Efficient and Robust Distributed Network Monitoring using Dynamic Group Formation

In this paper, we propose middleware for efficient and robust distributed network monitoring. In our technique, we specify network monitoring items, detection methods of viruses and DDoS attacks and the corresponding reactive actions as a management scenario. At the same time, we specify logical neighboring relations between separate network segments so that we can efficiently detect problems occurring in multiple segments. When a problem occurs in multiple network segments, the corresponding monitoring nodes form a group dynamically using logical relations. Based on the pre-defined management scenario, the nodes in the group carry out the reactive actions autonomously. We have designed and implemented our middleware consisting of several useful APIs. We have confirmed effectiveness of our technique through ns-2 simulation