|
iDEA: Drexel E-repository and Archives >
Drexel Theses and Dissertations >
Drexel Theses and Dissertations >
Traffic rerouting optimization in network recovery: a performance study
Please use this identifier to cite or link to this item:
http://hdl.handle.net/1860/3138
|
| Title: | Traffic rerouting optimization in network recovery: a performance study |
| Authors: | Bao, Fei |
| Keywords: | Computer engineering
Traffic engineering
Network performance (telecommunication) |
| Issue Date: | 26-Oct-2009 |
| Abstract: | In recent decades, the Internet has evolved from a special purpose computer network to a central platform of our daily communications. As it takes an increasingly important role in our everyday life, network reliability becomes even more critical. This is also one of the reasons why the Internet has been widely ubiquitous, due to its recovery ability. Internet protocol (IP) routing, such as Open Shortest Path First (OSPF) protocol, updates the forwarding tables to reflect alternative routes when the network topology changes. Due to the time consumed by the rerouting process after a failure, most recovery mechanisms are proposed to be proactive instead of reactive [1], [2], [3], [4], [5]. That is because in proactive approaches backup paths are preconfigured while reactive schemes such as that of OSPF usually are too slow to satisfy the recovery time requirements. After failures, all the traffic is moved to the backup path. After the traffic is transferred to the backup path, the fault on the primary path is r paired. Following the repair of the primary path, the trafficmay or may not be switched back to the primary path. There are two important points that need to be considered when deciding on rerouting traffic: i) the primary path may be flapping (fails again), ii) traffic rerouting cost may be too high. In this thesis, we investigate the performance of a mathematical model for traffic rerouting after repairs. Our contribution is to verify the the optimal traffic rerouting threshold-type control algorithm with simulations and an experiment. The simulations were performed in OPNET and the experiments were built on Sun VM VirtualBox technology. |
| URI: | http://hdl.handle.net/1860/3138 |
| Appears in Collections: | Drexel Theses and Dissertations
|
Items in iDEA are protected by copyright, with all rights reserved, unless otherwise indicated.
|
www.drexel.edu
