A cyber range, a special type of computer network emulation environment, allows users to evaluate network security technologies. Traditional network emulation methods, such as packet-level discrete event simulation, have difficulty meeting the realistic network security evaluation requirements of a cyber range. Thus, a series of new technologies, such as container-based network emulation and virtualization-based network emulation, have emerged. The key problem in network emulation is the contradiction between fidelity and cost. To solve this problem, considering the respective advantages of mathematical modeling, packet-level discrete event simulation, container, virtualization and physical networks, this paper introduces a multiscale network emulation method based on the above five methods, named MSCVP, to achieve a tradeoff between fidelity and cost. To improve the emulation performance, on the basis of cloud computing platforms, MSCVP focuses on the seamless integration of packet-level discrete event simulation, container, virtualization and physical networks. Given the dynamics of network topologies, MSCVP further integrates mathematical modeling. The experiments show that MSCVP can achieve seamless integration of the above five methods, enabling worm propagation at the million-node level and dynamic network emulation of space–ground integrated information networks. Compared with the other methods, MSCVP enhances the efficiency of network emulation deployment, achieving an implemented of 56.87%.