High quality queueing information from accelerated active network tomography

Abstract

Monitoring network state can be crucial in Future Internet infrastructures. Passive monitoring of all the routers is expensive and prohibitive. Storing, accessing and sharing the data is a technological challenge among networks with conflicting economic interests. Active monitoring methods can be attractive alternatives as they are free from most of these issues. Here we demonstrate that it is possible to improve the active network tomography methodology to such extent that the quality of the extracted link or router level delay is comparable to the passively measurable information. We show that the temporal precision of the measurements and the performance of the data analysis should be simultaneously improved to achieve this goal. In this paper we not only introduce a new efficient message-passing based algorithm but we also show that it is applicable for data collected by the ETOMIC high precision active measurement infrastructure. The measurements are conducted in the GEANT2 high speed academic network connecting the sites, which is an ideal test ground for such Future Internet applications.