We study how to embed virtual software-defined networks (vSDNs) cost-effectively over a substrate network with distributed network virtualization hypervisors (NVHs) for virtual datacenter (vDC) formulation to support Big Data applications. Specifically, we try to jointly optimize the embedding schemes of the control and data planes of each vSDN, i.e., to minimize the data plane's resource consumption and limit the number of NVHs used in the control plane simultaneously. We first formulate an integer linear programming (ILP) model to solve the problem exactly, and then design a heuristic to reduce the time complexity. Simulation results suggest that our proposed algorithms can embed vSDNs cost-effectively and significantly outperform the existing scheme based on global resource capacity (GRC).