Recently, there has been strong interest in distributed schemes for intrusion detection in order to defend against distributed attacks such as DDoS attacks. In this paper, we focus on a placement problem of a detection system across large-scale networks for distributed intrusion detection approaches. We formulate the placement problem as a set packing problem that is NP-hard. We then present an efficient algorithm for minimizing the number of detection systems and finding the optimal placement while limiting the impact of distributed attacks.