Since the development of broadband Internet access grows day-by-day, there has been a considerable increase in the number of users in media streaming services. At present, Peer-to-Peer (P2P) streaming has gained more interest and with the aim of solving the difficulties in conventional server-based streaming. Al the same time, in P2P network, peers possibly will hop from one group to another dynamically and it is turned out to be an essential matter to potentially deliver the exact contents to peers. Conventional approaches establish high control overhead to handle with several dynamic overlays. Moreover, when users repeatedly hop from one channel to another channel, overlay reorganization becomes very costly and might produce huge packet losses. In the Dynamic and Traffic-Efficient Data Delivery (DTDD) infrastructure, incremental relaxation method is used for tree construction but it is not much effective in case of large number of nodes to constructively built and update the tree. Hence proposed a Dynamic and Traffic-Efficient Data Delivery with Multilevel Relaxation Algorithm (DTDD-MRA) infrastructure is built in this paper to overcome these drawbacks. It is revealed from the experimental results that DTDD-MRA considerably reduces the link stress and reduction in the network resource along with the execution time.