Active and passive measurements on campus, regional and national network backbone paths

Abstract

It has become a common practice for Internet service providers (ISPs) to instrument their networks with network measurement infrastructures (NMIs). These NMIs support network-wide "active" and "passive" measurement data collection and analysis to: 1) identify end-to-end performance bottlenecks in network paths and 2) broadly understand Internet traffic characteristics, on an ongoing basis. In this paper, we present our analysis of the active and passive measurement data collected along network backbone paths within typical campus, regional and national networks which carry traffic of cutting-edge Internet applications such as high-quality voice and video conferencing, multimedia streaming and distributed file sharing. The active measurement data has been obtained by using "ActiveMon" software, which we have developed and deployed along the above network backbone paths. The passive measurement data has been obtained using SNMP, Syslog and NetFlow data available at the intermediate routers located at strategic points along the same network backbone paths. Our analysis of the measurement data includes studying notable trends, network events and relative performance issues of the network backbone paths which are reflected in the active and passive measurement data collected regularly over several months. Our results thus provide valuable insights regarding traffic dynamics in the different academic network backbones and can be used for better design and control of networks and also to develop traffic source models based on empirical data from real-networks.