next up previous contents
Next: metrics for burstiness Up: network analysis Previous: Internet traffic flow

network workload visualization

The most interesting flow metrics are also the most difficult to visualize; we have explored methods for visualizing network phenomena and characteristics. We have previously focused on the geographic representation of traffic flow between countries, animating country-by-country traffic matrices to produce a visual time series.

This year we have made considerable progress with software for visualizing IP flows, using a variety of tools, including xgraph, xmgr, splus, data explorer, gnuplot, and customized programs that we developed using the SGI GL library. Progress with commercially available visualization systems was slow, as they are written for visualization experts to optimize flexibility. We eventually decided to design our own software for traffic flow visualization using the GL library that is available on our RS6000 as well as SGI workstations. Jarom Smith, an SDSC REU, took an NSFNET data visualization project he completed using GL and installed it on a Cerfnet web server. We would like to expand the portability of these tools, and compare their utility and functionality to other tools in the community, for example Netramet from the University of Auckland, New Zealand [12].

Hans-Werner Braun
Wed Apr 19 20:12:08 PDT 1995