Selecting Measurement Paths for Efficient Network Monitoring and Diagnosis under Operational Constraints

End-to-end network diagnostics tools for identifying degraded links along an Internet path have been studied extensively, but few are found on the design and development of deployable solutions based on these approaches for actual large-scale networks. This paper presents a scalable measurement system that utilizes many users' off-the-shelf PCs as measurement end-point nodes (called gbeaconsh) to actively monitor network performance along many end-to-end paths and to conduct network diagnosis to infer degraded links when end-to-end performance degradation is detected. The network paths of end-to-end performance measurement and network diagnosis among beacons are dynamically and adaptively selected taking operational constraints such as the network resources along the path and the machine resource of the beacons into consideration. The system is evaluated through large-scale simulations and small-scale real-world experiments using a prototype implementation, and its considerable potential for practical network operations is demonstrated.