The expansion of the Internet routing system results in a number of research challenges, in particular, the Border Gateway Protocol (BGP) starts to show its limits a.o. in terms of the number of routing table entries it can dynamically process and control. More scalable routing protocols are thus under investigation. However, because deploying under-development routing protocols on the Internet is not practicable at a large-scale (due to the size of the Internet topology), simulation is an unavoidable step to validate the properties of a newly proposed routing scheme. Unfortunately, the simulation of interdomain routing protocols over large networks (order of tens of thousands of nodes) poses real challenges due to the limited memory and computational power that computers impose.
Until recently, the NS simulator, widely used in the networking
community, did not comprise any BGP routing model. The recently
introduced BGP++ module solves this limitation and allows NS users to
perform inter-domain routing protocol simulations. BGP++ chooses to port
an existing BGP daemon from Zebra to NS. This allows BGP++ to support
most of Zebra daemon capabilities. Thus, it is now possible to use this
daemon to build realistic inter-domain routing scenarios but not on
large-scale networks. SSFNet, another discrete event simulator, took a
different approach than BGP++. All the simulators previously cited here
above share many properties in common. They all rely on
discrete event simulation. They are faster than real implementations and
they support other protocols such as TCP, which is appropriate for
simulating protocol dynamics. However, they are too detailed for BGP
simulations (microscopic-level simulation): state machines of protocols
are fully modeled. As a consequence, they need so much resources to
simulate the behavior of the protocol that they are limited to networks
of usually of few hundred nodes: large-scale simulations are out of
reach.
This is partly what motivates us to build a network simulator
tailored to effective routing: only the necessary components for
simulating routing protocols over large-scale topologies are
modeled.
DRMSim is a Java based software simulator that aims at the evaluation of routing models at a large scale (up to ten thousand nodes on typical workstations). The design/development philosophy of DRMSim is make to compromise on the quality of the code written so that extensibility and reusability are maximized. Its development has started in the context of a study focusing on dynamic compact routing. A careful analysis of the data structures that will be used in the network model as well as on the granularity and time management of the simulation model has been performed in order to reach our objectives.
This joint research project is conducted by Alcatel-Lucent Bell, Universitée de Bordeaux (LaBri) and INRIA at Sophia Antipolis (Mascotte project) and is supported by the Euler Fire Project for the investigation of a new routing scheme suitable for the future Internet and its evolution.