Novel techniques for monitoring network traffic at the flow level

Viac o knihe

Research in Internet measurement offers new insights into the operation and enhancement of the Internet. Analyzing network traffic data necessitates a carefully selected set of techniques. This book explores innovative methods applied to large data sets, particularly focusing on network-level traffic data from commercial routers in the form of flow metadata (e.g., NetFlow). This approach facilitates the analysis of increasing traffic volumes with minimal measurement infrastructure demands. The book is divided into two main sections. The first part investigates unsolicited traffic without requiring unpopulated IP address ranges, as previously done. Instead, we examine one-way traffic—packets that do not receive replies in live networks—introducing a novel classification scheme for one-way traffic at the flow level. This scheme is validated using a comprehensive data set, revealing insights into the composition of one-way traffic and illustrating how the class of “Unreachable Services” can help detect network outages through flow-level data analysis. Additionally, we conduct an extensive study over eight years of traffic data, uncovering the evolution of this unique traffic type. In the second part, we present new visualization methods to summarize and interpret big data sets effectively. We propose a top-down workflow to identify hidden patterns in large networks, followed by detailed examinations of individual end systems o