Fast dynamic internet mapping

Abstract

Probing network devices using the Traceroute tool is a most popular network monitoring technique for troubleshooting and network mapping. Distributed scanning systems perform Traceroute periodically in order to detect routing anomalies. We propose in this paper a six-step methodology for creating a more efficient profile-based probing strategy that reduces both the number of probes and the time needed to map the network topology. Our proposed methodology takes the existence of load balancers into account when building the profile-based strategies and thus overcomes any inconveniences that these load balancers may cause. The basic idea behind our methodology is to examine how often routing changes occur in the routing path at the different times of the day. This insight will be used to provide a high probing weight for parts of the routing path that change frequently and a lower probing weight for parts that change rarely during the different time periods of the day. Since routing changes may occur more frequently at certain periods of the day, we propose an approach to determine the duration of the periods when the routing changes occur with a similar frequency. A profile-based probing strategy is later assigned to each one of these periods separately in order to further reduce the required number of probes. The experimental results show that our approach exploits temporal regularities in the routing changes and achieves great savings in the number of probes. In fact, our approach achieves a 66% reduction in the number of probes as compared to classical Traceroute that is launched periodically. This is an important enhancement for systems that scan the Internet repeatedly. Furthermore, we show another enhancement that cuts down scanning time by 90%. This is achieved by scanning multiple hop levels at the same time after marking the probes in order to match them with the received responses. © 2014 Elsevier B.V. All rights reserved.