High-reliability network system architecture using dynamic reconfigurable technology

Network reliability consists of the reliability of nodes and links. This paper focuses on node reliability. Nodes have applied reconfigurable technology to improve their functions. This technology can bring flexibility by making the configuration data of a function easy to change. Using this feature, this technology is applied to network nodes that provide high-speed packet-processing such as packet classification on a high-speed network interface. This paper describes the application of reconfiguration technology to enhance the reliability of packet processing nodes. The conventional redundant method requires an increase in the number of spare boards to raise system reliability, so it has a high equipment cost in exchange for high reliability. To solve this problem, we propose bidirectional shared serial N + k redundancy as a new redundant method. This can reduce the number of spare boards required and increase the recovery patterns for board failure. By evaluating system reliability, we show that our method can enhance the reliability of equipment better than the conventional method. Thus, this proposed method can attain high reliability and node miniaturization by means of reconfigurable technology and will lead to better network reliability.