Social networks have got increasing attention in recent years, and social networking services (SNSs), such as YouTube, Facebook and Twitter, are among the most popular network applications on the internet. The popularity of these SNSs attracts more and more researchers to focus on the study of the characteristics of the social network, especially the topology of networks, so as to improve current systems and to design new applications of social networks. However, most existing studies only focus on the dynamical network structure changing with time accumulation, and they do not take into account of other properties of the social network structure, such as link life. The link life reflects the phenomenon that the link established in the past may not be permanent and may vanish as time goes by. In this paper, we focus on the influence of link life on the dynamical network structure evolution, more specifically, on the basic and important parameters of the social network structure, including degree, diameter and average clustering coefficients. Experiments on the real datasets from the DBLP shows that the link life is necessary and important, which makes a great difference to the social network structure evolution. Particularly, the trivial perturbation of link life can lead to a dramatic change of the network diameter.