Unstructured P2P networks remain widely deployed in file-sharing systems, due to their simple features. However, the P2P traffic, mainly composed of repetitive query messages, contributes the largest portion of the Internet traffic. The principal causes of this critical issue are the search inefficiency and the construction of the P2P overlay without any knowledge of the underlying topology. In order to reduce the P2P redundant traffic and to address the limitations of existing solutions, we propose a solution that performs index caching and efficient query routing while supporting keyword search. We aim at improving the probability of finding available copies of requested files by leveraging file replication. In addition, our scheme tries to direct queries to close results, by using topological information in terms of file physical distribution. We believe that the traffic can be reduced and the user experience ameliorated in terms of faster downloads, with minimum overhead.