Most data placement schemes of P2P systems are based on open P2P systems like Gnutella where a node can freely access all data in the system after the node joins the system, which is not suitable for private file sharing. This paper suggests a closed P2P model where a node is connected only with some specific nodes and data are shared only within those nodes. In this model, since shared files usually contain large multimedia data, the data placement scheme employed should maximize the space available to each node by increasing the degree of sharing and by reducing redundant data placement. In this paper, we prove that the problem to maximize the space is NP-hard and suggest several heuristic approaches that can be realized on a peer-to-peer architecture. These approaches tend to be unfair to participating nodes since they are only focused on maximizing the total available space. To address this issue, we suggest another scheme where space available to a node is proportional to the space allocated by the node for file sharing. We also consider the limited network capacity problem, and decentralization problem. Performance issues of the above algorithms are explored through simulation studies.