Network Service Diagnostics based on Packet Analysis
This dissertation is a compilation of peer-reviewed papers published during the doctoral study. The papers, as well as the thesis itself, focus on the problem of fault diagnosis in computer networks. Network faults can significantly affect the performance, availability, and security of networks, resulting in significant financial losses and degradation of user experience. To mitigate these problems, network diagnostics has become a critical area that focuses on identifying, locating, and resolving network faults. Although computer network diagnostics is not a new field and there are several tools and publications available, it is still an unsolved problem. The goal of this work was the development of new diagnostic methods, each of which is focused on a different type of error, and in the whole the individual methods are complementary to each other. The developed methods aim at automating the diagnostic process, which is time-consuming in case of manual execution and requires good knowledge of network protocols and technologies. Network packets saved in PCAP format are used as a data source for diagnostics, which allows to detect errors without the need to interact with the network. Another advantage of the developed methods is their wide applicability, as the solution is not limited to a small number of network protocols, and at the same time the methods can be used by network administrators and not only by researchers or programmers.
An integral part of the thesis is a description of the contribution of the dissertation and the results achieved during the study. In addition to the acceptance of the published articles, the successful commercialization of the product Flowmon Packet Investigator proves the correctness of the chosen approach. Finally, the research areas are summarized and further steps to improve the integration of the results into practice are presented.
computer networks diagnostics, troubleshooting network errors, diagnostic methods