Data centers are expected to consume 3–13% of the global electricity in 2030. Thus, different measures such as reducing the carbon footprint and utilizing renewable energy sources should be explored to limit the environmental impact of cloud computing. By configuring data centers as microgrids, able to operate both in connected and disconnected mode from the main grid, and by exploiting electrical storage and renewable energy production, carbon reductions might be possible. Default workload schedulers are not yet fitted with the capabilities of knowing the source of energy and thus cannot act accordingly. Therefore, we propose a way of using the Scheduling Framework for Kubernetes to implement two different green strategies, one focusing on renewable energy self-consumption, the other taking into account also the carbon intensity of the local main grid. Additionally, we propose a framework for simulating microgrids and nodes, where the effects of scheduling are observable. By simulating a large number of microgrids and configuring them with servers (also known in this paper as nodes), our plugin determines the best possible microgrid according to our green strategies to schedule a workload. We simulate workloads based on a real Azure worktrace, and we show that our proposed plugin can either improve self-consumption by \(35.86\%\) , or have a much smaller carbon footprint in comparison with the default Kubernetes scheduler. Further study could introduce different workload types and specific hardware requirements, a non-heterogeneous node specification or different, better-fitting renewable energy sources tailored to the geographical region. All code and data used in this study is available open source, more information under the Declarations section.