Network virtualization is a new technology that provides a transparent abstraction of the networking resources. The most important challenge in network virtualization is the allocation of the physical substrate network among the pool of the active virtual networks (VNs). Our work in progress aims at allocating the node resources in a virtualized networking infrastructure. We believe that this allocation should be dynamic to lead to higher performance and better utilization of the physical resources. In this paper, we propose two models for dynamic node allocation for multiple VNs. The first model uses game theory and market-based approach in order to better allocate the physical node. The second one proposes a dynamic weighted round robin (WRR) approach where each VN receives a fraction of the physical node according to an estimation of its current number of waiting packets and its weight. Both models use a distributed approach to minimize the packet delays inside the physical router and to fairly allocate the nodes between different VN.