Proactive Peer-to-Peer Architecture for Sharing Files on a Cloud Network

摘要

Peer-to-peer (P2P) applications can deliver large files efficiently by dividing files into small chunks. However, P2P methods employ greedy pull-based algorithms that can easily congest networks, such that transmission of other packets is blocked. Via P2P methods, network bandwidth is occupied easily by peers. This limits the ability of applications to communicate using P2P methods. In other words, P2P applications flood a network with packets, leading to network congestion. Consequently, such P2P applications are banned on enterprise networks, where bandwidth is an expensive resource. File sharing among networked devices may be important for some businesses. An efficient P2P application is a tempting solution for these companies; however, congestion remains a problem. In fact, even with today's private clouds for business use, P2P applications remain prohibited. Of course, P2P applications achieve outstanding file-sharing performance, but networking congestion is typically high. An efficient P2P application in cloud computing is ideal; however, flooding a network with P2P chunks must be controlled. To control the delivery of chunks, an active method is preferable to a passive method. Thus, a P2P file-sharing architecture that proactively manages network traffic would be ideal for cloud networks. In this work, a proactive P2P file-sharing architecture (PP2P) is proposed. Transmission time and the congestion ratio are determined through 534 experiments used to test the performance of the PP2P application in a real cloud environment (HiCloud, CaaS) with 20 peers supported by ChungHwa Telecom Co., Ltd. Experimental results indicate that congestion of the cloud network is minimized by the proposed PP2P architecture. The PP2P file-sharing architecture may be used by managers of cloud businesses to strike a balance between delivery efficiency and network traffic load.