An efficient active probing approach based on the combination of online and offline strategies

Real-time problem diagnosis in large distributed system and network is a challenging task which requires effective and accurate inference and localization. In this paper, we use active probing to present an approach to develop tools for performing fault diagnosis. Active probing method is an active network monitoring technique that based on end-to-end techniques and requires minimizing the cost of probes while maximizing the diagnostic quality of the selected probe set. Existing works in active measurement contain offline and online probing strategies. However, both approaches have their limitation to the large-scale network. In this paper we propose a new algorithm that integrates the pre-planning and online probing strategies to find an approximate optimal subset. Experiments have shown that our approach is much more efficient than the existing strategies in selecting an optimal set of probes without degradation of diagnosis quality in large scale and noisy networks.