The deployment of cache servers in P2P networks for improved performance in content-delivery
T. Oh-ishi, Kazuya Sakai, Tomoharu Iwata, Akira Kurokawa · 2004
The peer-to-peer (P2P) model generally requires that bidirectional and direct communications for content-delivery be set up between user peers after discovery of the desired content; however, firewalls and the currently popular NAT/NAPT (Network Address (Port) Translation) boxes prevent most end-users from freely participating in bidirectional communications. The preferred solution to this problem is to apply rendezvous points (RPs) as mediators in content-delivery. An RP is transparently accessible through firewalls to end-users. However, content delivery with the RP scheme takes ages, since there are two phases in the transfer-uploading from the holder-peer to the RP and downloading from the RP to a requestor-peer. We propose a way of improving performance in P2P content-delivery where an RP mediates among peers within firewalls. This is based on the deployment of cache servers, which are currently popular as a means for getting greater efficiency out of the Web. This solution provides the same performance in content-delivery as is seen in direct communication between user peers.