Network Design Method Based on Link Protection Taking Account of the Connectivity and Distance Between Sites

Abstract

High reliability is required in networks, and it is important to build robust networks that are tolerant to network failures. In content delivery services in particular, service interruptions due to disconnection of communication paths between the server and nodes that receive the service must be avoided. Content delivery services use a master server that contains the original content and multiple edge-servers (mirror servers) that hold copies of the content so that communication paths exist between the master server and the edge-servers even in the event of a failure. In addition to the guarantee of network connectivity, large increases in path lengths when changing from the paths used during normal operation to alternative paths in the event of a failure must be avoided from the viewpoint of suppressing degradation of the communication quality and the possibility of congestion. However, building such high-reliability networks involves huge costs. Therefore, it is considered necessary to devise a network design that can realize continuity of communication between the master server and the edge-servers by suppressing sudden increases in path length, even if a failure occurs. In this paper, we formulate a network design problem and show that it is NP-hard. We also design a polynomial-time approximation algorithm for the case where the number of failed links is constant, and we evaluated its performance in various actual network topologies.