Finding a good balance between link congestion and over-provisioning for network service providers is a challenge. Recent advances in transport protocols promise a smoother transfer of bulk background data — trafﬁc that P2P (Peer-toPeer) ﬁle sharing systems are signiﬁcantly contributing to still today. This paper focuses on reducing the link congestion, and trafﬁc peaks originating from P2P systems by replicating data during periods of low network trafﬁc. For this purpose, RBTracker is proposed as an extension of B-Tracker in order to automatically manage the overlay network and replication of content, similar to a Content Delivery Network (CDN). RBTracker is designed to manage autonomously a P2P network and to use state-of-the-art mechanisms, such as biased neighbor selection.