Hierarchical control is a widely used strategy that can increase resilience and improve the reliability of the electrical network based on microgrid global variables. The large amounts of data required during transitions prompt the use of more reliable and flexible communications to achieve the control objectives. Such communications can involve potential cyber vulnerabilities and latency restrictions, which cannot be always addressed in real-time. To accurately capture the system’s overall operation, this paper proposes a co-simulation framework driven by flexible communications and a resilient control algorithm to regulate the frequency and voltage deviations in a networked microgrid. Model-based predictive control has been implemented, to avoid slow transient response associated with linear hierarchical control. Software-Defined Networking (SDN) is responsible for increasing the communication intelligence during the power-sharing process. The effects of critical communications and overall system performance are reviewed and compared for different co-simulation scenarios. Graphical Network Simulator (GNS3) is used in combination with model-based predictive control and SDN, to provide latency below 100 ms, as defined in IEC 61850. Testing of the proposed system under different cyber attack scenarios demonstrate its excellent performance. The novel control architecture presented in the paper provides a reference framework for future cloud computing-based microgrids.