Home :: Short Bio :: Contact :: Publications :: Talks :: Tutorials :: Software :: Supervision
Abstract—Trace analysis is the behavioral study of network traffic to reveal patterns, quantify usage, measure feature deployment, and identify protocol shortcomings. An excellent repository of traces collected at several monitoring points is maintained by the National Laboratory for Applied Network Research (NLANR/PMA). Although the repository features a wide variety of traces, it does not provide software to analyze the binary traces. We fill this void in a platform-independent way with Net::Traces::TSH, a pure Perl module, which provides methods to analyze IP packet traces in Time Sequenced Headers (TSH). Net::Traces::TSH, available from CPAN, can examine TSH traces, collect Meta trace information, measure protocol usage, and quantify IP Fragmentation, Differentiated Services and ECN deployment. Packet and segment size distributions are also generated and stored in comma-separated values (CSV), a platform-independent text format. Furthermore, one can extract all TCP sender transmissions and essential information about each data-carrying TCP segment. The module can convert binary TSH traces to (a) tcpdump text format for visual inspection or presentation in a classroom; and (b) binary Traffic Trace format for input to the network simulator (ns-2). Our research goal is to develop a "TCP-centered" error model (TEM) based on the measured experience of real world TCP flows, and demonstrate that TEMs provide a more realistic simulation tool than rate-based models for TCP performance evaluation. Taking advantage of the fast features of Perl, we enjoyed rapid development without having to make costly sacrifices in performance. We were able to focus on the research problem and explore many ideas instead of spending time on implementation. In fact, based on the same code, we expanded our research to quantify the deployment of TCP options using traces from the NLANR/PMA repository. We believe that Net::Traces::TSH can become the standard framework for TSH trace analysis.
Conference web site and software
Tags—Traffic measurements and analysis
Home :: Short Bio :: Contact :: Publications :: Talks :: Tutorials :: Software :: Supervision
Last updated: 16.08.2008