BRITE was designed to be a flexible topology generator, not restricted to any particular way of generating topologies. As such, it supports multiple generation models. In this section we describe how this design goal was approached and how BRITE is implemented. Figure 1 depicts a schematic view of the structure of BRITE as it is being used at Boston University. The different components are labeled (1)-(4).
BRITE reads the generation parameters from a configuration file (1) that can be either hand written by the user or automatically generated by BRITE's GUI (described in Section 4). BRITE provides the capability of importing topologies (2) generated by other topology generators (GT-ITM [5], Inet [13], Tiers [8], BRITE 1.0 [16]) or topological data gathered directly from the Internet (NLANR [17], Skitter [6]). Note that we include BRITE in the ``imported'' file formats, because it is possible to generate topologies using BRITE and then reusing them to generate other topologies by combining them with BRITE models or other imported formats. In the current distribution BRITE produces a topology in its own file format (3), and output capabilities for producing topologies that can be used directly by the Network Simulator (NS [18]) and the Scalable Simulation Framework (SSF [22]) simulators are currently being developed.
We developed a piece of software, separate from BRITE's generation tool, and we call it the BRITE Analysis Engine or BRIANA (4). BRIANA takes advantage of the flexible design approach of BRITE. The idea of BRIANA is to provide a set of analysis routines that may be applied to any topology that can be imported into BRITE. If we need to analyze a new topology, we just add a parsing procedure to BRITE for that new format, and once that is done, the set of analysis routines can be used on the new topology. Section 7 summarizes BRIANA's features.
