Visualization of the Growth and Topology of the NLANR Caching Hierarchy

Bradley Huffaker
Cooperative Association for Internet Data Analysis - CAIDA
San Diego Supercomputer Center,
University of California, San Diego

Jaeyeon Jung
Korean Advanced Institute of Science and Technology and
Cooperative Association for Internet Data Analysis - CAIDA
San Diego Supercomputer Center,
University of California, San Diego

Duane Wessels
National Laboratory for Applied Network Research - NLANR

k claffy
Cooperative Association for Internet Data Analysis - CAIDA
San Diego Supercomputer Center,
University of California, San Diego

As the NLANR Caching Hierarchy [http://ircache.nlanr.net], has increased in size and complexity, researchers and users have had a more difficult time depicting a comprehensive view of the overall topology. Planet Cache [ http://ircache.nlanr.net/Cache/cacheviz.html] was one of the first attempts at visualizing the hierarchy, but was not sufficiently flexible to customize views of subsets of the topology by specific attributes. CAIDA has expanded Planet Cache's visualization to include both topological and geographical depictions of the hierarchy. We have also invested in converting from VRML to a Java implementation to facilitate more interactive visualization. Our Java implementation allows the user to tailor the complexity, focus, and topological layout to match their specific visualization objectives. This ability to customize the visualization, combined with the universal availability that the web and Java provide, allows a much larger set of users to access data on the global caching infrastructure and gain insight into their own position in the hierarchy. While our goal is a universally accessible visualization tool, problems with the platform independence of Java and privacy sensitivities with respect to transcribing even logical topology information have resulted in our rendering some of the more useful aspects of the visualization only in standalone (`non-applet') mode.

Despite these drawbacks, the user-friendly interface and the visually informative and dynamic nature of both the standalone and the applet version are vast improvements over currently available tools, and provide a good basis for future analysis.