Distributed emulation in support of large networks

Abstract

Network emulation is a valuable, though potentially resource intensive, method for virtualizing networks for analysis or testing. Though high-powered servers are becoming increasingly accessible, the size and complexity of physical networks have increased in a similar fashion, thereby limiting the type and size of networks that can be emulated on a single physical machine. In this thesis, we present a tool that allows the developers of ground truth topologies to distribute the emulation requirements across multiple physical machines, thereby increasing the size of networks that can be emulated. First, we reexamine existing tools to discover current methods for emulating synthetic and physical networks. Then we modify an existing platform to enable execution on multiple machines, while increasing flexibility for future extensions. We then develop methods for efficiently distributing the topology among the available resources in order to maximize the potential scale. Finally, we run a series of scenarios simulating real world events, such as a Border Gateway Protocol (BGP) hijack attack, in order to demonstrate the utility and efficiency of the system.