Network systems, such as Internet, smart grids, transportation networks, social networks, etc., play a critical role in human society. However, due to their inherent vulnerability as well as the limited management and operational capability, these network systems are constantly under the threat of malicious attackers. Therefore, in such attack-defense scenarios, it is particularly significant to make the best use of defenders’ limited resources and capability. In this paper, we propose a networked Colonel Blotto game for the attack–defense strategy, where the attackers and defenders allocate the limited resources on each node, and their utility depends on certain network performance metrics, which are defined for evaluating the performance of the whole network system. Furthermore, considering the complexity of the equilibrium analysis in large-scale network systems, a coevolution-based algorithm is proposed for obtaining the practical action sets as well as achieving the mixed-strategy Nash equilibrium. Finally, relying on four real-world network systems, i.e., computer networks, Internet of vehicles, air transportation systems, and social networks, simulation results show the effectiveness and feasibility of our proposed model, which is conducive to the design, management, and maintenance of network systems.