Survivability is a key attribute to measure large-scale network security and has become a hot research issue. Large-scale network is a complex system, which has the typical survivability characteristics of complex system. The current network survivability studies usually aim for a specific aspect of network. In this paper, we attempt to study the survivability of large-scale network based on the characteristics of complex system. Firstly, the concepts and features of large-scale network survivability are described. Secondly, the definition and process of large-scale network survivability are proposed based upon the information entropy theory. Thirdly, the distance and time measurement of large-scale network survivability are proposed to measure the degree of large-scale network survivability. Finally, the proposed model are validated to accurately measure large-scale network survivability through case studies.