In this paper we consider gossip-based peer-to-peer streaming applications where multiple sources exist and they work serially. More specifically, we tackle the problem of fast source switching to minimize the startup delay of the new source. We model the source switch process and formulate it into an optimization problem. Then we propose a practical greedy algorithm that can approximate the optimal solution by properly interleaving the data delivery of the old source and the new source. We perform simulations on various real-trace overlay topologies to demonstrate the effectiveness of our algorithm. The simulation results show that our proposed algorithm outperforms the normal source switch algorithm by reducing the source switch time by 20%-30% without bringing extra communication overhead, and the reduction ratio tends to increase when the network scale expands.