Computing Tasks Scheduling in Virtualized Geographically Distributed Data Centers Based on Energy Sharing

With the wide application of geographically distributed data centers in cloud computing, how to efficiently schedule directed acyclic graphs (DAG) with complex dependencies, workflow, and reducing energy consumption cost have become key challenges. This paper proposes a virtuality geographical distributed data center scheduling framework based on energy sharing. Firstly, a multi-DAG mergence model is constructed by introducing dummy sources and sink nodes to achieve the structural unification of heterogeneous workflows. Secondly, establish a multi-objective optimization model including makespan and energy consumption cost. Finally, use the NSGA-II dynamic scheduling algorithm to support elastic resource allocation and task migration between data centers. In contrast to the independent scheduling strategy, the method proposed in this paper maintains that the deadline violation rate only increases by 0.43% while reducing the energy consumption cost by 6.72%, the effectiveness of the spatio-temporal collaborative scheduling mechanism. This method provides a way to balance service quality and energy cost for geographically distributed data centers.