Traffic replay tools that generate traffic in order to recreate real traffic scenarios play an important role in network operation. Nowadays, the Internet backbone links are increasing in capacity as well as in transmission speed making trace-based replay difficult. This paper proposes a novel approach to high-speed backbone traffic replaying. We bring out an original granular split of traffic that provides a smaller number of manageable entities than in case of traffic packets and flows. The proposed method uses IP aggregation to split traffic which allows us to incorporate social phenomena into traffic replay. The practical use and the advantages in real-time replay are the two major points established in this paper.