Overlay Network Virtual Link Technology

Overlay network is introduced to solve the continually appeared all kinds of applications. It is a virtual network overlaid upon the existing networks, constructed with a lot of overlay nodes distributed in different ASes(Autonomous Systems) in Internet and the virtual links between each of them. It can easily be employed and provide much better quality of service with little change to the current network architecture. Spines is an open source overlay network system that allows easy deployment and testing of overlay protocols. It runs in the user space, does not need root access or kernel modifications, and encapsulates packets on top of UDP. Spines offers a two-level hierarchy in which applications connect to the closest overlay node, and then the node is responsible for forwarding and delivering data to the final destination through the overlay network. The benefit of this hierarchy is that it limits the size of the overlay network, thus reducing the amount of control traffic exchanged. Overlay nodes act both as servers and as routers. Applications may reside either locally with the Spines nodes or on machines different from the overlay node they connect to.Virtual links on which this thesis focus on is one of the key items of Spines overlay networks research. After introduces the two-level hierarchy and software architecture of Spines, this thesis introduces the current condition of available bandwidth technology and propose an active available bandwidth algorithm named BELR, BELR is an available bandwidth estimation method for hop-by-hop, different from the ones for end-to-end, it can work well, otherwise, the algorithm can be put in practice simply. Then, researches Spines virtual link protocol-hop-by-hop reliable communication protocol and then analysis its performances by simulation with NS2, which is a network simulation running under Linux and come up with a proposal improving its performance by routing with available bandwidth. The simulation results indicate that our new routing strategy performs better than the existing routing strategy, especially for the latency, loss rate and throughput. Meanwhile, we change the value of retransmission rate of real-time recovery protocol, results in low loss rate and making more packets arrive at final destination in time , we also presente how to integrate with SIP and Spines in the the current Voice over IP infrastructure.