On network traffic concentration and updating interval for proactive recovery method against large-scale network failures

Abstract

Proactive recovery methods from network failures, based on multiple routing configurations, are effective for quick failure recovery, as compared with reactive recovery methods. However, there are two major problems, especially when we consider recovering from large-scale network failures: updating interval for recalculation of routing configurations against network growth, and network traffic concentration on specific nodes and links after recovering failures. In this paper, we first propose a light-weight and distributed algorithm for updating routing configurations when new nodes and links join the network, which does not need overall recalculation. We then evaluate the proactive recovery method for large-scale network failures, with proposed algorithm from the viewpoint of above two problems. Through numerical evaluation results, we find that to maintain the recovery performance, we should recalculate the routing configurations when the network grows by roughly 5–10%. We also present that the degree of network traffic concentration decreases when we employ a hop-by-hop selection mechanism of routing configuration.