In this paper, a real-time tertiary control algorithm for DC microgrids will be developed and implemented. Tertiary control, in the context of this paper, refers to the control layer that transcends the boundaries of a single microgrid, and hence supervises a group of microgrid central controllers (MGCC). Microgrids are considered as dynamic resources and loads that can inject or draw active and/or reactive power to/from the grid, depending on their loading condition. This paper focuses on microgrids with high penetration of renewable resources. To overcome renewable energy intermittency, the developed algorithm virtually aggregates neighboring microgrids into clusters, which virtually exchange energy through the distribution grid infrastructure stabilizing the main power infeed. For improved short-term planning, the developed algorithm periodically forecasts the net available energy from each microgrid. In addition, it solves a set of load flow equations, governed by predefined constraints, to determine (near-) optimal set points to the various microgrids, which consequently improves the efficiency and voltage stability of the distribution network, and increase its level of survivability. This study highlights some key examples of how the microgrids can be used to support the operation of the main grid, e.g. enhancing the voltage profile on a distribution feeder. The algorithm was validated experimentally using a laboratory scale testbed setup.