The increasing popularity of the World Wide Web has led to significant research targeting the reduction of the latency perceived by users. Cache servers proved to be an important resource in accomplishing this goal.Although the use of cache servers have reduced the data traffic in the Internet, the cooperation strategies employed while building clusters of those servers scale poorly and are not able to sustain their throughput under increasing demands.In this work, we present a new cooperation strategy that does not affect significantly the server performance. The use of this strategy allows clusters of cache servers to scale well with their usage growth.