Abstract: In P2P streaming systems, content servers which supply the source data and drive the data to spread in the whole system, are important components. The deployment of the servers obviously affects the network traffic which is related with the benefits of both P2P application providers and Internet service providers. But little research work focuses on the server deployment areas. The authors discuss the problem on server deployment from three optimization directions: load balance, minimum traffic in core network and minimum allocation costs. The adaptive algorithms, including load-balance-based greedy algorithms and the 0-1 linear-programming-based branch and bound algorithm, are presented. Some experiments are conducted to prove the effects of server deployments on the system, and it can be seen that different optimizations can achieve the different objectives.