to fog devices and cloud servers. In addition, we show that using this approach, the total cost can be reduced by about 29%.