Virtual data center (VDC) embedding is being regarded as a critical issue to provide performance guarantees for the future success of cloud computing. Most existing works neglected the embedding of virtual switches (VSs), especially, on the physical switches (PSs), resulting in low utilization of physical resources. Moreover, most works assumed that multiple virtual nodes in the same request cannot be embedded in the same physical node, leading to low embedding efficiency in data centers. To address the above two problems, in this paper, we first propose two fine granularity models to formulate the VDC embedding problem efficiently with the consideration of two possible embedding positions of VSs. Then traditional flow conservation law is modified to allow multiple virtual nodes to coexist on the same physical node. Finally, we propose multiple efficient embedding algorithms to solve two NP-hard problems. Comparing with existing methods, our algorithms employing heuristic information from the relaxed model can find sub-optimal solutions in polynomial time. Extensive simulation results show that our proposals outperform the existing methods in terms of acceptance ratio, revenue, and utilization.

虚拟数据中心（VDC）的嵌入被视为为云计算的未来成功提供性能保证的关键问题。现有的大多数工作都忽略了虚拟交换机（VS）的嵌入，尤其是在物理交换机（PS）上的嵌入，从而导致物理资源的利用率较低。此外，大多数工作都假定不能将同一请求中的多个虚拟节点嵌入同一物理节点中，从而导致数据中心的嵌入效率低下。为了解决上述两个问题，本文首先提出了两个精细的粒度模型，以考虑VS的两个可能的嵌入位置来有效地表达VDC嵌入问题。然后修改传统的流量守恒定律，以允许多个虚拟节点共存于同一物理节点上。最后，我们提出了多种有效的嵌入算法来解决两个NP难题。与现有方法相比，我们的算法采用松弛模型中的启发式信息，可以在多项式时间内找到次优解。大量的仿真结果表明，我们的建议在接受率，收入和利用率方面都优于现有方法。