Data Measures and Collection Points to Detect Traffic Condition Changes on Large-Scale Computer Networks

Abstract

Computer networks, which play a crucial role in the operation of many organizations, are vulnerable to various problems that may cause traffic condition changes, with possible negative impact. In computer and network systems management, it is desirable to detect such changes and correct them before localized problems propagate to an entire network. For large networks, monitoring large amounts of data at all points is inefficient. This study aims to direct network management to those data measures and collection points that are most effective for efficiently detecting traffic condition changes; minimizing the amount of data required for accurate analysis. We design and build a network model to experiment under normal and problem network conditions. Our results indicate that IP traffic received is a good metric for detecting traffic condition changes on our network. The best point for collecting this metric is at popular routers at the edge of collections of sub-networks.