Automated Network Node Discovery and Topology Analysis

Abstract: This Master's Thesis describes the design and development of an architecture for automated network node discovery and topology analysis, implemented as an extension to the network management and provisioning system NETadmin. The architecture includes functionality for flexible network model assessment, using a method for versatile comparison between off-line database models and real-world models. These models are populated by current node data collected by network sensors. The presented architecture supports (1) efficient creation and synchronization of network topology information (2) accurate recognition of new, replaced and upgraded nodes, including rogue nodes that may exhibit malicious behavior, and (3) provides an extension of an existing vendor-neutral enterprise network management and provisioning system. An evaluation of the implementation shows evidence of accurate discovery and classification of unmatched hosts in a live customer production network with over 400 nodes, and presents data on performance and scalability levels. The work was carried out at Netadmin System i Sverige AB, in Linköping, Sweden.