With the development of cloud manufacturing technology, Manufacturing Service Ecosystem (MSE) is emerging as a typical complex cyber-social system. On the one hand, service strategy (cyber layer) drives the evolution of manufacturing community (social layer); on the other hand, the initial conditions of manufacturing community (social layer) affect the performance of service strategy. In order to promote the evolution of MSE in the expected direction, it is necessary to clarify the loop feedback mechanism between heterogeneous networks. However, how to analyze and intervene in the possible evolution directions of MSE has become a serious challenge in the field. In order to face this challenge, this paper proposes a parallel system theory-based research framework to study the evolution and controlling of MSE. Firstly, the corresponding digital system of MSE is constructed from the perspective of supply and demand matching. Secondly, the specific computational experiment is executed to present the effect of different service strategies (cyber layer) and different initial conditions (social layer) on the evolution of MSE. Furthermore, the comparison of experiment results with real data verifies the credibility of the proposed approach. It demonstrates that our approach can provide a new way for analyzing the complexity of MSE.